# Building Tools for Edge Based Control by ps94506

VIEWS: 18 PAGES: 51

• pg 1
									Building Tools for Edge Based Control
Understanding Edge IPv6 versus Backbone IPv6
VoIP and Vonage - When Customers Become Competitors
Open Spectrum Versus the Spectrum-as-Property Worldview
We are finding more evidence that the        January 7, 2003. http://shirky.com/writ-
phone companies are caught in an un-         ings/zapmail.html What is wrong here         Volume XI, No.12, March 2003
dertow from which they cannot escape -       is not speed of connection but rather        ISSN 1071 - 6327
- unless the FCC and, or Congress does       control over the technology. The phone      sands and eventually millions of fax ma-
something truly idiotic like grant them a    companies cannot see the world in any       chines. Rather than rely on Fed Ex for
monopoly on fiber to the home. “Grant        terms other than those of control.          the faxing service, Fed Ex’s customers
them complete control over the glass                                                     bought their own fax machines and did
and then they will invest” will run the      A Matter of Mindset --                      it for themselves. Today rather than rely
tired argument. The first problem is that                                                on centrally controlled circuit switched
they have pledged this before and done       ZapMail and the Telcos
technology, increasingly large numbers
nothing. The second problem is that if                                                   of phone company customers are taking
they were given yet another opportunity      Consider what happened to the hot new
company Federal Express in the early        telecommunications into their own
there is and will be no enforcement for                                                  hands.
any of the pledges they make.                1980s. There was this new fangled de-
vice called a fax machine that scanned a
document and sent the resulting digital     It is a simple matter of economics. The
To borrow the metaphor from the 19th                                                     cost of communication via IP is but a
century, the result of granting them a       bit map over a phone line. The time was
just before the split up of ATT and the     fraction of the cost of doing it the phone
fiber monopoly would be to hamstring                                                     company way. In our January-February
the entire American economy into re-         explosion of customer premises phone
equipment. Fed Ex totally missed what       issue we saw how the large corporate
liance on “canals” in order to scare off                                                 enterprises are beginning to pull their
this new and chaotic world called “rail-     was happening. Thinking that its com-
petitors were the other over-night deliv-   voice service from the PSTN. This
roads.” While other countries are build-                                                 issue examines why Ipv6 is unlikely to
ing “railroads” - that is broadband - for    ery companies, it spent 200 million dol-
lars in an attempt to one-up them by        ever be significantly deployed in back-
us not to do so would irreparably handi-                                                 bone of the Internet. It also will show
cap what is becoming one of the most         buying expensive new fangled fax ma-
chines and building a dedicated phone       how IPv6 deployed at the edge of the
basic infrastructures of a modern econo-                                                 network, in the hands of the end user
my. We are already behind. The Orga-         network to run them on.
customers of the phone companies,
nization for Economic Cooperation and                                                    could do a great deal to redress the on-
Development (OECD) ranks United              As Shirkey writes in his essay, they
failed to see that the breakup of ATTand    going consolidation of power into the
States only 17th in utilization of com-                                                  hands of the central control minded tel-
munication services. [Quoted in Feb 1        the consequent opening of the network
Pulver.com letter to FCC Chairman            would allow their customers to buy their
own fax machines and by being able to
Powell.]                                                                                         On the Inside
use the PSTN, become their competi-
tors. They underwent a huge build out             Tools for Edge-based
But fiber is fast you say and speed of
connection is the issue – not control over   for a business that wasn’t there. With                  Telecom
access. What is wrong with such a sce-       the network opened up, Fed Ex’s cus-
nario? Well consider the ZAP mail ex-        tomer bought thousands and then tens of      Contents                   p. 2
perience as written by Clay Shirky on        thousand and then hundreds of thou-
Contents
Building Tools for Edge Based Control -- Understanding Edge IPv6 versus Backbone IPv6
VoIP and Vonage - When Customers Become Competitors - Open Spectrum Versus the Spectrum-as-
Property Worldview                                                                                                      pp. 1 –4

IPv6 Going No Where - Political Push Fails to Propel Elegant Solution Lacking Market Pull -
Former Drivers of Address Space, Device Addressing and Wireless Seen As No Longer Critical -
While Very Important at Edge, v6 to See only Niche Backbone Deployment                                                  pp. 5 - 13

Is IPv6 Necessary? - One Year Later                                                                                     p. 10

IPv6 at the Edges -- IPv6 Seen Not as a Backbone orTransport Solution But Rather as User-Applied
Edge-Based Overlay Supporting End-to-end Applications                                                                   pp. 11 -16

Two Internet Futures - With Edge IPv6 and Without Edge IPv6                                                             p. 17

Customer Owned Networks --ZapMail and the Telecommunications Industry
by Clay Shirky                                                                                                          pp. 18 – 21

Discussion of Clay Shirky's ZapMail Essay
Unlicensed, User Financed, Edge Based Connectivity Technology -- Locustworld Meshbox in
Context of Building Edge Based Wireless Transport                                                                       pp. 22 – 24

Open Spectrum - Property Rights World View Dies Hard
Exploring the Problems with the Farber-Faulhaber Have-Your-Cake-and-Eat-it-Too Spectrum Arguments                       pp. 25 - 30

ICANN and the Failure of "Self Regulation"
How the National Science Board was Overruled by the Clique that Became ICANN - Part One                                 pp. 31 - 33

Governance by Lawrence Lessig
Lessig Demonstrates How the Would Be "Self Regulators" Took Control - Part Two of How ICANN Came to Be
pp. 34 - 39

Interview, Discussion, and Article Highlights                                                                           pp. 40 - 47

Executive Summary                                                                                                       pp. 48 -51

cos and cable companies.                     grey market has never been discussed in     they could for a minute of data traffic is
detail in the press. Beginning with our     rapidly diminishing. Recently the differ-
We have been learning a lot more about       April issue we shall do so.                 ence has been as high as seven to one.
the technology of VoIP. In our April                                                     That is if a telco could make a penny for
issue we shall return to VoIP and shall      The stark fact is that the blades of the    a minute of data transfer, it could make
show how new developments are already        VoIP scissors are closing in on the         seven cents for each minute of voice
beginning to lower the artificially high     telco’s cash flow. On the one hand one      transmission.
costs of international phone tariffs. We     blade is the result of large corporations
anticipate writing about the technology      withdrawing voice traffic from the          For the most part the seven-cent differ-
and mechanics of the use of VoIP on a        PSTN and running it over their corpo-       ential is no longer there. Bits are bits.
global basis as a substitute for expensive   rate IP networks. On the other hand the     One cannot really distinguish voice from
circuit switching. The cost spread be-       other blade is derived from international   data bits. That any price difference ex-
tween international circuit switched calls   VoIP wholesaling by companies like          ists at all is increasingly a regulatory ar-
and VoIPcalls which can be routed from       ITXC and activities by thousands of         tifact. In two or three years market and
one part of the PSTN to another is now       phone card middle-men hammering             technology pressures will have driven
so huge that it has spawned a global grey    long distance rates ever downward. The      the differential to zero. When this point
market. In part because so many people       ability of the phone companies to charge    is reached, the telcos could find their rev-
are so busy making money from it, this       more for a minute of voice traffic than     enues slashed by two thirds. They will

2
The COOK Report on Internet March 2003
than have all the relevance of Zap Mail.     viding consumers with a direct substitute
to their traditional wireline phones.”       For the first time Cisco is beginning to
Canadians Abandon                            “These various sources of competition        understand that these radios can be used
have contributed to the first declines in    not just on a corporate campus but rather
Faith in Facilities Based total access lines for the four major                           can be used to replace the local loop in
Regulation                                   ILECs since 1933 (the only previous year     community based applications be it
where access lines declined).”               Nepal or in connecting community net-
Meanwhile the regulators cannot keep                                                      works in Wales where Forster, at Hughes'
pace. Testifying before the Senate Com- Vonage and Cisco                                  urging, has also involved Cisco with pos-
merce Committee on January 14, Chair-                                                     itive results. Of course, if they work in
man Powell said “The Commission has With his mention of Vonage, we can cer-               Nepal, and in Wales, they will work in
before it a number of major proceedings tainly see that Powell has more clue than         the US. Unless acting in ignorance and
that will attempt to improve and advance he possessed a year ago. We are about to         on behalf of the telcos, our political and
the goals of the 1996 Act. With the bene- sign up for this service that for the first     regulatory system forbids it.
fit of hindsight, we will be able to assess time takes a Cisco product (the AT186)
the last seven years and consider how we VoIP gateway and treats it as a consumer         The bottom line of all these events signi -
might improve the regulatory environ- product. The gateway plugs into the RJ-             fies only one thing. The local telephone
ment to more aggressively promote facil- 11 jack at the back of the phone and Eth-        company’s standard business model is
ities-based competition, to promote ernet into the cable modem in the back of             dead. Rendered extinct by users taking
major investment in advanced communi- the gateway. The result is unlimited long           control of inexpensive technology and
cation infrastructure, and to reduce regu- distance in the fifty states for $40 a using it for their own purposes. Because lation—all hallmarks of the Act.” (p. 6) month – plus very attractive rates to the Cisco, unlike Nortel and Lucent, always rest of the world. It is important to note had its major business outside that of the As we showed last fall in our asset-based that Cisco, as a device selling company carriers and the ILECs, it is in the wire- telecom issue (Vol. 11 Nos. 8 – 10), the and not a phone company, is well posi- less and VoIPareas much better equipped Canadians have essentially given up on tioned to profit from the VoIP price scis- to deal with the world in the aftermath of trying to make facilities based competi - sors. the death of the carriers than its more tion work. It is too bad that the FCC feels telco-oriented sister companies. trapped in the requirements of the 96 Act. Furthermore those who have read the As Powell describes it, the FCC is in dan- New York Times January 23 coverage We signed up with Comcast Cable Inter- ger of striking out having taken two (http://www.nytimes.com/2003/01/23/te net earlier in January in order to install swings and failed on both occasions. The chnology/circuits/23sher.html) of our Vonage. On Monday January 27 we or- Canadians, realizing the impossibility of work with Tsering Gyaltsen Sherpa will dered the package. The question of se- the task, have gone onto better things. see another interesting aspect of what lecting the Vonage phone number was could be a new found Cisco view of the not immediately clear although with Powell has shown some interesting world. Cisco donated Aironet 350 radios hindsight it seems obvious. The service changes during the past year. Among to the Everest base camp project at Dave gives a separate phone line with its own them is a journey from saying that he Hughes’ urging. These are radios that phone number. We still have our 609 didn’t know what the public interest was Cisco markets to connect LANs inside of 882-2572 number. Dial our new Vonage to the following remarkable statement. “. building in large corporations. Prior to number 703 738-6031 and you will also . . we will be guided exclusively by the this it seems never to have occurred to ring our desktop phone. Moreover if you public interest, and resist the pressure to Cisco marketing people that these radios are in the Washington DC suburbs and view our exercise as awarding benefits can be used to bridge a LAN to a VSAT 703 is a local call, dialing 703 738 6031 and burdens to corporate interest.” (Page at 5500 meters over a distance of two gets you through to us for a local call re- i). Of course the proof will be not in kilometers. The Times wrote: “Mr. gardless of whether you are a Vonage words but rather in actions. Forster eagerly donated three Wi-Fi ra- customer. The Vonage web pages dios on behalf of his company. Such ra- http://vonage.com did not have an 800 On page four of his text he said: “In ad- dios enable the creation of wireless net- number listed for dial in. Frustrated. We dition, broadband connections have also works that can relay data within a couple sent email asking to talk to a real live put pressure on wireline networks as of hundred feet or as far as several miles human. Not five minutes later our phone many consumers that migrate to broad- as the crow flies, much the way that rang and a very helpful resident of Con- band for their Internet services have local-area networks, or LAN's, work in necticut who worked from his home as dropped their second telephone lines offices. "What I like about this project is well answered our questions. We com- (which were used for dial-up Internet that it demonstrates that the technology pleted the order chose the Vonage phone services). Moreover, 2002 saw the intro- developed for a LAN in a building can be number and were billed$40 for the first
duction of reliable Internet telephony applicable beyond that," Mr. Forster said.         month service, $30 for account activation services through a broadband connec- "This may be as far outside the building and$10 shipping for the Cisco ATA 186
tion. Companies such as Vonage are pro- as you can get."                                  gateway.

3
COOK Network Consultants, 431 Greenway Ave. Ewing, NJ 08618 USA
and play! It was not even necessary to within the Internet. Indeed we have
The package arrived the next day. The open a network panel and configure an begun a fairly in depth exploration. It is
only hassle was buying a two line phone IP number for the gateway. Our Apple not yet really clear what Microsoft will
(price range 30 to 60 dollars) and the Airport Base Station did that transparent- offer in order to make edge based IP v6
right RJ11/14 connector. Installation ly acting as firewall and router. They applications plug and play. Standards
was a breeze. Total plug and pay. Well quality is excellent. Our first interna- would help enormously. Five to ten
almost. We signed up for Free World tional call was to Arcady Khotin in St years ago the IETF would have been the
Dial Up and that does not work with Petersburg, Russia. The cost was seven place to turn. Today it might be the
Vonage. Why? Because the gateway is cents a minute. The same rate that we IEEE.
shipped pass word protected and to work were paying ATT for domestic long dis-
with FWD a proxy IP number must be tance.                                           Or it might even be the Consumer Elec-
inserted.                                                                           tronics Association. At some point, we
The Center is Dead                        hope to offer input from Virginia
The Cisco gateway retailing for $200 Williams who is active at CEAin leading and available wholesale at$135 and The center is dead. Forward movement an effort to enable whole families of de-
“free” from Vonage is very compact. is at the edges. The major focal point for vices plugged in at the edges of IP net-
Roughly 6 inches by 6 inches and less this issue is IP v6. Farooq Hussain works to find each other. In a conversa-
than 2 inches high. Plug in the power shows why its chance for significant de- tion with her on January 30 we learned
cord. Plug in the RJ11 from the phone ployment in backbones at the core of the that there are several consortia of com-
and connect an Ethernet cable from the Internet is effectively zero. However in panies within the consumer electronics
gateway to our 8 port fast Ethernet a discussion with David Reed, Bob field that are exploring a range of issues
switch ($45). With line 2 on the phone Frankston, Francois Menard and Farooq that could be described as loosely related set as the default (the VoIP line plugged we are introduced to the concept of V6 at to Edge Based v6. We hope to describe into the gateway) pick up the phone and the edge of the network. We begin to un- these efforts in more detail in a future dial. The gateway has a large red light derstand how V6, in the hands of end issue. on the top that glows red when working. users at the edges of the network, could redress the shift toward the center that The experience actually was totally plug has taken place in the balance of control Editorial Calendar In the next issue we shall return to Voice over IP. In the one after that we likely shall do a reprise of asset based telecom which is now going global in major ways. 4 IPv6 Going No Where - Political Push Fails to Propel Elegant Solution Lacking Market Pull Former Drivers of Address Space, Device Addressing and Wireless Seen As No Longer Critical While Very Important at Edge, v6 to See only Niche Backbone Deployment Highlights rope and North America and is financial- to the existing v4. Some critics now ly stable.) I hadn’t paid a lot of attention would say that part of the problem is that Editor’s Note: Farooq Hussain was the to what had been going on with IPv6 be- the whole goal of expanded address Principal Investigator for the Sprint fore mid 2000 or so. Like everyone else space is just propping up the established NAP and moved shortly after the I had been reading all the announce- concept that every device reachable NSFNET transition from Sprint to MCI ments that it was “about to happen” and from the Internet needs at least one per- joining the team directed by Vint Cerf. my first inkling was that as long as it was manent layer-three address. He left MCI just prior to the completion about to happen, perhaps this client of the merger with WorldCom having should be doing something about it. The Presumed Address worked on both the merger plan with BT Space Shortage and subsequently WorldCom for the In- At the time there were three or four large ternet components of MCI. He was with US Operators, most notably WorldCom Ten years ago this was actually not such AGIS for a little over a year helping to and Sprint who were saying that they an unsound approach. We then had this establish a business relationship with had v6 networks operating. idea that the car would have its IP ad- Telia of Sweden who subsequently dress and that within the car maybe the bought AGIS out of bankruptcy. Cur- COOK Report: In the sense of test net- air conditioning system and carburetor rently, he is a partner in a research and works or were they really production? also needed their own IP addresses. Just consulting firm Network Conceptions as every house has a phone number, together with Phil Jacobson [also an ex- Hussain: I believe the old vBNS had v6 everything was to have its own IP ad- MCIer]. We interviewed Farooq on Jan- going. You had in the engineering com- dress. But things have turned out rather uary 3, 2003. munity a lot of tension between those differently. We are much more sensitive who were strong proponents of v6. to devices and uses being session orient- There was some middle ground among Is IPv6 a Deployable those who were not really bothered one ed. And having, as a result, temporary addresses. Protocol? way of the other while on the other side there were and still are some very very Now we are looking at problems of the Hussain: IPv6 and the question its de- strong critics of v6. ployment is wrapped up in a series of Internet in going forward a decade later quite complicated tensions which are and it will not necessarily be appropriate There has been, from the very beginning, to say that what has happened over the difficult to articulate. My interest is in a considerable amount of tension within focusing on the policy issues that sur- past 10 years to the way that v6 has de- the IETF about the need for an approach veloped actually applies very well to the round it as well as the lack of any rea- to IPv6. Lying at the very foundation of sonable way to determine what the com- current situation. an understanding of where v6 is going is mercial value of deploying it would be. the necessity of understanding the ra- There are two camps. One says IPv6 is COOK Report: In terms of current op- tionale for its creation back in 1992 - 93. erational economic and technology con- not needed and won’t happen and those Everyone said then that we were going who say it is absolutely necessary and cerns? to run out of address space. This con- will happen. These diametrically op- cern about address space continues up to posed positions all stem from a very Hussain; Exactly! On all levels! But today to be stated as the key rationale for the difference of opinion in the engi- fundamental issue of where we are with IPv6. protocol development. neering community is really substantial. Seen in this light we have had a parallel The reality is that the problems with the path of the pursuit of the development of Two years ago a major international car- Internet protocol that v6 was designed to rier whose networks were certainly ap- v6 while, at the same time, IP Sec, solve have been managed during the MPLS, NAT all of these things, let alone plicable to IP v6 commissioned me to course of the intervening decade both develop an IPv6 strategy for them. the management of address space, have without v6 being available and without it happened and, in their respective ways, (This carrier has operations in Asia, Eu- having become a convincing alternative 5 COOK Network Consultants, 431 Greenway Ave. Ewing, NJ 08618 USA have extended the viability of v4. sion to support it and, whatever the rea- than all of China. sons for doing so, are not really com- COOK Report: And there is now a lot of pletely clear to me. The other parts of the Hussain: Quite true. Take therefore the infrastructure in place that depends on US government don’t seem to care. Japanese position that Japan is very tight what people thought might be only tem- on address space and that it is required porary patches. These patches are in fact In Europe it is very strange to see that the therefore to manage it very carefully. now turning into very permanent looking European Commission is hugely in sup- Consequently it is going to mandate the fixtures. port of v6. They have quite a few initia- use of v6 because doing so frees Japan tives, including a couple of major ones, from any constraints imposed by the ar- Hussain: Yes. As time passes, it be- on-going to push forward the protocol. bitrary nature of the way in which the ini- comes less and less appropriate to call it Meanwhile Japan has long been in favor tial allocations were made. only a patch. I think that if you sudden- of v6 and indeed has become the one ly started telling people that NAT ad- government to actually mandate v6. You The Position of Japan dressing is only a band-aid, they’d look have then a significant portion of the at you as though you were more than a OECD countries in terms of their respec- In this context the most important paper bit loony. It is here and working fine. tive economic power who are in favor of that I would direct your readers to is one When you get into these discussions, you v6. But looking at the over all situation, that they really should read before get- have arguments that are about issues of you must say that the US is not quite ting absorbed into the detail and finer is- technical and architectural elegance. there. That Japan, from the government sues of why v6 is in my view unlikely to People will look you in the eye and say perspective, is totally pushing it. Europe be anything more than a niche protocol. but v6 was designed to have security as is trying to push it and, in fact, there is an This paper was published by Glocom in an integral component. It has auto con- international alliance between the Euro- January of 2002. Its title is “Is IPv6 Nec- figuration as part of its design. We know pean Commission and Japan to endorse essary?” It is by Nobuo Ikeda and Ha- all this but if we look around we are and promote IPv6. jime Yamada. See forced to acknowledge that it still isn’t http://www.glocom.org/tech_reviews/tec here. It has a lot of nice “features’ - yet But looking at all this official support h_bulle/20020227_s2/ The paper is well people still are not using it. you need to ask what is going on here? Is put together with a very balanced argu- it not good enough to get adopted on the ment. But note also that it is from Japan! COOK Report: Shades of OSI! It is the face of things? It is rather unusual to outlook that says I will promise you look at a protocol and proclaim that The authors estimate that we are unlikely everything if only you are patient. somehow it is the key to some economic to run out of v4 address space for anoth- power. Or that it will lead to some terrif- er 15 years – if ever. I haven’t seen this Hussain: It definitely is afflicted with ic economic advantage. It seems to me paper really challenged. When I read the bits of OSI. But the road to v6 started that this outlook is one that fights the last paper, I wondered what would be the EC out in a fever pitch rather more like the battle. It says that the US gained great reaction? Would the EC just quietly de- march to Y2000 fixes because everyone advantage from IP v4 so let’s try to gain fuse its support? There has been an enor- was propelled forward by the idea that comparable advantage from being the mous push back from European ISPs the exhaustion of address space would first with a replacement for v4. who fear that they might be mandated to kill the new-born Internet. Also what deploy v6 just as ISPs in Japan were. In may prove to be the most damaging thing I am not at all sure that this makes much Europe there is push back against the EC for IPv6 is that governments have man- sense anymore because the rational for directive as well as all the hype that you dated its use. One might ask why on v6 is about controlling and managing ad- hear for it. What I do see is that, in Eu- earth they would do this? Why would dress space. Where you find the heaviest rope, the conclusions of the paper are there be official political battles, at the endorsement of v6 is where the routing being wished away. national level on behalf of a communica- registries have the most severe policies. tions protocol? Japan certainly falls into this category. Since the Ikeda -Yamada paper is basi- APNIC pushes v6, but within APNIC, cally a research paper, the proponents of Institutional Proponents Japan pushes especially hard. v6 breathe easier knowing that it won’t fall into the hands of the trade press that of v6 In the initial allocation of v4 address goes on cobbling out simplistic argu- The main source of institutional support space, the claim was and remains that the ments that we better hurry before address for IPv6 now in the US is to be found al- United States allocated address space in space is gone and the huge numbers of most exclusively within the Department such a way that certain countries were wireless users all of whom will have de- of Defense. No one else really battles for left very short changed. vice dependent IP addresses arrive. All it. But even then it is really difficult to these assertions go unchallenged except COOK Report: If you had a Class A ad- within that core community that had seri- say exactly how strongly DoD is really pushing it. Someone has made a deci- dress block and many universities did ous issues with v6 from the very begin- and still do, you had more address space ning. 6 The COOK Report on Internet March 2003 complexity to undergo. To arrive at v6 cation that it is more elegant than what COOK Report: The people with the is- you will need to do serious levels of pro- we have as an alternative. The debate be- sues are those who have v4 infrastructure tocol translation at the edges because ob- tween elegant v6 versus plain old v4 is in place, are running happily and do not viously all but a negligible fraction of beginning to bear the marks of the dis- want to have to make the huge invest- your traffic will be originating and termi- putes of MACs versus their PC brethren. ment in changing? nating as v4. From an operational stand- There is precious little that v6 a decade point, as a large network, saying v6 does ago was designed to do that cannot now Hussain: Yes. But furthermore the huge this and that better than v4 for me makes be done in other ways. You can almost investment in changing would require a no sense because no such a network lives certainly say that there are some things rationale propelling the change. Why in isolation from the Internet. You have cannot be done with v4 in ways that are would we be making a huge financial to be dealing with v4 anyway and what as elegant as those to be afforded by v6. commitment? What would we anticipate you end up with therefore is in effect a The problem was that v6 has simply not our return to be on such an investment? dual direction that is now being pursued. been there for other purposes because its whole design rationale had been driven If you have a large network and are re- So where are we now? I would say that by the warnings of v4 address space ex- quired to implement this protocol, you v6 is a pretty solid protocol. There is a lot haustion. can derive an operational or internal ben- being done to address transition. Most of efit. Or it can come because there is mar- the key core router manufacturers – COOK Report: All the talk was of the 60 ket pull. It is something that customers Cisco, Juniper – and a couple of others MPH collision with the brick wall which want. Now we have been told that cus- such as Hitachi have announced releases because of Cider and DHCP didn’t hap- tomers will want v6. But the window of for v6. They are basically offering their pen. when customers will really want it has routers with dual stacks. Networks that been moving outward now by 2 to 3 deploy v6 will be doing so with dual Hussain: Don’t forget NAT. All of this years every six to eight months. stacks. This means that you will have v4, has become part of a fabric that is global v6, MPLS, and must have a dual stack in scope. If you now try to envisage a Window for Alleged DNS - in short you will have a lot more transition to IPv6 set against this existing complexity to deal with. Heading in this installed infrastructure of v4, I think the Market Pull Keeps direction does not mean that you have Glocom paper not sarcastically suggests Receding chosen a path to operational efficiency that it will take centuries. If there was and cost savings in the core of the net- some market pull, one might say there is When I first started looking at this in the work. But in tough economic times this a rationale for it to happen. year 2000, the period 2001 –2003 was is the direction in which everyone must going to be the big and explosive period head. of IPv6 adoption. Two years later we are Isolated Rational – looking at a period of somewhere be- COOK Report: Well suppose a universi- Wireless 3GPP tween three and five years before there is ty wanted to operate v6 only on its cam- any indication of a recognizable market I think the rationale for IPv6 exists only pus? But even doing it just on its own pull in the wireless arena. Projected pull in very small isolated cases. Let me look campus would increase the cost of opera- that is 3 to five years distant is something at them by putting the small isolated case tion? that is too uncertain to be a reason for us that is the most contentious of the lot to commit to capital expenditure now. In first. Wireless. The wireless environ- Hussain: You have to ask just what it is short I think it quite safe to assert that ment has really had a number of interest- that they would gain from v6? Do they currently, there is no reason to deploy v6 ing twists and turns. V6 has taken a need to run v6 because they don’t know because of market pull. decade to declare that it has solved the how to do NAT? Or because they won’t address space problem by essentially giv- have enough address space? When I was There are ways to implement v6 as tun- ing everyone infinite space. But in paral- evaluating v6, I found a very ambivalent neled within v4 within a backbone net- lel our way of handling address space has position on the part of educational insti- work. You might consider doing this as a become so good that we no longer need tutions. The 6 Net that has six or seven means of gaining experience with it as a the solution that v6 has labored so long to hundred institutions is hosting the net- protocol. Most players out there who say achieve. It is very unclear that we have works in general of small research de- they have v6 are implementing it in this an address space exhaustion problem that partments. I really don’t think that these sort of marginalized way. When you cannot be managed. Moreover we have departments are representative of the look at what operational benefits are to managed it quite well so far. campus network of the entire the univer- be gained by turning a backbone network sity. at the Internet core into an IPv6 network, The other issue is why does every device there are really precious few. To turn a need an IP address and the conclusion is The bottom line is that we are having a backbone network into a v6 network, that it probably doesn’t. So put these two problem in finding a commercial ration- there are actually quite a few levels of things aside and look at what you have. ale for deploying v6 solely on the justifi- You have networks that are carrying 7 COOK Network Consultants, 431 Greenway Ave. Ewing, NJ 08618 USA IPv6, IP SEC, MPLS and Ipv4. My con- of all this, at some point the mobile op- going to be 2007. tention is that in the future IPv6 will be erators decided to have a competition to the smallest niche component of this traf- establish exchange points for mobile In- COOK Report: By 2005 your cell phone fic. ternet operators that were also delivering will sync to other cell phones in the other kinds of Internet services. neighborhood and likely be able to figure COOK Report: But once upon a time out what kind of address grid it is in. In wireless devices were going to need If you were an Internet operator you this sense a geographical addressing sys- fixed addresses. Do we now have the could have an exchange agreement. tem could become possible? equivalent of DHCP for wireless? (Cable and Wireless, the Amsterdam Am Six, Sprint were among those involved.) No Impact Before 2007 Hussain: I think the situation about A whole bunch of mobile operators got wireless is fundamentally unclear and into this group that was interested in cre- But by Then Whole quite contentious. 3GPP, which is the ating exchange points. These mobile op- Nature of Wireless Will third generation mobile project, adopted erators were trying to create an insulated Be Changed IPv6 as their protocol of choice in 1999. domain that was outside of the manage- In doing so they probably gave v6 the ment of the routing registries. The ef- Hussain: Exactly. Other than this belief strongest endorsement that it has ever re- forts never really took hold. No body that we are going to run out of address ceived. It claimed that each cell phone complained about it but also nobody space, the only rationale for v6 is that we would have its own IP address and that pointed out that it was really a very are some how going to have billions of there would be billions of handsets. The flawed approach. mobile users whose operators are going requirement for using IPv6 to handle to need v6. My analysis of this has such addressing issues seemed to make a COOK Report: They were adding anoth- brought me to the estimates of others that lot of sense. But there were a couple of er layer of complexity. claim by 2007 mobile requirements problems. could make an impact. The problem is Hussain: If mobile Internet had actually that by 2007 there will likely be enough COOK Report: For one until a cell started to take hold, I think they would other changes in the way mobile works phone becomes totally digital it doesn’t have seen a problem of huge dimensions. such that no one else will want these need an IP address. Right? IPv6 related capabilities. COOK Report: Why? Hussain: Correct. And furthermore they COOK Report: Because there will be may never get to that point because there Hussain: The exchanges we have now other better and cheaper ways of doing is something else going on with the wire- are just hanging on. Segmenting the it? less operators in terms of their selecting market further into exchanges for just v6 as a protocol. The mobile operators mobile operators would not I think have Hussain; And these are already showing and certainly those outside the United made much sense. I think there were fac- up now. The compelling arguments for States have been very pleased, and right- tors at work here beyond just ones of get- v6 are based on two things. Address ly so for that matter, in terms of their ting IP connectivity to your cell phone space considerations and mobile devel- ability to establish mobile roaming. that had slowed up and disrupted things opments that might represent an uncon- When they approached third generation in the mobile market place. This slowing trollable growth problem that would ex- roaming requirements for data, it was and disruption was I think a fortunate acerbate the address space issue. their intention to have a third generation side effect for those of us concerned wireless network run as an IP network. about the Internet’s strategic direction. COOK Report: If Powell carries the But their idea was there would be the old open spectrum reform forward, history Internet and a new 3GPPInternet with its So now what we actually have to ask is may show that it was this effort that ren- own addressing and its own domains. If whether it will be 2007 when 3GPPstarts dered IPv6 unnecessary. you want to send traffic to it (3GPP) you to happen and we are all going to have to would have to connect to it and peer with be ready with V6 because this is the kind SONY Proclaims v6 it. of forecast date they are asking us to look at right now. Hussain; Precisely. However, here is a There was a moment in time during the final issue. About 18 months ago a high height of the bubble when for the blink of COOK Report: But if we have software SONY executive declared that all future an eye you might have said “my god radios coming on line right now by then SONY devices would be IPv6 address- these people are trying to compete with we shall have software defined cell able and warned that all service providers and over take the global Internet with one phones. had better deploy v6 to be ready to take of their own construction!” They simply advantage of Sony’s roll out. didn’t seem to understand the most fun- Hussain: Exactly. At the beginning of damental points of what they were deal- 2001 they were talking 2005 at the be- The problem is that even if SONY’s ing with in terms of the Internet. On top ginning of 2002 they were saying it was strategy were to work, those v6 devices 8 The COOK Report on Internet March 2003 would have to talk v4 as well because all now have two. worms what would we do? networks will never convert over night. In order for the device to operate howev- One has to ask whether the purported Hussain: Exactly! There are pluses and er the protocol must be transparent and it benefits – address space, security and minuses on both sides but from the per- will have to work as well on v4 as on v6. auto-configuration are worth it. Do these spective of a network operator the worst You are just adding another level of com- benefits outweigh the aggravation of of all worlds is emerging in that they plexity with no real payoff since if it will having to manage two versions of IP in know that they will have to deal with v4 operate v4 there is no real need for it to the network? In other words IP Sec for the vast majority of their customer operate v6. would be nice, but if the cost of getting it and transit traffic but that there will also on a meaningful scale is a multi billion be certain circumstance in which they COOK Report: So even if the networks dollar global reconfiguration program, will have to envision carrying v6. were impressed enough to start investing are there other less expensive ways of en- and even if they had enough cash to do so suring security? The answer is very like- They will have to manage both and they (which they do not), the issue off added ly yes. may or may not make a decision in a operation complexity would suggest that couple of years time that their core net- they don’t go forward with v6? Still No Market Pull work would be v6. Hussain: Yes. In a couple of years you COOK Report: At one point a good v6 COOK Report: Is there any reason why will see all the major core routers with stack in Windows was supposed to bring they might go to v6 in the core? dual v6 and v4 stacks. You will be able on the v6 revolution? to serve customers by setting up v6 tun- Hussain: Only because they routers are nels inside of v4 for those customers Hussain: The irony is that a lot of peo- already enabled and they could do it which for some reason or other just have ple have now come out with good v6 without having to spend any significant to have v6. You could also use MPLS to stacks and the existence of these stacks extra money. Doing it would increase set up a native v6 PVC. But the aggra- isn’t doing anything for anyone. costs and add complexity and in the ab- vation to do this is extensive and even if sence of market pull it is unlikely that it were cheap, the idea that your engi- COOK Report: Once upon a time you they would do it. Right now I don’t see neering team will be eager to rush out were going to use SIP to be able to turn a pull. and embrace v6 just isn’t likely. on and off your home air-conditioner. Now however the air-conditioner and a DREN did exists as a defense research We may well be faced with quite an irony bunch of other stuff sit behind your home network run originally by ATT. They lost if we are faced with the need to run two firewall and you don’t want it to be uni- it and Global Crossing had picked it up. versions of IP on the internet – v4 which versally addressable? But when Global Crossing went bank- will likely never go away and v6 for rupt more than a year ago WorldCom which there may be a few niche markets? Hussain: That is about where we are. having had in the vBNS experience in IPv6 was intended to replace v4. It is un- These applications are suggested as playing with v6 won he contract. But the likely that it will ever achieve this goal. things that are doable with IPv6 but they only market pull here is coming from the But it certainly has its advocates and its are not market driven applications. US DoD that, as I said earlier, is the only niche applications. Given the current di- There is no evidence that there are large substantial advocate for v6 in the USA. rection in which we are going we will not numbers of folk out there who want to do DoD might indeed want vast numbers of have a permanent address for every de- this. devices with fixed v6 addresses. But the vice as envisioned in v6. There could be DoD's requirements are likely to have a some circumstances that include the pos- COOK Report: At one point people were military rationale and not one that will sibility of a global 3GPPnetwork that in- complaining about firewalls holding translate into creating a broad commer- sists on having fixed IP addresses for back end-to-end v6 capable architectures cial market pull for the general imple- every device dependency. So instead of but if we didn‘t have firewalls out there mentation of IPv6. one IP protocol to be managed you will protecting our cable modems from Klez 9 Is IPv6 Necessary? - One Year Later COOK Report: We asked Nobuo Ikeda, As Larry Lessig emphasized in his book one of the authors of Is IPv6 Neces- "The Future of Ideas", it was the E2E ar- The cheapest way to recover the E2E sary?” http://www.glocom.org/tech_re- chitecture that made possible the explo- would be to reallocate v4 addresses of views/tech_bulle/20020227_s2/ to give sive innovations on the Internet. But which only 3% are used. I recommend his opinion on the subject of his paper a today it has drifted far away from this that ICANN to have "address buyouts" year later. He replied principle. The Internet is complicated, to buy back idle addresses by reverse opaque, and controlled by service auctions and sell them through auctions. Ikeda: I suggest you ask comments from providers. Can we take it back to the [Editor: Not trusting ICANN, we are Jun Murai, the global leader of IPv6. He E2E ideal? I'm not optimistic about that. glad that the chances of this happening and I discussed this problem in a recent Today, five years after RFC 2460 that are quite remote. But as a general prin- IETF meeting and agreed that the prob- recommended IPv6, the number of v6 ciple we see the point that Ikeda makes.] lem is not the "shortage" of addresses sites has increased to 1259 from 1046 in I proposed a similar mechanism for but rather applications to take advantage November 2001 - about 180 sites per opening spectrum in my article "The of v6. year among more than 40 million sites Spectrum as Commons". on the Internet. How many millenniums COOK Report: May I send him your does it take to replace v4? See http://www.rieti.go.jp/jp/publications/su suggestion? h t t p : / / w w w . c s - mmary/02030001.html ipv6.lancs.ac.uk/ipv6/6Bone/Whois/ind Ikeda: Yes, but I'm not very much inter- ex.html#full Indeed the problem with IP addresses is ested in v6 because we have reached a similar to that of the spectrum "short- conclusion. Yet it's important to make it Some people argue that "ubiquitous age". In fact, spectrum is not running clear because there are still many people computing" requires IPv6. However, it is short; it is only monopolized by incum- who believe in v6 without knowing it. not v6 but Auto-ID that is prevailing as bents who can't use it efficiently. Internet Our paper on IPv6 was downloaded by the international standard for RFID. And people are accusing the incumbents of 17000 people last year. I don't think it would be so ubiquitous as stifling the innovations made possible by they imagine. Indeed NAT is ugly, but it new radio technologies. I would make a COOK Report: In other words v6 is protects average users from direct at- similar accusation for MIT, Apple Com- dead? tacks on their IP addresses. We even puter, Hewlett-Packard, and other organ- don't receive e-mail by E2E. It's the real- izations that each have more addresses Ikeda: I think it is still alive as an ideal. ity of the Net whether you like it or not. than all of China. 10 IPv6 at the Edges IPv6 Seen Not as a Backbone or Transport Solution But Rather as User-Applied Edge-Based Overlay Supporting End-to-end Applications Highlights COOK Report: Asking for comments, ty. IPv6 has been designed as a protocol ing security between end points not only we sent the Interview with Farooq to Bob that tries to meet the needs of the user do we have a chance of understanding Frankston (one of the developers of Visi- (application) layer and the transport layer what is happening, we can also choose Calc), and to David P Reed, and Francois at the same time. While IPv6 does a rea- our own policies. Barriers between sys- Menard whose names should be quite fa- sonable job at meeting both requirements tems (including firewalls) seem more fo- miliar to our readers. the deployment model is seriously flawed cused on fear than on allowing organiza- because it ignores the dynamics of the In- tions to create value. On January 20, 2003, Bob Frankston ternet as a marketplace driven by the replied: It's very important to distinguish needs of each user. [snip] There is no requirement that the edge between V6 at the edges and V6 in the protocols and the transport protocols be backbone. The reason that V6 is not cur- IPv4 (or just "IP") represented the birth the same. It should be consistent and rently available is that those who are the of the Internet by shifting the power to convenient to leverage common formats. guardians of the net -- the backbone peo- define the network to the users at the Those of us at the edges have already ple are just worrying about their internal edges. paid a high price in waiting on those of us issues and there is no concept hereof ac- who are tweaking IPv6 for use within the tually using the network. The Internet has thrived because supply backbone for the Internet. This continues is driven by demand. New application to be a dysfunctional dependency. We Here is what I wrote last summer: Edge services are supported by simply provid - must learn from the success of the Inter- Protocol (EPv6) rather than IPv6 ing more transport (or IP) capacity. net itself and treat the relationship be- http://www.satn.org/archive/2002_06_30 Rather than wait for new capabilities to tween the IPv6 and what I am calling _archive.html#85208157 be defined, users will create their own so - EPv6 as similar to the separation of UDP lutions. (When I say "users" I don't mean from IP or IP from Ethernet packet for- I recently (June 21st, 2002) spoke at the all users create applications. It only takes mats. IPv6 (Internet Protocol version 6) sum- one motivated, creative individual with mit (http://www.ipv6summit.com/ipv6- some time on their hands to create an ap- Though this is really Edge Protocol ver- program.html). I was invited to speak plication that will be adopted by millions sion v1, I am calling it v6 for marketing about the issues raised in my essay on the of others. We just don't know which user reasons and it looks a lot like IPv6. The Importance of Encrypted IPv6. In that that will be.) big difference is in the requirements. We essay I pointed out we need to assure that can deploy EPv6 on the existing IPv4 In- every system connected to the Internet [snip] The solution is severing the de- ternet now. Not only does this avoid de- has its own (IP) address so that it can be pendency upon IPv6 as a way to meet the pendency upon unmotivated service a full peer participant. Encryption is im- needs of the transport layer. Instead we providers, it also allows us to ignore portant because the separation of the ap- need to focus on the requirements at the those who are trying to build out an IPv6 plication layer (TCP) and the transport edge of the network. network since we shouldn't care whether layer (IP) has been weakened by their efficiencies come from adding ca- providers who are second-guessing the Edge Protocol 6 pacity or clever protocols. In fact, we traffic on the network. should discourage any cleverness in I'm proposing a new protocol called Edge favor of just adding capacity. Despite the urgency there are many who Protocol 6 to give us the benefits of the wonder if we'll ever be able to make the larger address space and simplicity. It Key EPv6 transition from IPv4 to IPv6. gives us the ability to make immediate use of IPV6 technology at the edges Characteristics: The answer is "no" because that is the using the Internet as-is. • Supports a larger address space. Ad- wrong question. The idea of transitioning dresses can be composed using the exist- the entire Internet to a new protocol rep- We must not lose sight of what is really ing IPv4 addresses as a prefix so we can resents a failure to understand that the In- important, namely recovering the sim- use the existing infrastructure. ternet has thrived because it is defined by plicity of the Internet by giving each end its users rather than by a central authori- point a public presence. By implement- • Address resolution can use the existing 11 COOK Network Consultants, 431 Greenway Ave. Ewing, NJ 08618 USA IPv4 DNS Entries (A records) as well as out such a demand isn't very interesting. a priority. This includes being willing to the newer AAAA (IPv6) records. Thus I do whatever it takes even if that means could say V6 in the Backbone and using TCP tunnels through recalcitrant "rmf19.myhouse.frankston.com" where NATs until they can be brought into line. "myhouse.frankston.com" is a V4 ad- V6 at the Edge is Efficiency without connectivity is a form dress part. Entirely Different of death. It's also vital to address other is- sues like working with the existing V4 • Connections are assumed to be encrypt- On January 20 Bob Frankston reminded DNS and extended it with dynamically ed in order to discourage favors from us in response to the IPv6-in-the-back- finding local end points. those who are fixated on "efficiency" and bone focus of our interview with Farooq: other meddlers. And, of course, the edge users must have The purposes of V6 in the backbone and encryption because of all those bell heads • While there are elegant approaches to V6 at the edges do not have any relation- lurking around trying to impose their the problems of NATs (those routers you ship whatsoever. Period. No qualifica- terms of service and other forms of buy for your home), EPv6 implementa- tions. This has lead to the tragedy of the smart-assed meddling. tions can even fall back to TCP tunnels. misperception of a commons. The back- The advantage of TCPis that it maintains bone has indeed accommodated itself to Converting the Internet to V6 is a bad a connection through NATs but the dis- V4 since trying to address each atom on idea. It must be adopted at the edges to advantage is that it can impose arbitrary the net individually is a very big problem meet needs and can coexist just fine with delays and overhead. From a marketing and unnecessary. The IP "address" is like the old Internet. New apps can thrive. It standpoint, however, it means we can use the circuit ID in the phone network and it probably makes sense to have some EPv6 without changing the NATs and encodes a routing though not necessarily V4<=>V6 http but for the most part you that can create a demand for better solu- a precise one. want to go to a v6 server you must have tions. The risk is that the pain won't be a v6 client great enough to force a change but that V6 in the backbone has become a feeding isn't all that bad. frenzy for those who miss the PSTN and In your newsletter it is vital to distinguish want to bring back QoS (AKA discrimi- between EV6 and BV6. Backbone V6 is • EPv6 is meant to enable new applica- nation in favor of legacy traffic and to not at all interesting to me though some tions. Transitioning existing services is a justify maintaining scarcity) and MPLS of your readers may care. As long as BV6 secondary priority though being able to (circuits are forever). There's also the bad is an opportunity to bring back dead access EPv6 web sites from older sys- idea of providing mobility at the IP lay- ideas, however, it may actually make the tems is important but can be done at the ers. (Yes, saying this is in conflict with net worse. Steve Deering agrees that the application level. complaining about temporary IPaddress- ideas are bad though I don't know if he'll es but that's a longer discussion). agree they make the net worse. In short The Internet has been seriously weak- the lack of EEV6 (End/Encrypted) V6 is ened by the need to share IP addresses There is indeed no market pressure form a definite liability for those of us who among a set of computers. We're ten the sheep at the edges so there is no way want an end-to-end edge controlled net- years overdue on remedying the situa- ISPs will make it a priority and waiting work that I open to innovation. tion. The availability of EPv6 is a key for them is pointless. The reason we need part the rebirth of the Internet. The P2P addresses at the edges is to give every David P Reed: Bob doesn't need me to (Peer to Peer, including Instant Messag- end point a first-class public presence on agree with him. His points above are ing and other collaboration tools) com- the net with a modicum of stability. dead on, and in many ways put more suc- munity already represents a significant NATs, and VPNs and Firewalls have al- cinctly than I would be able to. pent up demand that is ready to catalyze ready destroyed the Internet replaced it around a commonly accepted way to pro- with a series of walled realms that don't Needed: an End-to-End vide a large address space with direct trust each other and are constantly being connectivity between systems at the edge invaded. Each invasion is seen as fatal Overlay of the network. The use of encryption thanks to the Maginot line mentality. My only amplification would be that helps assure that the connection is indeed EV6 rides very well over the V4 network those of us who see no need for BV6 or a direct. (modulo NATs) though native implemen- twisty convoluted web of walled gardens with trolls at all the gates may need to re- Transport providers who do want to take tations would be nice. V4 tunneling does volt, and do an end-to-end overlay net- advantage of the IPv6 addresses to sim- have its issues. The reason we don't work of our own (just as the original In- plify routing will also benefit by having a have V6 is that it is a repeat of my expe- ternet was an end-to-end overlay net- demand for their services. The process rience with home network. I am not work). When is a revolution necessary? will start by building on IPv4 but the around to harass the network people to When the current market leaders keep ability of EPv6 will also make it easier to make sure they get rid of excess baggage building instruments of control -- like meet the demand using IPv6. IPv6 with- and recognize a “just-do-it” mentality as NATs, like usage policies that bar certain 12 The COOK Report on Internet March 2003 kinds of uses, and like attempts to charge keep the local addresses constant even if commercially or operationally unwork- merchants a percentage on every transac- your ISP connection flaps but you then able. Just at the moment, its really hard tion on the net -- rather than what the have to make sure you know the purpose for me to see what would commercially users really want. of each address in order to expose appro- motivate any carrier to deploy v6 other priate. I don't see that this should be a big than sensible forward positioning at a COOK Report: How can the users at the issue but it adds complexity in explaining very low level of commitment. edge install their own IPv6? How do the things to users and forestalls "just works" users revolt? What is the cost of doing unnecessarily. Encryption without pre- (3) I certainly agree with the comments so? What can users do on their own? arrangement. I presume it is very doable of David and Bob that the Edge has more What do they have to get an ISP to do? but haven't drilled down. Local and dy- value, rationale, and probable commer- namic DNS -- not sure if the protocols cial viability than Bv6 at this point. Still Frankston: Let me start with the ideal- are in place we'll need to understand the kind of com- ized answer -- the one I believe would be mitment that Microsoft is willing to put the case were I still at Microsoft. It's the But none of this is a killer. Roll your behind v6 as well as others like Sony for same as the one for home networking. It own? Sure. The problem is getting some- example. I'm not sure that I grasp the pic- would "just work". In fact, there is a V6 one with the right time and expertise to ture of the grass roots 'brush fire" that implementation in XP right now but it is do it. In the MS world everyone is wait- might accelerate the promulgation of missing key "just work" elements: ing (and waiting) for Microsoft. Ev6. I fear that Ev6 is caught in the same Linux/Unix? There’s no telling. grip of conflicting government policy, (1) It doesn't try to "just work" and you giant multi-national enterprise, and other have to do all sorts of setting and tweak- I think the first step is to simply get the institutional interests. If Ev6 can make a ing. The .Net server version doesn't seem backbone and edge agendas separate and compelling case for adoption based on a much better. (2) It requires a cooperating then we can make some progress. Too business/commercial rationale, this - NAT and ISP for type 41 packets. It few people understand the Internet and based on the points David has made ear- should support various alternatives even just treat it as a telephone-like shopping lier - applies to a different segment of the if they represent large performance hits. network. There is no pain because legacy market than that of the backbone carriers (3) It doesn't do encryption well or auto- apps work and few people see beyond the though they will be impacted by it. matically or universally and confuses en- capabilities of that past. The one middle cryption with authentication. (4) It does- ground is to write apps for V6 and then Frankston: Just to avoid ambiguity -- n't allow me to simply use my V4 DNS. go over V4 as an accommodation instead the E in EV6 means EdgeV6 or Encrypt- (5) It's concerned about transition instead of thinking "V4" ed Edge V6. I presume the government of simply giving me V6 capabilities. policy question is about Encryption and, Farooq Hussain: I certainly have no at this point, I would make a strong case None of this is fundamental. issue at all - rather I'm in agreement - that an enlightened government would about making a distinction between the push strongly in favor encryption -- were To oversimplify, V6 is simply a 128 bit value of deploying v6 at the Edge (Ev6) we to have an enlightened government. address. It is typically divided into a pre- and the deploying it in the backbone Encryption is merely like expecting peo- fix portion and a local portion. One of (Bv6). There are definitely reasonable ple to take responsibility and lock their these is reserved for a particular way of rationales for placing emphasis on the own doors rather than saying only police using the V4 address as a prefix but it's edge deployment and good reasons to are allowed to have keys. possible to define additional approaches. stay away from Bv6. But assuming that The local portion is arbitrary and can be Ev6 gathers some impulsion in the years Role of Microsoft based on the MAC address, for example. ahead, Backbones will have to route v6 support http v6/v4 and v4 to v6 and vice The reality is that much of this depends There are some difficulties: versa aside from the necessity for dual on an enlightened Microsoft and, to be stack DNS. very very specific, Christian Huitema. The tunneling packets are a different type He's doing everything "just right" but I so may run into barriers. There is no rea- The points that I was trying to emphasize haven't seen a sign of his going beyond son for this to be necessary. Since it is a with Gordon were that: that and separating edge from backbone different type of packet you can't just tell requirements and emphasizing encryp- your NAT to forward the packets to a (1) v6 should stop being viewed as a re- tion and edge deployment. given port. There lots of ways to name a placement for v4 particularly on the machine and you can have global prefix- grounds that there will be address space When I worked there well before Christ- es and local only. Making sure you have exhaustion. ian joined the company there was no the right one for the return path is a chal- awareness about making Internet connec- lenge. It's not clear why this has to be a (2) Having 2 IPprotocols on top of all the tivity simple and I see that as my main problem but there seems to an effort to rest that happened to backbones is both accomplishment. Nevertheless, it's still a have special addresses. One reason is to painful and inelegant, but not necessarily very difficult idea to remain faithful to. 13 COOK Network Consultants, 431 Greenway Ave. Ewing, NJ 08618 USA Consumers never asked for home net- COOK Report: On Friday January 16 I As to complexity. The V4 prefix ap- working and they are not asking for V6. had Comcast cable modem service in- proach means the net works as-is. Native They are asking for higher and higher stalled. Howver my mail comes from V6 routing means that the Ev6 machines firewalls and, as you can see from the fast.net. Fast Net also does my DNS use Bv6 addresses rather than those with MSN ads, they want only bits with good service for cookreport and hosts my web v4 prefixes but otherwise the edge apps intentions to be allowed through. Con- site. Therefore Fast Net is my (Internet work the same. The Bv6 infrastructure is sumers are still asking for more and more Service Provider (isp) and comcast my then able to use native V6 routing rules. broadband for browsing and not for shar- Internet Access Provider (iap.) Farooq would know more about that than ing. I do. I presume they have more explicit Frankston: For now, let's get EEv6 ship- structure that allows the backend routers As to http v6/v4, I would argue that too is ping. And Bv6, doesn't matter to me, to do something better. Although there is entirely edge. Those who care to listen on that's between Farooq and his manager the risk that better might not really be both Internets would have dual ported and shouldn't interest us users though better as things like MPLS and QoS are servers. The big public servers like CNN your readers might care, maybe too unproven conjectures and, I would argue, will do that, especially if they are strong- much. bad ideas. But I presume that those can ly encouraged and the Microsoft IIS and be shed and V6 will put more knowledge Apaches do it automatically. The old V4 COOK Report: Just to be 100% clear. about the net into the bits. But, again, servers will still be accessible to all sys- You meant Edge V6 or Encrypted Edge that's a mixed bag. tems since I don't expect the V6 systems V6. Right? to drop V4 client support even if they Hussain: I agree that the MPLS is prob- don't have a public appearance. I don't The Necessary Tools ably a bad idea in this mix - but it may mind the NATs as a legacy accommoda- prove to be a fact of life in many back- tion -- the problem is there is no alterna- Frankston: When Farooq wrote Ev6, I bones. My feeling is (also based on Bob's tive. wanted to make sure we're explicit. I observations above) that Ev6 *may* would like it to mean Encrypted and drive Bv6 but doesn't have to. Left on Bv6 is then freed to give more routing Edge but one can talk about them sepa- their own backbones don't have sufficient flexibility. In fact one can run V4 over a rately. Encryption of the pipes can be rationale or commercial incentive to go V6 backbone or visa versa. It's just an in- done independently of the applications. to B6. ternal design issue. But why bother with The purpose is to bring back the naive Bv6? To the extent that Bv6 is a better simplicity that allowed us to assume our Francois Menard: Bob's dead on. route description than using a native pre- conversations were not being overheard. fix, then you might as well use it. But if App-to-app encryption is more appropri- I'm trying very hard to get [Canadian] the routing portion is really only used to ate when we want strong security be- municipalities to implement IPv6 open get you to the a Ev6 connection point, tween apps that have their own security access across municipal FTTH networks then it's not as vital. We would greatly re- model. The pipe security is a compro- so that MPLS doesn't squeeze-in and duce the pressure on the address space. mise as is TCP(which gives the apps cir- end-users become required to run PEs. cuits at the price of potentially very long I'm seeing ISP's provide value added There is one big problem in my nice delays vs. UDP). In this context, Encryp- services by offering commercial access story. I don't like the notion of a single tion is then just an edge V6 issue. to tunneling servers on their premises level of V4<=>V6 switching. There is a which bridge to the good old legacy In- point to which the V4 prefix takes you Dual port boxes? You get them now. ternet. For as long as two service from where you switch to your local v6 Every XP with V6 is dual port. It's just providers across two different municipal router. One can nest additional routing that the current V6 implementations are a FTTH system would want to intercon- internally so you can cascade but the bit pain to configure and use and get past the nect with IPv6, there would then be a layout story isn't as clean. Still, even with NATs and you don't have easily usable parallel Internet. that caveat, we must get started and give encryption. But I should've emphasized every system a public presence and, with that we do have dual V6/V4 in XP. May This is in my view a (the) killer app for encryption, bring the meddler-free Inter- complaint is that they have 90% of the IPv6 ... I'm not sure why it seems so dif- net. mechanism there but haven't put the ef- ficult for people to believe in this ... And fort in to the take it all the way to "just if its true in North America, it'll be even Next we can start teaching people that an work". There is a very strong case for more so once open access is properly Internet Access Provider and Internet saying V6 already is deployed but just provided in Japan. Service Provider are unrelated functions not tweaked. And not encouraged. IIS -- and just because ATTBI changes its name Microsoft's Web server and Apache is the I guess that, rather than debating this to Comcast is no reason to change their primary one on Linux. Get those two philosophically once again, I'd rather email address. But that's the dotDNS working smoothly and you have dual throw everything I have into regulatory agenda and more. support everywhere (at least potentially). interventions, like the one which is going out tomorrow in the context of a PartVII 14 The COOK Report on Internet March 2003 of CAIP to the CRTC. This is going to apps (hardware and software) industry TCP gives us those despised circuits but result in IPv6 being deployed. I'm total- just doing something like Ev6 for its at the application level where they pro- ly convinced that it’s going to end the own needs, with or without the IETF. vide some convenience but are still not tight control that incumbents have over Anyone want to form an IEEE commit- intrinsic. The advantage of Ev6 is that it last mile DSLand cable modem in Cana- tee for edge-to-edge scalable overlay leverages the intellectual energy that has da and which intentionally destroys in- network standards? We could call it gone into the Internet protocols and novation through mandating IPv4 and "VigorNet" because it would regain the gives us a minimal commonality that denying IPv6. vigor of the original Internet. happens to parallel the minimalness of the basic IPv4 Internet. So Where’s Market Bob Frankston Jan 22: I want to clari- fy the "Xbox" argument. I compare Edge v6 with MIME which Pull? became the common way to extend People claim we don't need EV6 because email rather than having to choose be- COOK Report: What is to create a mar- we can also work around each problem tween a lot of different ways to transport ket pull? How do you package and ex- on a case- by-case bases. But that case- binary and multipart messages. We did- plain and then ignite that? by-case basis has brought us a mess. n't transition email to MIME, we just Look at how many of the resulting prod- made it available first to those who un- David Reed: Products. Too many prod- ucts require setting proxy information derstood the need and later to those who ucts are limited by NATs. and firewall settings and all sorts of just like pretty stuff. The mistake is to try other stuff. Still other applications re- to transition the existing Internet to V6. If Ev6 had the property that it automati- quire external servers to act as relays. The real need is for enabling the applica- cally ran over NATs, which would not be tions that don't work well though the ex- hard, it would be adopted by lots of new This problem is not only just creeping isting protocols. Depending on how products. Multiplayer video games, CE ossification but also the inability to do deep their V4 assumptions are, we will devices like Tivo and Replay, home se- new things without a lot of arcane find that existing applications can be re- curity cameras, VoIP phones using SIP... knowledge that locks one into the acci- implemented atop V6 with modest ef- all of which are fundamentally "edge-to- dental properties of each of these work- fort. edge" devices (not edge-to-server) and arounds. And as the work-arounds fester the need to traverse NATs is causing the resulting scab is confused as a some- Without V6 we have no synergy for each huge customer support and marketing how necessary part of the environment. new application we have to make new problems. It is reminiscent of the Heidelberg scars arrangements to work around each of the that showed that the student was a great myriad problems. However, with en- Yeah, I know about STUN and MID- swordsman and, by extension a scholar. cryption and the assumption that new ap- COM and UPnP for IGD's, but when all Firewalls have become the condoms of plications aren't hopelessly naïve, we is said and done, Ev6 overlay network- computing and NAT's inability to pass can reposition the firewall as a tempo- ing would do the job better, and be more interesting traffic makes them into fire- rary scab rather than as protection standard, especially since each of those walls. Finally we have the "marketplace" solutions are limited in scope (STUN for assuming that all bits have intrinsic In building our edge architecture, we home networks, but not corporate ones, meaning, and asking for these filters to should then go on to complete the pic- MIDCOM for corporate if only the cor- become omniscient. ture with "dotDNS" so we can avoid porate firewall people didn't believe that making ICANN the ultimate authority new applications are evil, and UPnP All of this works very very well. At least on meaning. Building an Edge IPv6 ar- IGD's are essentially restricted to Wintel by comparison with the ancient world of chitetcure would also subsume much of clients with a fig leaf of a Linux imple- scribes and quill pens and 1990. We can P2P. The P2P effort is about two things: mentation). browse and we can download (a terribly (1) -applications and (2) work-arounds. asymmetric word) and we can use those Each P2P effort has its own novel solu- But I'm not holding my breath for Ev6 old telephones without having t crank tion to tunneling through the barriers and from anyone focused on communica- the magneto. How could anything be its own unique way of generating per- tions (operators, vendors to operators, better? After all, doesn't all this changing sistent handles (names). Mostly these are IETF). The communications industry stuff threaten all that we have? I would just idiosyncratic and poor reinventions seems not to care about enabling new argue that the answer is no. of the common mechanisms and divert products at the edge. They seem to want efforts from actually doing anything in- to control and tax any new innovation, Edge V6 as a Sub teresting. strangling it in its crib. None of this "ris- ing tide lifts all" nonsense for them. :-) Routine Library While I'm a great believer in market- places, I find that they don't automatical- I have some hope for the Consumer One way to think about EV6 is as a com- ly give the optimal path between two Electronics industry and the computer mon subroutine library just like TCP. 15 COOK Network Consultants, 431 Greenway Ave. Ewing, NJ 08618 USA points. If we just look at existing appli- viously we don’t need anything else. purposed as V6 router at the edges of the cations, then, by definition, they do not There should be a clamor to bring back local network. That would give us the "need" v6. Just like we didn't need the In- simple connectivity. Yet everyone seems biggest performance improvements. ternet for incrementally better TV or for to be in love with firewalls, NATs and better faxing. gargoyles of all sorts. All this is com- Taking Advantage of V6 bined with a menagerie of hobgoblins It's only by accident that we got to expe- such as QoS, MPLS and cleverness at The major value of V6 is in allowing rience the web because the Internet had working around problems instead of users to connect devices and not just big lowered the barrier to creativity enough solving them. The Edge V6 need is there. iron and web sites. In doing this, the role for Tim Berners-Lee to hack it together But in order to understand the need, peo- of the DNS in providing a stable handle in his basement office. ple need to understand the Internet first. becomes very important. Not only do we not need the .com semantics, in this en- IPV6 is similar. Those of us who have There are existing implementations of deavor we must be assured that the worked with trying to connect things rec- IPV6 on XP but they don’t “just work”. names are unique and valid basically for- ognize the importance of a consensus Perhaps writing an application that sim- ever. This is all the more reason for cre- that would allow us to have devices that ply does the configuration would go ating a TLD (I call it .DNS) that simply “just connect” instead of constantly hav- along way towards usability. Having provides unique identifiers and NS ing to work around impediments. The done that we would still have to address records – the pointers to the actual DNS marketplace does have this just connect the lack of encryption, the inability to get records which would be maintained by need but some many IPv4 accretions are past recalcitrant NATs, and the inability the owner of the identifier. standing in the way that it can't articulate to leverage the IPv4 DNS entries to name it. interior systems as the norm. Unfortu- Note that mobile IP seeks to provide a nately encryption has gotten entangled stable relationships but it does it at the The simplest solution would be a for a with authentication. I'm not a crypto ex- plumbing layer. Such application level high profile end point player like Mi- pert but we should be able to have a mod- stability belongs at the application level crosoft (more than just like) do to do the est level of crypto between two systems and not the network level. Mobile is an- right thing. Given that the “IP Stack” that don't know each other. other example of an experiment masking business is problematic in the presence of as a basic protocol and it has contributed players who bundle theirs in ways that A secondary problem is that the applica- to the confusion over IPV6. make unbundling extremely difficult, an tion support is uneven but I'm not wor- interim alternative would best be done as ried about that since it can come later. I COOK Report: Above the Fold for Janu- public spirited project, perhaps by stu- want to be able to do simple things like ary 29, commented - “Convinced that dents or others with a need. It can be im- have a VoIP application that just streams large "enterprise" networks of the future plemented as an application level library between two end points and doesn't use will be shaped by the Internet, by ever- written atop UDP – it doesn’t have to be complicated protocols. The fact that such increasing needs for security and mobili- deep in the system. The application applications are not ready attests to the ty, and by the convergence of voice and would then listen on an EPV6 port for importance of early Edge v6 availability data, Hewlett-Packard's new network TCP and UDP connections that serve as so we can work out such problems. strategy is to move more intelligence and the IPv6 tunnel from another system. These are the technical issues. They may control from the core of a network to its have changed greatly in the last year edges, using cheap switches populating This outcome would serve as a V6 shim since I looked at the stuff! those edges.” We sent the url to Bob at the application layer. It can let us as- Frankston and asked for his evaluation. sume V6 while waiting for “official” im- Deployment plementations. Frankston: It is an example of screwing We should build EV6 on the specifica- things up. Notice level four prioritization What is Holding Back tions for BV6 to the extent we can. Doing -- that's a synonym for breaking the end- so will give us an extended address struc- to-end connectivity for users by second Use and Deployment ture. In deployment the most important guessing the applications. And not a The major impediment is a lack of un- step will be to use the IPV4 address as a word about V6 or extending the address- derstanding that the Internet is really routing prefix. We may need an addition- ing model. But lots about security in the about simple end-to-end connectivity al option for a form of routing that is able network which means more and more and the rest is but a detail. But instead of to get past older NATs. speed bumps, twisting passages and viewing the net as a future opportunity, meddling police biddies. we find that there is the normal tendency This is entirely separate from the ques- to confuse it accidental properties with tion of an IPV6 backbone – that is only a Note their press release “Its ProCurve what they could be and because those performance issue. It would be nice, 5300 series switches delivered last sum- properties work then we assume that ob- however, if the NAT boxes could be re- mer, for example, which cost about$65

16
The COOK Report on Internet March 2003
per 10/100 port, implement a broad           said Clark.                                 gence and discrimination into the heart
range of security features as well as                                                    of the network (the edge is in my PC, not
Layer 3 and Layer 4 traffic prioritization   This sure looks like another one of them    in the IT switch) is just more of the old
features. "They let the customer deploy      bellheads run amuck. It would be won-       telco control paradigm.
work's edge at a commodity price point,"     about the network but adding intelli-

Two Internet Futures - With Edge
IPv6 and Without Edge IPv6
By Bob Frankston and from http://www.Frankston.com/public/ESSAYS/EncryptedIPV6.asp

We can loosely separate two agendas:

The backbone agenda is about improving the efficiency of the Internet infrastructure. For the sake of this essay I will only note
these issues to the extent they seem to interfere with the edge agenda.

The edge agenda is about making more addresses available so each host can have a public presence. It is about making more
addresses available as well as improved protocols for automatically assigning addresses. For simplicity I'm focusing on the in-
crease in the number of addresses. IPV6 can be deployed at the edges of the network using the existing IPV4 network as a
transport.

The two agendas are intertwined to the extent that there must be an agreement on the format of an IPV6 packet and the lay-
out of the IP address. But now that there is agreement on the packet format, we can and must deploy IPV6 from the edges.

With and Without
To understand the importance of IPV6 we can compare two scenarios.

Without: If we continue business we will simply accept that the Internet used to be exciting but we have to get back to busi-
ness as usual. Experiments at public access will have mixed results and all-to-often will fail. Hotels will provide some access
but it will be limited and expensive. We will find the Internet is increasingly like television with the transport providers care-
fully selecting which services will work and how well they will work. To most people this won't seem to be a problem and the
economic doldrums will seem to be a higher priority. After all, this is the post Internet era and we should reduce our expecta-
tions.

With: I'll have to tone this down to be taken seriously. But think about being able to take your computer anywhere and it would
just be connected. But why not? Especially if I could just drop an access point anywhere and connect simply and securely.
What might not be obvious is that the kind of "Moore's Law" price/performance improvements that have made email free
(once one has paid for a pipe to the rest of the Internet) would operate to make these access points act as part of a common
good in the same way that we generally allow others to benefit from porch light or a restaurant doesn't charge for tap water.
These aren't free either but it would seem counter-productive to try to charge a passerby who uses that light to read a map. The
key to driving this cycle is simplicity. This is not the post-Internet era. We haven't even started to explore the possibilities.

One lesson I've learned with VisiCalc is that seemingly minor decisions can make a big difference. In making home network-
ing a normal retail product I took a step towards demystifying the Internet and making connectivity just another commodity.
But I was only able to take the first step. I accepted the evil of NATs (Network Address Translation) as I awaited the deploy-
ment of encrypted IPV6.

We have waited too long and there is no reason to wait any more since IPv6 can be deployed from the edges without waiting
for any changes to the Internet itself!

17
Customer Owned Networks
ZapMail and the Telecommunications Industry
by Clay Shirky            Highlights

First published January 7, 2003 on the                                                      better to buy a device to tap directly into
'Networks, Economics, and Culture'            First of all, Federal Express didn't get      that than to allow FedEx to overcharge
mailing list. [COOK Report: See http://       that faxing was a product, not a service.     for an interface to that network that cre-
Shirky.com/nec.html in order to sub-          FedEx understood that faxing would be         ated no additional value. The competitive
scribe to Clay Shirky’s very good mail        cheaper than physical delivery. What          force that killed ZapMail was the com-
list. Republished here with permission.]      they missed, however, was that their cus-     mon sense of its putative users.
tomers understood this too. The impor-
To understand what's going to happen to       tant business decision wasn't when to pay     ZapPhone
the telephone companies this year thanks      for individual faxes, as the ZapMail
to WiFi (otherwise known as 802.11b)          model assumed, but rather when to buy a       The business Fred Smith imagined being
and Voice over IP (VoIP) you only need        fax machine. The service was enabled by       in -- build a network that's cheap to run
to know one story: ZapMail.                   the device, and the business opportunity      but charge customers as it if were expen-
was in selling the devices.                   sive -- is the business the telephone com-
The story goes like this. In 1984, flush
panies are in today. They are selling us a
from the success of their overnight deliv-    Second, because FedEx thought of fax-         kind of ZapPhone service, where they've
ery business, Federal Express announced       ing as a service, it failed to understand     digitized their entire network up to the
a new service called ZapMail, which           how the fax network would be built.           last mile, but are still charging the high
guaranteed document delivery in 2 hours.      FedEx was correct in assuming it would        and confusing rates established when the
They built this service not by replacing      take hundreds of millions of dollars to       network was analog.
their planes with rockets, but with fax       create a useful network. (It has taken bil-
machines.                                     lions, in fact, over the last two decades.)   The original design of the circuit-
This was CEO Fred Smith's next big idea       However, instead of the single massive        switched telephone network required the
after the original delivery business.         build out FedEx undertook, the network        customers to lease a real circuit of copper
Putting a fax machine in every FedEx of-      was constructed by individual customers       wire for the duration of their call. Those
fice would radically reconfigure the cen-     buying one fax machine at a time. The         days are long over, as copper wires have
ter of their network, thus slashing costs:    capital expenditure was indeed huge, but      been largely replaced by fiber optic
toner would replace jet fuel, bike mes-       it was paid for in tiny chunks, at the        cable. Every long distance phone call and
senger's hourly rates would replace           edges of the network.                         virtually every local call is now digitized
pilot's salaries, and so on. With a much                                                    for at least some part of its journey.
less expensive network, FedEx could at-       Finally, because it misunderstood how
tract customers with a discount on regu-      the fax network would be built, FedEx         As FedEx was about faxes, the telephone
lar delivery rates, but with the dramati-     misunderstood who its competition was.        companies are in deep denial about the
cally lower costs, profit margins would       Seeing itself in the delivery business, it    change from analog to digital. A particu-
be huge compared to actually moving           thought it had only UPS and DHL to            larly clueless report written for the tele-
packages point to point. Lower prices,        worry about. What FedEx didn't see was        phone companies offers this choice bit of
higher margins, and to top it all off, the    that its customers were its competition.      advice:
customer would get their documents in 2       ZapMail offered two hour delivery for
hours instead of 24. What's not to love?      slightly reduced prices, charged each         Telcos gain billions in service fees from
time a message was sent. A business with      [...] services like Call Forwarding and
Abject failure was not to love, as it         a fax machine, on the other hand, could       Call Waiting [...]. Hence, capex programs
turned out. Two years and hundreds of         send and receive an unlimited number of       that shift a telco, say, from TDM to IP, as
millions of dollars later, FedEx pulled the   messages almost instantaneously and at        in a softswitch approach that might have
plug on ZapMail, allowing it to vanish        little cost, for a one-time hardware fee of   less capital intensity, must absolutely
without a trace. And the story of Zap-        a few hundred dollars.                        preserve the revenue stream. [
Mail's collapse holds a crucial lesson for                                                  http://www.proberesearch.com/alerts/re-
the telephone companies today.                There was simply no competition. Zap-         focusing.htm]
Mail looked good next to FedEx's physi-
The Customer is the                           cal delivery option, but compared to the      You don't need to know telephone com-
Competitor                                    advantages enjoyed by the owners of fax       pany jargon to see that this is the Zap-
machines, it was laughable. If the phone      Mail strategy.
ZapMail had three fatal weaknesses.           network offered cheap service, it was

18
The COOK Report on Internet March 2003
never materialize.                            mitted to charging the old prices by pre-
Step #1: Scrap the existing network,                                                         tending that they are running an analog
which relies on pricey hardware switch-        The wireless ISPs are likely to fare no       network.
es and voice-specific protocols like Time      better. Most people do their computing
Division Multiplexing (TDM).                   at home or at work, and deploying WiFi        Voice quality is just one
Step #2: Replace it with a network that        to those two areas will cost at worst a
runs on inexpensive software switches          couple hundred bucks, assuming no one         feature among many
and Internet Protocol (IP). This new net-      to split the cost with. There may be a        True to form, the telephone companies
work will cost less to build and be much       small business in wiring "third places" -     also misunderstand the threat from VoIP
cheaper to run.                                - coffee shops, hotels, and meeting           (though here it is in part because people
Step #3: "Preserve the revenue stream"         rooms -- but that will be a marginal busi-    have been predicting VoIPs rise since
by continuing to charge the prices from        ness at best. WiFi is the new fax ma-         1996.) The core of the misunderstanding
the old, expensive network.                    chine, a huge value for consumers that        is the MP3 mistake: believing that users
generates little new revenue for the          care about audio quality above all else.
This will not work, because the cus-           phone companies. And, like the fax net-       Audiophiles confidently predicted that
tomers don't need to wait for the tele-        work, the WiFi extension to the internet      MP3s would be no big deal, because the
phone companies to offer services based        will cost hundreds of millions of dollars,    sound quality was less than perfect. Lis-
on IP. The customers already have ac-          but it will not be built by a few compa-      teners, however, turned out to be inter-
cess to an IP network -- it's called the in-   nies with deep pockets. It will be built by   ested in a mix of things, including acces-
ternet. And like the fax machine, they are     millions of individual customers, a hun-      sibility, convenience, and price. The av-
going to buy devices that enable the           dred dollars at a time.                       erage music lover was willing, even
services they want on top of this net-                                                       eager, to give up driving to the mall to
work, without additional involvement by        VoIP -- Phone calls at                        buy high quality but expensive CDs,
the telephone companies.
internet prices                               once Napster made it possible to down-
load lower quality but free music.
Two cheap consumer devices loom large          Voice over IP is another area where a
on this front, devices that create enor-       service is becoming a product. Cisco          Phone calls are like that. Voice over IP
mous value for the owners while gener-         now manufactures an analog telephone          doesn't sound as good as a regular phone
ating little revenue for the phone compa-      adapter (ATA) with a phone jack in the        call, and everyone knows it. But like
nies. The first is WiFi access points,         front and an ethernet jack in the back.       music, people don't want the best voice
which allow the effortless sharing of          The box couldn't be simpler, and does         quality they can get no matter what the
broadband connections, and the second          exactly what you'd expect a box with a        cost, they want a minimum threshold of
is VoIP converters, which provide the          phone jack in the front and an ethernet       quality, after which they will choose
ability to route phone calls over the in-      jack in the back to do. The big advantage     phone service based on an overall mix of
ternet from a regular phone.                   is that unlike the earlier generation of      features. And now that VoIPhas reached
VoIP products -- "Now you can use your        that minimum quality, VoIP offers one
WiFi -- Wireless local                         computer as a phone!" -- the ATA lets         feature the phone companies can't touch:
networks                                       you use your phone as a phone, allowing       price.
new competitors to offer voice service
In classic ZapMail fashion, the tele-          over any high-speed internet connection.      The service fees charged by the average
phone companies misunderstand the                                                            telephone company (call waiting, caller
WiFi business. WiFi is a product, not a        Vonage.com, for example, is giving            ID, dial-tone and number portability
service, and they assume their competi-        away ATAs and offering phone service          fees, etc) add enough to the cost of a
tion is limited to other service compa-        for $40 a month. Unlike the complex phone that a two-line household that nies. There are now half a dozen compa- billing structures of the existing tele- moved only its second line to VoIPcould nies selling wireless access points; at the phone companies, Vonage prices the save$40 a month before making their
low end, Linksys sells a hundred dollar        phone like an ISP subscription. A Von-        first actual phone call. By simply paying
device for the home that connects to           age customer can make an unlimited            for the costs of the related services, a
DSL or cable modems, provides wireless         number of unlimited-length domestic           VoIP customer can get all their domestic
access, and has a built-in ethernet hub to     long distance calls for their forty bucks,    phone calls thrown in as a freebie.
boot. The industry has visions of the          with call waiting, call forwarding, call
"2nd phone line" effect coming to data         transfer, web-accessible voicemail and        As with ZapMail, the principal threat to
networking, where multi-computer               caller ID thrown in free. Vonage can do       the telephone companies' ability to
households will have multiple accounts,        this because, like the telephone compa-       shrink costs but not revenues is their cus-
but if customers can share a high-speed        nies, they are offering voice as an appli-    tomers' common sense. Given the
connection among several devices with a        cation on a digital network, but unlike       choice, an increasing number of cus-
single product, the service business will      the phone companies, they are not com-        tomers will simply bypass the phone

19
COOK Network Consultants, 431 Greenway Ave. Ewing, NJ 08618 USA
company and buy the hardware neces-            of an internet connection rises with the       suggest that the user should be the
sary to acquire the service on their own.      number of users on the network. Howev-         builder and owner of the network infra-
er, the phone companies do not get to          structure.
And hardware symbiosis will further            raise their prices in return for that in-
magnify the threat of WiFi and VoIP. The       crease in value. This is a matter of con-      The creation of the fax network was the
hardest part of setting up VoIP is simply      siderable frustration to them.                 first time this happened, but it won't be
getting a network hub in place. Once a                                                        the last. WiFi hubs and VoIP adapters
hub is installed, adding an analog tele-       The economic logic of the market sug-          allow the users to build out the edges of
phone adapter is literally a three-plug set-   gests that capital should be invested by       the network without needing to ask the
up: power, network, phone. Meanwhile,          whoever captures the value of the invest-      phone companies for either help or per-
one of the side-effects of installing WiFi     ment. The telephone companies are using        mission. Thanks to the move from analog
is getting a hub with open ethernet ports.     that argument to suggest that they should      to digital networks, the telephone compa-
The synergy is obvious: Installing WiFi?       either be given monopoly pricing power         nies' most significant competition is now
You've done most of the work towards           over the last mile, or that they should be     their customers, because if the customer
adding VoIP. Want VoIP? Since you need         allowed to vertically integrate content        can buy a simple device that makes wire-
to add a hub, why not get a WiFi-enabled       with conduit. Either strategy would allow      less connectivity or IP phone calls possi-
hub? (There are obvious opportunities          them to raise prices by locking out the        ble, then anything the phone companies
here for bundling, and later for integra-      competition, thus restoring their coercive     offer by way of competition is nothing
tion -- a single box with WiFi, Ethernet       power over the customer and helping            more than the latest version of ZapMail.
ports, and phone jacks for VoIP.)              them extract new revenues from their in-
ternet subscribers.                            First published January 7, 2003 on the
The economic logic of                                                                         'Networks, Economics, and Culture'
However, a second possibility has ap-          mailing list.
customer owned                                 peared. If the economics of internet con-
networks                                       nectivity lets the user rather than the net-
work operator capture the residual value
According to Metcalfe's Law, the value         of the network, the economics likewise

20
Discussion of Clay Shirky's ZapMail Essay

Unlicensed, User Financed, Edge Based Connectivity
Technology -- Locustworld Meshbox in Context of
Building Edge Based Wireless Transport Highlights
Editor's Note: From a private mail list       Keweny wites: “While the learned are         share the costs with all his parishioners -
and used with permission, we present a        laughing at Negroponte's fantastic "futur-   without any technical expertise." Editor –
discussion on Clay Shirky's ZapMail           istic" vision of a mesh of interconnected    Readers should without fail read to the
essay.                                        wireless LANs "like lilypads which you       rest of the article at http://www.theregis-
hop from one to another" a UK company        ter.co.uk/content/59/28972.html
Andrew Odlyzko: A great essay. I agree        has produced Mesh wireless technology
with the general conclusions. Still, a few    which you can buy and install, today, for    David P. Reed: Unlicensed, user-fi-
vital points are missing from your piece.     under £300. Fancy setting up as a rival to   nanced wireless technologies potentially
The main one is that ZapMail failed pri-      BT Openworld? Even in a remote vil-          do (allow us to avoid the costs of wiring
marily because FedEx underestimated           lage? Easy: buy a Locustworld Mesh-          up every household.) Wi-Fi is a brand
how popular faxes would be. Initially fax     Box; half the price of a home PC. You're     name for something that is much more
machines were expensive and usage was         in business."                                narrowly defined than that, which does
low. Under those conditions it made                                                        not scale to handle the first mile problem.
sense to offer a service that would be        "The software is the key to Locustworld.     I think Wi-Fi's success may initiate a
used by all those small businesses that       Written by text-message pioneer Jon An-      business model change here. However,
could not afford their own fax machine.       derson, it configures a group of wireless    what you said is like saying the Apple ][
But technology and exploding demand           access points into a coherent "mesh" and     will replace timesharing. At least say
led to dramatic declines in prices, so that   connects them to any broadband Internet      "personal computers" and "unlicensed
business model went out the window.           node available."                             wireless networking technologies".

Another key point that was missing is       "Most experts regard the mesh approach         Andrew Odlyzko: Yes, a very good
that the reason faxing could become a       as hugely complex, because of the effort       point. I was using WiFi just because that
product instead of a service was that it    needed to set up the mesh. The system          was what Clay had in his essay. Other
rode on the existing telephone infrastruc-  used to be known as a "parasitic net-          forms of wireless are likely to be more
ture. What do you think would have hap-     work" - although the fashionable term          important in the long run, as WiFi seems
pened if you had to build a totally new     these days is "symbiotic" - the idea is that   to have too many problems.
physical network for faxes?                 you turn a group of wireless nodes loose,
and tell them to introduce themselves to       On the other hand, I would dispute the
VoIP is indeed destroying the old busi - each other. Then you set up routes                claim that unlicensed user-financed wire-
ness models, and in particular is leading through the mesh. It can be fiendishly           less technologies are the only serious
to a flat-rate pricing regime. However, complex, but Locustworld's mesh does               competitor. Licensed centrally run net-
VoIP by itself does not deal with the this for you. You just buy the node from             works could also be a disruptive factor.
basic problem of providing connectivity. them: the current model is £250 plus              The key issue is that of cost. So far we
It does not lower the costs of the first VAT."                                             have been stuck for the last couple of
mile, which dominate. WiFi potentially                                                     decades in a situation that it cost around
does (and I emphasize potentially, be- "The last legal obstacle, according to              $1,000 per household to provide connec- cause we don't know yet how this will founder Richard Lander, was the deci- tivity, whether it was through copper or play out), since it may allow us to avoid sion by Oftel, allowing people to share coax. Because of monopoly rents, such the costs of wiring up every household. their broadband with up to 20 others. The connections then got valued at$2,000 to
excitement in the UK hasn't been quite as      $5,000 each. However, with wireless COOK Report: What Andrew is calling high as it was in the US, but even there, (and it can be licensed wireless) we may for may already be on the verge of hap- it seems only "nerds" really picked up on potentially get to (1) lower cost per pening. On January 20, 2003 Guy it - probably due to an article by Ander- household, as the need to wire up each Kewney, of Newswireless.net published son which was flagged on SlashDot in residence disappears and (2) a more com- an article, "Become a wireless ISP: for December. It should have hit the head- petitive situation, in which each connec- £300," at the Register web site. The arti- lines big time, since it allows a street to tion is valued closer to the replacement cle certainly gives a partial answer to An- share all their broadband nodes, at a huge cost. drew's question. cost saving. It would allow a vicar in a small village to hire a leased line, and 21 COOK Network Consultants, 431 Greenway Ave. Ewing, NJ 08618 USA “Spectrum Control one could not. Otherwise, it would not COOK Report: Almost three weeks later be "better" enough to generate adoption. (January 28, 2003) on the telecom reg Thinking” So, look for two variables to change at mail list the IRU’s Robert Shaw under- once, both implementation cost and sales lined Shirky’s perception that wireless, Anders Comstedt (CEO of the Swedish channel. One variable alone does not broadband wireline and VoIP could fiber utility) [For wireless to take hold in usually do the trick, tricking analysts. shake up the market in unexpected ways. the local loop, such wireless] really does Look for the second enabler to be the big not need to be "unlicensed" even. The hit. In this case, look for people to adopt Shaw: “The Japanese market is under- question is how low a threshold you can a crummy sound quality in exchange for going an extremely dynamic and com- create for the "local micro group" in really cheap service. Along the way, they plex transformation with regards to IP solving the access problem. Right now will probably accept poor security, laten- telephony. Its genesis is probably in the spectrum control thinking, from ITU cy, down time all in the name of radical- Japan's having decided that the deploy- downwards, prohibits anything but big ly lower costs. In some way, look for dis- ment of broadband networks was ex- business to be players. In fact, I am in intermediation of some middle-man. tremely critical to Japan's future. In Jan- the middle of trying to "sub-license" to uary 2001, a Cabinet-level Task Force on individual users and small groups a Odlyzko: Clay: Another key point that IT Strategy (led by the Prime Minister) chunk of spectrum to use as "fibre exten- ders" where we right now cannot offer was missing is that the reason faxing announced its "e-Japan Strategy". The fiber (suburban, rural). It turns out to be could become a product instead of a strategy set ambitious targets for 30 mil- illegal, not only nationally, but some also service was that it rode on the existing lion households with high-speed Internet claim violation of international spectrum telephone infrastructure. What do you access (e.g., DSL, CATV, FWA) and 10 treaties. think would have happened if you had to million with ultra-high-speed access build a totally new physical network for (e.g. FTTH) within five years.” Odlyzko: Clay, a few vital points are faxes? missing from your piece. “In 2000, the Ministry of Public Man- Shirky: This is probably the biggest agement, Home Affairs, Posts and Shirky: You know that moment where omission. FedEx actually built a propri- Telecommunications (MPHPT), in order you get a cute little Yamaha dirt bike, etary network to handle the data trans- to encourage deployment of DSL, estab- and then you drive it by the Hell's Angels mission, because the planning stages lished rules for local-loop unbundling clubhouse? That's what I felt like posting were pre-Judge Green. This makes and co-location. These rules made it that URL here. FedEx's mistake less confusing -- need- much easier for new ADSL operators to ing a proprietary network for data trans- interconnect with the local networks of All your points above are, of course, cor- mission really did look like a lock-out the incumbent, NTT. Since then, a num- rect, and there were several economic scenario -- but the increase in flexibility ber of new ADSL service providers en- details I glossed over in order to get the of end user devices caught them by sur- tered the market, and you've noted, one Zapmail story into 150 words. FedEx prise. of the most successful has been went into an oil-intensive business in Yahoo!BB (http://bbpromo.yahoo.co.jp), 1973 (ruh roh), and so they were used to Adding that to the article would have which started an ADSL service in Sep- managing inflation, and misunderstood made it more historically accurate, but (I tember 2001 at a very low monthly how quickly technology prices could think) less illustrative for the general charge of JPY 2400 (~US$ 20) for up to
fall.                                          reader, for whom the breakup of ATT is        1.5 Mbit/s connection speed. It also pro-
right up there with stagflation as a mem-     vided specific broadband content &
They also misunderstood that poor qual -       orable event.                                 services, including a VoIP service avail-
ity/high convenience was a tradeoff                                                          able exclusively to subscribers.”
users were willing to make (their fax ma-      VoIP is indeed destroying the old busi-
chines were letter quality at a time when      ness models, and in particular is leading     “Like Hanaro Telecom in Korea,
Type 3 faxes sucked). This mistake was         to a flat-rate pricing regime. However,       Yahoo!BB's entry from outside the nor-
also made by the music industry vis-à-         VoIP by itself does not deal with the         mal telecom circles, energized competi-
vis MP3s, and is now being made by the         basic problem of providing connectivity.      tion among ADSL providers and set a
ILECs vis-à-vis VoIP.                          The "Got WiFi? Why not throw in VoIP?         price benchmark. Most providers'
Want VoIP? Might as well add WiFi" ef -       monthly charges fell to around JPY3000
Roxane Googin: In my opinion, "The             fect is going to make these two technolo -    (US$25), and the quality of service has Crummiest technology always wins". I gies more intertwined than they are been rapidly enhanced from 1.5 Mbit/s use that statement (a quote of myself) today, and I am betting that the inter - to 8 Mbit/s. In the fall of 2002, several often. Also, "the next paradigm never twining happens fast. [Editor's Note - providers started offering access at 12 mimics the prior one". The reason for for us the dynamic was get cable modem Mbit/s. The combination of low prices that is that the new paradigm by defini- in order to get VoIP.] and higher speeds is clearly working. tion is able to solve a problem the old The number of ADSL subscribers in 22 The COOK Report on Internet March 2003 Japan now stands at almost 6 million looked like it would break$1000 per           being asked at this very moment to elim -
with around one half a million being         subscriber, but I haven't followed what        inate unlicensed operation and replace it
added per month (MHPHT stat's are at         came of it. Most recently I noticed anoth-     with WiFi (the trademarked and patented
http://www.soumu.go.jp/joho_tsusin/eng       er form of WLL being deployed by               and IEEE 802.11 controlled standard).
/Statistics/dsl/index.html).”                newly licensed competing fixed opera-          Outlawing any other radio systems in the
tors in India. It's called CDMA (IS-95)        U-NII bands. Letting Microsoft define
“Cable modem service providers have          and it's being provided without cell-to-       what WiFi is or is not.
also been affected by competition from       cell handoffs or roaming, i.e. with "limit-
ADSL. Most CATV operators have up-           ed mobility" (thus not infringing on those     Yeah, stupid people colonize terms. Also
graded their access speed to over 2          with mobile service licenses). The good        very smart people take advantage of that.
Mbit/s and brought down their prices.”       news is that WLL seems to have gotten          Synecdoche is worse than metaphor.
down in price to costing no more than          There is a huge difference between a
“It's interesting that number exhaust was    mobile (by adopting mobile technology).        mandated but "limited" MAC layer (what
mentioned as Japan has now taken an un-                                                     Microsoft says on the Hill) and an open
usual (if not unique?) regulatory ap-        My question (for David or Andrew or            wireless space. It's the difference be-
proach to numbering plan allocation for      anyone): If it were done centrally, are        tween the ITU and the old IETF.
IP terminal devices. Yahoo!BB had been       there any realistic prospects for breaking
bundling VoIP with subscriptions which       the $1000 per subscriber barrier anytime It's not just a question of language pref- allowed free calls to other Yahoo!BB soon? erence. You know that. users and cheap calls to regular phones either in Japan or internationally. How- Odlyzko: That is the big question. The Clay Shirky wrote: David, I'm going to ever, you still had to keep your other costs of all the electronics are coming differ with you on the subject of lan- number/phone for incoming calls as there down, so this should happen, but we re- guage. The general audience, for whom was no way to address/terminate calls. ally need some real data from the field that piece was written, often adopts tech- This is about to change as MHPHT de- (including management costs). nological synecdoche, where an example cided last year to issue telephone num- of a class stands in for the whole class. bers specifically for IP devices (starting David Reed: No. The cost would be (Palm describes many PDAs, the Web with a 050 prefix). In November 2002, very high, because the local laws bar an- means the whole of the publicly accessi- MHPHT handed out ~7 million numbers tennas without full town review. Need- ble internet, and so on.) to ISPs (see ham (my town) is currently holding up a http://www.techinformer.com/english/cr point-to-point high-speed provider by re- Anyone who struggled to write about d_ip_967540.html). Consortiums of quiring that the Zoning Board review "pen-based computing" remembers what ISPs are now deciding they're going to every antenna. The particular antenna it was like when the Palm came along -- work together to terminate calls, in some being blocked is a 6 inch square surface you could get a light to go off in the read- cases for free (see mount patch antenna on a wall of a tall ers head much more easily, but the trade- http://www.idg.net/english/crd_ip_9647 building, facing another tall building off was a loss of generality in the de- 01.html).” with line of sight. The patch antenna scription. When writing for the public, it would be painted the same color as the was often a tradeoff worth making. Declining Price’s Impact building, and looks like a flat-surfaced piece of construction material. In this case, I believe that WiFi has come on Market Viability of to mean "unlicensed wireless networking New Technology What are they doing this for? The answer technologies", and I will not be surprised is money. Because the solution being de- if non-802.11b technologies get deployed Brough Turner: [Referring to Andrew’s ployed costs way less than$1000 per           under that name, or at least described at
emphasis on cost.] A few years ago we        point served, they seem to feel they can       their launch as being "like WiFi."
were providing components to some            tax the hell out of the service providers to
major equipment providers for wireless       bring the cost up to the current cost um-      Jim Forster: I liked Clay's paper too. I
local loop (WLL) systems. At that time, I    brella set by T1's. So the only solution is    made a comment to him on this part and
concluded there was no big upside for        getting customers to finance their own         thought I'd pass it on to you all as well:
Natural Micro Systems as WLLjust plain       networks, and vote out the idiots...
cost too much (2K+) per subscriber. Shirky: Cisco now manufactures an ana- Then three years ago, there was a flurry log telephone adapter (ATA) with a of WLL activity in Asia where the inten- The Problem of Generic phone jack in the front and an ethernet tion of new competitive operators (in Use of Wi-Fi jack in the back. The box couldn't be sim- Hong Kong for example) was to use pler, DECT (cordless technology) or Japanese I am working very hard to prevent the PHS technology to provide WLL within whole unlicensed thing becoming about a Foster: That's the first time I've heard of 30 story apartment buildings. That very limited technology. Legislators are 23 COOK Network Consultants, 431 Greenway Ave. Ewing, NJ 08618 USA cisco product that "couldn't be simpler"! I love competition but I have yet to see a Amazing, I didn't think we knew how to broadband access network that has a low Googin: To address David's and An- make simple things. Well, we're a big transaction cost layer below IP that facil- drew's question as to who runs what place so there's always some lunatics itates "ISPChoice". This was easy for di- when the incumbents get disintermediat- lurking somewhere.... Fortunately! alup Internet because of the switching ed. I have gotten that question since that's fundamental to the PSTN. ILECs 2000. The answer is "who knows"? By the way, folks like Vonage will put a naturally followed that model with DSL While technological trends are somewhat ceiling on the value of the voice service and put ATM in the middle, which is not deterministic, human behavior is not. that can be built on top of a broadband adding particularly any value but is The best we can do is set "boundary con- access service. It's a lot easier to build a adding a lot of complexity. Worse, they ditions", as trying to pinpoint a "solu- competitive voice service on top of never really did Switched Virtual Cir- tion" tends to blind you to alternatives broadband than it is to deal with compet- cuits. outside of our collective limited imagina- itive ISPs on top of broadband access, tion. That is, in my opinion, the most because the former case can take advan- David Isenberg: You've clearly struck a productive line of reasoning is along the tage of IP addressing to reduce to nearly chord with your ZapMail essay -- about a lines of: the next paradigm will most zero the transaction cost of setting up a dozen SMART People have forwarded it likely have these characteristics: (a)peer- voice service, whereas pretty much all to me. [snip] Who's going to run the con- ing, (b)fiber backbones (c)low budgets the schemes for setting up "competitive nectivity network after telephony-classic (d) a way around greedy city govern- ISPs" have significant transaction cost, and cable-TV-classic are dis-intermediat- ments, bla bla bla. How they are imple- since there's no switching layer below IP ed? This is the big unasked question, as mented beyond that will probably be de- in some cases. And adding such a switch- Andrew Odlyzko pointed out. (Fortu- termined by some wild-eyed 20 year-old ing layer just keeps the network from nately for Fedex, the runways are not with tatoos. CHAOS. GO FOR IT. staying dumb. owned by American, United, or the US Post Office.) 24 Open Spectrum - Property Rights World View Dies Hard Exploring the Problems with the Farber-Faulhaber Have-Your-Cake-and-Eat-it- Too Spectrum Arguments Highlights Editor’s Note: In August 2002 universi- would have "easements" that require the ty giveaway. One specific I am looking ty of Pennsylvania Professors Gerry property owners to allow for spectrum for are examples of how entities have Faulhaber and Dave Farber put them- "underlays" using wideband spread spec- used their power to squash similar selves squarely in opposition to the ideas trum or Ultrawideband and maybe also moves. of Open Spectrum and a commons for allow for cognitive radios to utilize local spectrum being explored by the FCC unused, but allocated spectrum. The few I have thought of so far are: spectrum task force with a paper sug- gesting a combination of spectrum auc- They claim that this would deliver the * NAB's uses Congress to eliminate low tions and easements. best of both worlds (the economist's de- power FM sire for market forces to allocate narrow- The whole idea of the open spectrum band, high power spectrum and the engi- * The general hobbling and then repudi- movement is that technology develop- neers desire for a spectrum commons. ation of the Telecom Act of 1996 through ments enable radios to be so sophisticat- lobbying and media manipulation ed that they can stay out of each other’s They even claim with a straight face that way in their use of spectrum. Conse- because of the strength of the commons, * The RBOCs' use of legal, regulatory quently exclusive property rights for that will force the price of spectrum and active non-compliance with the Tele- spectrum are no longer necessary. property to near zero. If this is true, it com Act to facilitate the downfall of seems to me to ask the question as to CLECs Farber and Faulhaber claim to have cob- why bother making it private property in bled together a property rights regime for the first place. * Content Owners' use of DMCA to ma- spectrum that makes consideration of an nipulate ISPs and software companies as open commons politically feasible. The My claim is that their approach is based well as to distort the Internet's end-to- approach is certainly contradictory on on a false assumption that such a market end paradigm the surface. One version of the debate would be open, unbiased and transpar- may be found at http://www.interesting- ent. If history is any guide, the people * RBOCs getting laws passed to block p e o p l e . o rg / a r c h i v e s / i n t e r e s t i n g - p e o- who will get the spectrum as property are Municipally owned telecom networks ple/200206/msg00083.html the top monopolists / oligopolists / lob- byists who will use their market, capital Nobuo Ikeda: As I said, their argument Recently it was debated again on the and political power to eliminate econom- is self-defeating: if the price goes down Open Spectrum mailing list in prepara- ic interference under the guise of elimi- to zero and spectrum becomes commons, tion for a seminar in Tokyo. nating technical interference. nobody will buy the property that will be eventually worthless. Conversely, if On January 9 Robert Berger wrote the The last thing that broadcasters, cellular someone buys it, he will do everything to Open Spectrum mail list: It looks like I operators, RBOCs, DMCA wielding maximize its value by monopolizing it. will be presenting the arguments promot- content owners and other existing giant ing a more pure Open Spectrum Com- corporations who get their monopoly - So the logical conclusion is: they assume mons vs. the Faulhaber/Farber argument oligopoly power from the government that there will be sufficient fools that buy for Property with Easements approach at enforced artificial scarcity will tolerate is the spectrum in "Big Bang" auctions and a GLOCOM Symposium Jan 21 (web making it free and abundant. And they lose money. I think it is an interesting site is have shown over and over that they will economic theory based on "irrational ex- http://w3.glocom.ac.jp/project/wireless/ use every legal, political, economic and pectations". but its all in Japanese) media manipulation tool at their disposal to stop any such competition. Robert Berger: I am sure that Faulhaber The core of the Faulhaber/Farber argu- and Farber are sincere in their belief that ment is that all spectrum should be pri- I would appreciate any suggestions, ad- they have proposed a win-win scenario. I vatized using something like the FCC's ditions or input on how to focus and am- just think its naive to believe that the Kwerel & Williams "Big Bang" auction. plify the support of promoting an Open people who will buy the spectrum (and [They] add that the private property Spectrum Commons vs. a private proper- who think they already own it, that is 25 COOK Network Consultants, 431 Greenway Ave. Ewing, NJ 08618 USA Broadcasters and cellular operators) ial. won't manipulate things to consolidate While [Faulhaber & Farber's] modified their power further. a. If spreading a signal over a range of system is much better than the pure prop- frequencies wider than most "spectrum" erty system, it is still substantially con- The problem is that the folks in power owners currently possess would result in straining to open wireless network de- will use this approach as a Trojan horse such low power spectral density as to not sign, and again it is Coase who helps us to get the government to effectively give displace anyone else's communication understand why. In both his Federal away the spectrum to the incumbents (because the contribution to din the other Communications Commission piece and under the guise of having easements. pair would need to overcome would be in the Nobel-winning article he wrote the Then they will use their power to at least trivial), transaction costs will prevent an following year, The Problem of Social make the easements useless and thus efficient collection of permissions to Cost, Coase introduces the problem of continue to force spectrum to be a scarce spread a signal to that bandwidth. No one the physician and the confectioner who resource. will undertake the transaction costs just are neighbors. The confectioner's equip- to permit a non-displacing, hence zero ment makes vibrations that make it diffi- Larry Lessig: That's certainly one argu- cost, transmission possible. This is rec- cult for the physician to see patients. ment. The problem is that it puts you out- ognized by Faulhaber and Farber as the Normal legal thinking at the time would side your space of expertise. They will anticommons problem, to which they treat the confectioner as "causing" dam- respond about conditions for competi- offer the easement solution. (of which age to the physician by making noise and tion, etc., and that will leave the argu- more below) vibrations. One of Coase's great insights ment at a stalemate. in that article was that the physician is b. It is impossible to price "spectrum" ef- "causing" the damage to the confectioner Why not focus on the costs of their prop- ficiently without computing the displace- by being so sensitive, just as much as the erty system--in particular, the costs it ment effect of all the possible opposite is true. would create for a spectrum commons. network/equipment/communication con- Then the argument looks something like figurations that any given sender receiv- Who should be shut down or made to pay this: (0) Coase said you should only er pair could actually use to achieve a cannot therefore be decided on the basis adopt a property system if the benefits of communication, given their own equip- of stating who is "causing harm," but the systems outweigh the costs; (1) they ment and present cooperative equipment. should rather be based on whose activity acknowledge a property system imposes is more socially valuable. The lesson is costs; (2) if you thought a meshed UWB Only when one does so, can one acurate- directly applicable to the proposition that etc. architecture was ideal, here are the ly state what the social cost of a wireless open wireless networks need not be ad- costs a property regime would impose on communication is. This computation is versely affected by an exhaustive Big getting to this architecture. precisely the computation needed to Bang auction of property rights as long achieve communications in an open as they are permitted to operate without This trades on the intuition that many wireless network. So all property-system interfering with rights owned under that have (in the face of the extraordinary costs associated with identifying de- regime. If, however, we define the oper- growth in commons architectures) that mand, clearing it, and pricing it, will lead ating parameters of open wireless net- the commons might indeed be the best to systematic divergence from efficient works based on the sensitivities of the architecture, without you having to prove pricing. That the direct transaction costs property-based services, we have effec- that the market would not work. of property are higher than those of com- tively treated the property-based system mons is also recognized by Faulhaber & as the physician, and the wide band de- Yochai Benkler: I agree with Larry that Farber, but they do not specify the effect vices and agile radios as the confection- there is no need to resort to the argument or magnitude of these costs--that is, that er. But saying that we will allow confec- that market actors are greedy. The struc- they undermine the one thing that prop- tioners so long as their equipment does ture of the arguments about limitations of erty in spectrum could do right--price not vibrate is not to say that we now property, effect of property on commons, bandwidth. (the main argument of the ar- allow both physicians and confectioners. and transaction costs, that I make in ticle is that commons have more capaci- It is to say that we have chosen to make <http://benkler.org/OwlEcon.html> 30- ty and scale it more rapidly for any given the world safe for physicians and con- 39 is roughly as follows. investment in total network infrastruc- strained for confectioners. This may be ture, except that property could improve the right decision or the wrong decision I Limitations of property sometimes, through efficient pricing. from a social welfare perspective, but it 1. Assumptions: That transaction costs cause systematic is a decision in favor of one approach, a. Demand for wireless communications divergence from efficient pricing is quite not an accommodation of both. is localized and highly dynamic. fatal to that approach). b. Transaction costs of identifying who To be less metaphoric and more specific needs to communicate, with what equip- II. Limitations of easement. Here's a di- let me be clear about the effect of high- ment, at which set of microseconds, in rect quote of the answer I give on pages powered property-based services in a fre- which 30-300 meter bubble, are nontriv- 35-39: quency band on open wireless systems. 26 The COOK Report on Internet March 2003 The level of non-cooperating radiation in spectrum can be solved by technology, it Also, the "big bang" auction provides no any given band affects the extent to should not be solved by market mecha- revisability when technology and appli- which a system needs processing and co- nisms for which social costs are very ex- cations change. WIthout such revisabili- operation gain to achieve a certain rate of pensive. The FCC economists should ty, there is no sensible price for a perma- information delivery through an open prove technically why the spectrum can't nent transfer from the public to today's wireless network. The more radiation be used efficiently as public goods. If investors. Markets cannot discount future there is, the greater the complexity of the they can't prove it, their economic argu- value in the face of technology change solution to the problem of communicat- ments don't make sense. over more than 3-5 years, much less ing information through the channel. The decades. Even if such revisability were greater the complexity of a system, the Berger: The fools aren't the people buy- built in (in the form of, say, 7-year greater the cost of the equipment needed ing the spectrum, its those of us who terms), effective revisability might be to implement it. So, holding all other would allow them to get away with blocked by political power considera- things equal, if you permit only open "owning" what is inherently a public tions as they are today in the case of TV wireless systems to operate in a given commons and then getting to set the licenses. range of frequencies, they will be able to rules. achieve a given throughput at lower cost The problem is that there is no clear eco- than they could if they need to achieve Kevin Werbach: For Farber and Faul- nomic value to selling the right to ex- the same throughput in the presence of haber, easements are a hedge. I don't clude, when simultaneous use is possible high powered communications. think they believe that prices will go to and getting easier. If there is a compro- zero in a commons, or that the commons mise that leads to auctioning some of the While the modified property right is will work at all, but they can't rule it out. spectrum, it's clear that society benefits much better than the perfect property As good economists and engineers, they by making that right revisable, with little rights regime in that it does not com- recognize that if the assumptions open or no cost for the revision. Sure, the price pletely prohibit open wireless systems, it spectrum advocates make are true, a paid to the gov't would be lower, but still imposes a burden on the develop- commons is indeed a better ordering even if the auction guys are right, the ment of those systems. Perhaps the pro- mechanism than a market for most spec- long term cash to the government would ponents of spectrum property rights are trum. be larger as a stream of renewal pay- correct, and that burden is socially justi- ments. fied given the relative value of both types The question at issue here is whether of approachesÅ\the proprietary and the government can create them in a way that And it's clear now (it probably should openÅ\to wireless communications. But ensures they can be viable. The assump- have been in 1934) that spectrum usage the modified property right does not tion is that, by defining the owner's prop- technologies and applications get obso- allow us to eat our cake and have it too. erty rights as stopping at the edge of the lete. Newer, more efficient technologies We must still choose how much we will easement, a win-win solution will should be given a chance to "buy" spec- have of each type of wireless communi- emerge. But that depends on real-world trum if that is what is needed, in compe- cations facility. conditions. At the last open spectrum tition with old technologies. meeting at Harvard, Andy Lippman Ikeda: In economic terms, the spectrum pointed out that the FCC's "interference Thus I would argue that if big bang auc- must be used as (pure) public goods, not temperature" might leave easements too tions are truly the best thing, let's require as commons (common pool resources). narrow to be commercially viable for un- that the purchasers buy their spectrum The difference is that commons is "rival" derlay uses. This becomes an empirical every 3-5 years in a new big bang. i.e., many people can't use the same re- question. It's a narrower one than source at the same time, while public whether spectrum owners will try to poi- Faulhaber: the author [Robert Berger] goods are non-rival. If sufficient spec- son competition in general. understood at least part of our message. trum is opened, it will be non-rival be- Not only do we suggest underlays, but cause we can increase the capacity by David Reed: Spectrum is not like real we also suggest that governments (not adding stations. property (land) in so many ways that the just the Feds) may want to maintain own- "easement" metaphor is inappropriate. ership of large (or small) swaths of spec- Typical public goods are parks. It is ab- trum to hold in commons. In fact, private surd to allocate parks in markets, because Carliss Baldwin recently reminded me firms and foundations may want to do the dividing parks will break its value. The that the human race had been dealing same thing (e.g., a firm may want a com- problem is to maintain its quality by pub- with land rights for many thousands of mons to promote its own phones/radio). lic administration. "Easement" is irrele- years before "fee simple" ownership was This happens today in spectrum (Part 15 vant because parks are totally open. invented. Even then, there are a lot of et al.) and it happens in land (Federal, limitations on what rights exist in land state and municipal parks, even private As Larry and Yochai said, the logical under such ownership. parks and gardens). What makes the au- order is important: if the problem of thor think it will be different in spec- 27 COOK Network Consultants, 431 Greenway Ave. Ewing, NJ 08618 USA trum? more capacity than demand by opening spectrum, it makes no sense to "fill that So is it really just a mater of calling for Noting that we claim "with a straight need". the right kind of easement and maybe face" that marginal price will be near saying, as Benkler does in his paper, that zero, so why should we want a market, Timothy X Brown (University of Col- there should be no big bangs before there this pretty much misses the point we orado) Consider what I'd call "The Dis- is more data? That instead we should have made forcefully: it is both the ef- ney Argument". Property values may not have the equivalent of easements on ex- fects of a market and (perhaps in the go anywhere close to zero even with isting spectrum allocations, allow more long run) the effect of a commons that competing alternatives and having large flexible use of existing allocations by will lead to short- to medium-term near- swaths set aside as a commons. Consid- current licensees and not do any perma- zero prices. But ownership for some uses er Disney who bought a big chunk of nent allocation to private property or makes good economic sense, despite the swampland and created value from noth- commons property for a 10 year period? religious fervor of some commons advo- ing. Disney is able to extract high rents cates, and property rights will fill that and nearby property values in Orlando My particular situation is that I was need. are relatively high. This is despite a asked to be a controversial rebuttal to the plethora of competeing theme parks and Faulhaber/Farber position at a seminar The diatribe about monopolies/oligopo- despite a big chunk of Florida being set that they are also going to be at. So I am lies trying to get their way by influenc- aside as a free park. looking for a bit of an artificially ex- ing government seems naive muckrak- treme position potentially. ing in the extreme. Of course everyone By analogy, if someone with deep pock- tries to influence the government in their ets locks consumers into a wireless serv- Michael Calabrese: You might find the own interest; that's what the Open Spec- ice in a particular band, the value of that attached excerpt from our SPTF Reply trum movement is about, e.g. And of band could be held quite far from zero. Comments helpful (filed by New Amer- course corporations do this, and have al- ica, Consumers Union, Consumers Fed ways done it. And of course the rest of us Berger: Ok, you all convinced me I was of America, Media Access Project, et al), do it as well. Is the author claiming this going in a lame direction. particularly par. 4 below. is some special evil from which we need protection? By what government? The I've been working on how to convey the Licensing will be with us for quite some one in which he gets all the influence? concepts that Benkler and Lessig sug- time; and so our emphasis is to retain the Welcome to Planet Earth. This is the way gested. Its tough to get it in without giv- presumption in current law that (a) the things work here; they always have and ing a tutorial on wireless communica- airwaves are a natural system owned in they always will. Deal with it; nobody's tion, process and cooperative gain. (Ben- common by all Americans (i.e., "public going to rescue you from reality. And kler does a fabulous job in his paper but airwaves"); (b) an exclusive, govern- guess what? It actually seems to work its 53 pages long :-) ment license to communicate impinges pretty well, even if it isn't what you read on First Amendment interests of both the in your high-school civics book. What's more to the point is that its easy licensee and of the public; (c) for both to argue against full on property rights. these reasons - and to preserve the flexi- Ikeda: As usual, Prof. Faulhaber's theo- But I find that, like Benkler's own com- bility of policymakers to adjust spectrum ry is very unique in economics. What ments on the mailing list, its hard to management priorities as technology does "marginal price" mean? He must come up with a strong rebuttal to Faul- changes - the current law requirement have invented it, because there are no haber / Farbers mix of Property Rights that licenses are TEMPORARY and such terms in any textbooks in econom- and Easements unless one wants to sup- transfer NO RESIDUAL PROPERTY ics. port a big bang conversion to all com- INTERESTS beyond the term is critical. mons. (This assumes that the Easements And why does ownership make sense? are big enough i.e. pretty much all spec- In Coaseian terms, there is virtually Maybe because he believes that com- trum and enough power). nothing useful to be gained by transfer- mons will bring about the "tragedy of the ring PERMANENT property rights in commons". It is a wrong application of Its true that property rights limit the full fee simple that could not be accom- Hardin's famous article. He emphasized capability of an open spectrum approach plished with complete but term-limited that the "tragedy" takes place when there in terms of transmit power / range and flexibility (ie, property-like rights) -- is no "technical solution". would require the open spectrum devices but, as you well know, there is a much to to have more computational power than lose. http://dieoff.com/page95.htm if there weren't private property narrow- band high power users. But computa- On January 13, Robert Berger: What So we should investigate whether there tional power is getting cheaper and we're would power management be like if we is a technical solution before we discuss not going to get rid of the incumbents had the ideal world where we had a economic solutions. If we can supply overnight. "pure commons", ie didn't have to worry 28 The COOK Report on Internet March 2003 about legacy issues and not be relegated But since we have enormous control of tual data! to being an "underlay". physical processes today, this is how we may have to phrase the question in the Tim Shepard: Ahhh. I fear that more test Assuming wideband spread spectrum long run. networks and examples would only fur- and/or ultra wideband techniquies. Is It ther confuse things. For almost 3 decades expected that power outputs would be Interference in radio systems is not that computer scientists have been playing much more dynamic in range? For in- well defined independent of a particular with radios and publishing papers extrap- stance in less dense areas where there technology, and it is almost entirely relat- olating what they've learned about the might not be a nearby unit available for ed to architectural limitations of de- particular radios they were playing with. relay, that a device might output in rela- ployed systems, not radiated power. tively high power? Would there be other Reed: I presume you mean that said ra- high power outputs envisioned in an Tim Brown: Many people have looked dios are not the kinds of radio systems Open Spectrum world? at decentralized low-power peer-to-peer that might spur innovative designs be- communication systems. Many of the cause they work quite differently. What happens to spread spectrum and technical elements of such community ultra wideband when there is a mix of networks have been worked out. The I am greatly in favor of exploring alter- low power and high power users? challenge now is how to make them work native systems that might be scalable, with users who do not know each other and testing them in the real world. My fa- David Reed: I don't think the issue is just or necessarily trust each other. Why vorite concept "societies of cognitive ra- power management. Tim Shepard would should anybody carry anyone else's traf- dios" is one such thing that is promising like to consider the worst case where fic? Why should I trust someone else to and ought to be tested - not because it is folks don't cooperate. (but in fact cooper- carry my traffic? Why should I use lower the only option, but because it is promis- ation would evolve by default, if only the transmit power if higher power will bet- ing. cooperation embodied in choosing not to ter ensure my packets get delivered? The "jam" others). key here is that there is no central opera- Theories tell you where to look. You've tor who can police user behavior and still got to test them. And in this case, our We don't know what the best tradeoff be- conversely be sued if service agreements theories incorporate theories about eco- tween the many choices we have might are not met. nomic behavior by future participants. be. Those can only be validated by tests in Thus, beyond the basic technology and the real commercial world. (Remember Information theory and EM physics have scaling issues open spectrum requires: 802.11 was a big surprise in terms of the not given us an "optimality criterion" - it (1) incentives for users to join, cooperate, economic behavior it has stimulated - so is clear, however, that the current archi- and behave well. (2) methods for estab- those who had a theory that 802.11 might tectures are incredibly far from optimum lishing trust between users' radios. (3) be big news ended up proving it by put- as the use of radio communications mechanisms for isolating non-conform- ting it on the market. Prior to that it was scales up. More research is needed. ing users. NOT obvious what would happen with short-range unreliable unlicensed wire- What we think we know about computa- These are tough issues that involve not less). tion is that computation only consumes just radio technology, but, economics, energy when information is destroyed. sociology, and security. Much work For example, if we could only open up Thus computation need not consume en- needs to be done. We are developing a the software radios in today's 802.11a/b ergy. community network testbed that we hope combo chips from Atheros, we could can address some of these challenges. conduct lots of experiments at low cost. What we know about communications is Not all important experiments, of course, less. The core of communications is the Reed: It is clear to me that considering but lots. 2nd law of thermodynamics - entropy in these externalities is important. A major a closed system never decreases. Infor- factor that drives standards is the value of Outcome of Symposium mation is negative entropy. So in a closed interoperability to all participants. Not system, communications is orthogonal calculating that value and incorporating I believe that the Glocom Symposium on (sort of) to energy use. And of course, we it into the model of technology, business, Spectrum Policy don't live in a closed system (we can use and user value leads to bad decisions. In- (http://w3.glocom.ac.jp/project/wireless/ mirrors and solar motors to manipulate teroperability is a "policing" force that if you read Japanese) went quite well. light from the sun to communicate, and exceeds what even a central operator can There were somewhere around 200 – 300 the received photons can potentially achieve. people from telecom companies, Inter- power the detectors). net, vendors, government and the press Berger: Sounds like we need to be build- there. All the presentations are on the I'm sure this isn't the answer you wanted. ing test networks so we can get some ac- website, but in Japanese. I will see if they 29 COOK Network Consultants, 431 Greenway Ave. Ewing, NJ 08618 USA can make the English versions available (Some of these issues might be power ether direction. (This is my interpreta- (they exist, just not at the website). The levels, infringement on property or prop- tion, it may not be correct!) Symposium was simultaneously translat- erty infringement on commons if we ed in English and Japanese. have the easements, etc). How do we Sunao Takatori , president of Yozan Inc I avoid having the FCC step into that role Japan gave kind of a strange speech (He Besides Faulhaber, Farber and myself, and maintaining a regulatory regime? spoke in Japanese, so I heard it through there was Professor Randy Katz from UC translation. Don't know how much that Berkeley who gave a very good overview I proposed that the technically related is- made it strange) His published talk keynote on the "Evolution of Broadband sues such as power levels, any need for looked very interesting about his service Wireless Tech and Community Access. coordination of modulations or protocols which offers IP wireless service via could be done through the IETF or an 802.11 and PHS, but he only talked about Masanobu Suzuki, President & CEO of IETF like process (rough consensus and how the Japanese people should trust NTT Communications gave a talk on running code). And I'm sure the courts their government bureaucrats and not Wireless Broadband, though it was really will get in there one way or another for question authority..... a state of the Japanese broadband indus- the easement stuff... try and NTT Comm's vision of a person- And then Dave Farber followed up with alized Internet service that is independent Then I gave my presentation which I commentary on all of the above. (Gerry of transport. NTT Comm is rolling out a called "Not Time for a Big Bang" where Faulhaber had to leave to catch a plane so pretty aggressive 802.11 HotSpot (They I mainly focused on how it is not appro- there wasn't really a debate). own the trademark for HotSpot in Japan). priate for us to jump right into turning all NTT Comm (which is the long distance spectrum into pure private property, but I did get to talk with Gerry and Dave for and other services portion of NTT when instead look at it as a transition period a while and even though there are obvi- NTT was split similar to ATT/RBOC where we need to not permanently block ous differences, primarily Gerry's strong split) seems pretty progressive in many any future solution and yet must move support for a significant position for pri- ways in their vision, but I think they still forward and start deploying new tech- vate property, I got the feeling that their want to be vertically integrated and offer nologies. You can download a copy of approach is more of a way to make some- a One-stop-solution. Its not clear if that is my presentation at thing happen that breaks the "GOS- one-stop for transport, security and relat- http://www.ibd.com/presentations/No- PLAN" style of spectrum allocation. My ed layers 1 – 3 or so or includes content BigBang.pdf. interpretation is that they also strongly (i.e. all layers). I believe the former. support the commons and they believe Hajime Yamada from Toyo University / that their proposal is a way to make it po- Lindsay Schroth of the Yankee Group Glocom and who I work for here at Glo- litically possible. There is probably more gave a presentation on the state of Broad- com gave a paper that shows what the we have in common with them than with band Wireless with a focus on licensed policy trends are within the Japanese people who want a pure property rights. technologies. FCC equivalent. They are going much more slowly. The good news is that they I also think that Faulhaber likes to argue Then Gerry Faulhaber gave his keynote don't want to jump into turning spectrum :-) which was basically a PowerPoint subset into property via privatization. The prob- of the Faulhaber/Farber paper. He did lem is that they don't want to abandon the If there was anyone else who attended bring up an issue that we in the Open "command and control" structure ether. the symposium, please put in your views Spectrum "camp" should address which They are though calling for investigation of how it went as well. is how will policy or etiquette or whatev- into new spectrum techniques and seem er we call the rules that facilitate the interested. But I would say that they are commons, will be set and coordinated? moving much slower than the FCC in 30 ICANN and the Failure of 'Self Regulation' How the National Science Board was Overruled by the Clique that Became ICANN - Part One by Gordon Cook, Highlights ceptable to Don Heath and Dave Farber but it was and is *something* -- over to of ISOC. During the summer of 1998 ICANN. Although during the past year ICANN Heath, Mike Roberts and Farber were has become almost universally reviled, it gunning for IFWP, and with Sims admin- Professor Post, who has given me his is unfortunately not yet dead. One rea- istering the coup de grace, they killed it. permission to quote him, was not the son is that the fundamental questions of Lessig shows how. only one with a good idea. I have long Internet governance posed by ICANN known what I shall now report. Starting are not yet solved. Six years ago the pri- Lessig also shows how these people took in the summer of 1996, Don Mitchell had mary question was the institutionaliza- advantage of the widely held libertarian pushed this very policy of early dise- tion of the IANAfunction. Today IANA perception that if government did some- nagement at the National Science Foun- works but the question of who pays for it thing, it was, by definition, bad. He dation. In what follows I shall to my best and who controls it is still critical. Also shows how Sim’s quote: "The single uni- to connect the dots to explain how sound today word leaked out from six clicks fying force is that we don't want the gov- policy taken with the best interests of the down in the NTIA web pages that the ernment running things" became the internet at heart was derailed and the IANA function without public discus- mantra behind which ICANN was set up take over by the ICANN clique (elo- sion has for four more years been sole by the IBM, MCI, ISOC ‘old boys.’ quently described by Larry Lessig in his sourced to ICANN. ICANN was created as a tool to ensure October 15 1998 CPSR keynote address) that this clique could maintain control of enabled. ICANN is an on-going travesty. I main- the global Internet. Finally Lessig in his tain that is important to understand how address, prophetically lays bare I have railed against ICANN from the ICANN was created, if we are ever to be ICANN’s anti-internet roots and philoso- very beginning. I can report now that I able kill the one that exists and to avoid phy by summarizing John Gilmore’s re- did so because Don Mitchell had been a similar miscarriages of justice in the fu- marks from September 1998 noting that key inside source on matters of Internet ture. ICANN was to be all about the glo- ICANN had been built in ways that we governance since January of 1995. Don ries of industry self–regulation. Instead completely antithetical to the earlier has given me his permission to be pub- it has given us an excellent demonstra- growth and development of the network. licly identified. Don should be justly tion of what happens when you let the proud of his career at NSF where from fox regulate life in the hen house. Or as ICANN was the aborted “child” of the 1987 on he had a hand in virtually every another observer commented: ICANN privatization process. Some observers important project on which the commer- isn't really about "industry self-regula- looking at the 1996 period have recently cial Internet was founded. [See text box tion." Rather ICANN is an example of wondered why the “feds” just didn’t on page 40 below.] To the extent I have large-industry government collusion to walk away and avoid further involve- been “right” about ICANN, I owe my safeguard mutual self-interests and lock ment. I point out below that it was gov- rightness to the education that he provid- out newer and smaller stakeholders. ernment policy to do just this until the ed. ICANN clique found out and executed a In late December 2002 we discovered veritable coup d’etat. From his joining DNCRI in 1987 to his Larry Lessig’s October 15, 1998 CPSR retirement from government service on keynote address. In this little noticed On January 6 2003 Temple Law School July 13, 2002, Don did his best to defend speech reprinted with Lessig’s permis- Professor David G Post wrote to the the independence of an “end-to-end”In- sion in full below, Larry did three impor- BWG list: As far as I know, I was the ternet described by David Reed in his fa- tant things. He recounted in detail the only person on the planet who called, in mous 1983 paper and later more broadly process of ICANN’s formation in 1998 1996, for the US Government to simply articulated by Larry Lessig in his books including directly identifying the action walk away from its expiring contract Code and Other laws of Cyberspace and taken by Joe Sims in September that tor- with NSI -- to do 'nothing,' to let the sys- the Future of Ideas. Don correctly ex- pedoed the IFWP process that had begun tem find its own 'private' equilibrium, plained to me as early as the the summer in June. The IFWP process in fact was without government intervention. My of '96 that the weak link was the fact that critical because it was the only effort advice was not followed. The US gov- the DNS was becoming a strategic lever aimed at creating what could have been ernment hardly walked away; it handed because the traditional Internet commu- an open ICANN. But an ICANN as cre- something -- and God knows its damned nity hadn't considered names important ated by the IFWP process was not ac- difficult to characterize what it was or is, and had allowed a single point of control 31 COOK Network Consultants, 431 Greenway Ave. Ewing, NJ 08618 USA to evolve. As the Internet bubble began By the summer of 1996 it was obvious did the private and unannounced decision to swell, the sharks who wanted control that the situation was headed down hill. of the NSF get so quickly reversed? I over the action perceived this weakness Mitchell began to draw up recommenda- think it now becomes possible to connect and began to circle Postel, the IANAand tions that the Foundation extricate itself. the dots and I encourage folk like Craig the DNS. In late 1996, a memorandum proposing Simon and Ken Cukier who are writing an "early conclusion" to the NSI cooper- books to do so with thorough research on Don emphasized again and again the ative agreement on March 31, 1997 was their own. problem of institutionalizing the IANA sent to the National Science Board. In a function. The function of the IANA had January 6, 2003 phone conversation Don My Hypothesis grown up in a closed community when reiterated what he had told me in Febru- Internet was largely an academic re- ary 1997. Namely, that The Science In interviews I conducted throughout search project. This authority was never Board Committee on Programs and Plans 1998 with Ira Magaziner, Ira explained formalized and had no legal standing as accepted that approach. Don notes that how as a part of his economic policy the ITU's Robert Shaw showed convinc- this memorandum should now be avail- work on Internet commercialization for ingly in his 1996 analysis. For the last able from the NSF public affairs office. President Clinton he had during the pre- two years of his life, Jon was whip sawed ceding three years. When I asked Maga- to and fro. Finally as Dave Farber point- An early draft of the memo now in my ziner when the critical issue of DNS and ed out to me in an interview in 1999 Jon possession states: “NSF has determined trademarks surfaced for the first time, he was told by his friends in early 1998 that that the best course is to disengage from told me that representatives of two large after the episode of redirecting the root our involvement in and oversight of the corporations came before his bi monthly servers he needed an attorney to keep registration activity. This action will be staff meetings in either late November or him out of jail. The "friends" found Jon taken March 31, 1997, the end on the 4th early December of 1996. They then ar- his attorney. Joe Sims. It was Sims who year of the present award, as the date for gued to argue that, unless trademark in- then took over the process allegedly on our formal disengagement . . .” However fringement in the DNS was brought Jon's behalf and gave us ICANN. Jon because a number of things like the for- under control, the Internet would never died six months after Sims came on the mation of ARIN in order to handle the succeed in becoming the engine of eco- scene and shortly before the first ICANN distribution of IP and ASN numbers had nomic growth for the economy that Ira board was announced. to be completed, the memo informed the had envisioned. I later kicked myself for Board that no advance announcement of not asking Ira which corporations came How to Institutionalize the planned withdrawal would be made. before him. Most likely it was IBM’s As the draft put it “it is important that our Rogger Cochetti and ATT’s Marilyn the IANA plan to withdraw from this activity not be Cade. This meeting seems to have oc- prematurely divulged to the general pub - curred in the aftermath of the National In the mid 1990s Don spent countless hours trying to solve what turned out to lic” in order that the necessary coordina- Science Board’s acceptance of the be an insoluble problem. How to institu- tion between Network Solutions and “the planned termination of NSF involvement tionalize the IANAfunction. Workshops appropriate Internet governing bodies” in DNS. in 1995 and 96 had concluded that this be achieved. was a necessary step but this was more In February 1997 I was briefed by The NSF Tried to Walk than the NSF could have been expected Away Mitchell and invited to a press confer- to take on. To insiders it was becoming ence scheduled to be held on March 18. increasingly clear in 1995 that the Inter- Had the NSB disagreed with what was The early conclusion of the cooperative net Society under Vint Cerf and Don presented in the memo, it would have agreement would be announced on that Heath felt it should become the ruler of objected and NSF efforts to disengage date. Two weeks later on the first Mon- the Internet. How Heath worked togeth- would have ended. It did not object. day of March the press conference was er with Educom and John Patrick’s large- Consequently participants carried away summarily cancelled. Unfortunately, by ly IBM funded GIP to build ICANN in from that meeting news that would have Monday March 2 1997 NSF had been or- collaboration with the trade mark com- substantial impact on the continued de- dered by OMB to cancel its plans. I was munity has been well documented in velopment of the commercial Internet. informed on March 2 that the early con- writing by Michael Froomkin and Milton Planning at the NSF went forward for clusion of the Network Solutions Coop- Mueller. What has not been documented some 90 days when suddenly on March erative Agreement was off. In the nick of is how Mitchell and the National Science 2, 1997 it was cut off at the knees. Poli- time those who would go on to create Foundation, almost freed the US Govern- cy control was effectively removed very ICANN got their claws inserted into the ment from the DNS tar baby before IBM, suddenly from the hands of NSF and right points of the political pressure MCI, ATT and ISOC could create taken over by Kahin, Burr and Magazin- process. Those who would attempt to ICANN as a shield behind which they er. What happened? rule the Internet through ICANN got the could execute their own personal coup. government to reassert control. I have I suspect I know. James Duderstadt, the often wondered what happened. How 32 The COOK Report on Internet March 2003 President of the University of Michigan place. ed, the problem of institutionalizing the was also President of the Science Board. IANA function would have been forced Doug van Howelling was the VP of IT at At year’s end the intellectual property out on an open table (or, possibly made the University and a fervent supporter of contingent was holding a full court press moot) by the demand for (and creation IBM. The NSFnet backbone had been to exert control over the DNS. In De- of) additional TLDs. It might also have built at Michigan by MERIT with joint cember 1996 the Patent and Trademark been forced into the courts. It certainly study partners IBM and MCI. Duderstadt Office held meetings led by Attorney would have become more clear to many himself has an association with IBM Lynn Beresford on how to handle the more people that one of the most critical going back to the days of project Andrew DNS. As 1996 turned to 1997 the IBM, underpinnings of the Internet, the IANA ay Carnegie Mellon. It would have been MCI, ISOC lobby was getting Magaziner function, had no basis in law. Neither do- very easy for Duderstadt to leak word collared and Becky Burr was being taken mestic nor international. If the play had about the highly sensitive plans at NSF from her place at the Federal Trade Com- been open, the high stakes mania that fes- through van Howelling to Cerf and mission and maneuvered into position tered into the Internet bubble might well Patrick who could then very quickly pull under Sally Katzen at OMB in order to have not reached such a fever pitch. The the necessary strings to get to Magaziner. pull the plug on NSF efforts to extricate industry might not have ridden so high Right now this is only a hypothesis. I itself from supervision and control of and fallen so hard. have no proof. I suggest though that DNS. since the existence of the NSB memo is The over ruling of NSF plans for termi- now known those doing formal research Since NSF wanted out, an interim man- nation by Burr and her ISOC clique and could use it to find out. ager had to be created. By April 1997 the resulting extension of that agreement Brian Kahin and Burr were moved to allowed a small number of high stakes One thing is certain. The would be con- NTIA in the Department of Commerce players to keep the game closed. The trollers of the Internet moved with great and put in charge of such a “manager” the game was still closed in June of 1999 speed after the NSF determined that the Interagency Task Force on Domain when in the ICANN board emails we DNS should be in effect declared fully Names. The announced goal was to de- published Esther Dyson, IBM, Vint Cerf mature and commercialized. Such a de- velop a coordinated DNS policy for the and Mike Roberts hatched a strategy to cision would have left IANA in the per- US government. The reality was that get money for ICANN from the venture son of Jon Postel as the sole outside au- Burr and her associates at the Department capitalists of Sand Hill Road by warning thority over Network Solutions Opera- of Commerce were making the decisions them that their investment were in danger tion of the DNS. Either the movement on and giving orders to NSF (which was still if ICANN did not succeed and by meet- the part of those behind ICANN occurred the legal point of authority over NSI) that ing with Tom Kalil in the White House to as a complete coincidence or it occurred the Cooperative agreement not allowed seek support. Today the investments of because word of the coming policy to expire. The chance to extricate gov- the Sand Hill VCs have largely vanished, change had leaked. Circumstantial evi- ernment from the process was lost. We the IANA function is still not institution- dence points to a leak. Judging by what maintain that it was lost because the alized. Indeed today February 3, 2003 happened it seems likely that the IBM, levers of ICANN control were not yet the IANA function was just handed back MCI, ATT, ISOC group decided that they ready. to the same closed group of high stakes had to put a brake on the process ap- players who profess to operate ICANN proved by the Science Board in order to Had the cooperative agreement conclud- with openness. They are lieing. In reali- give themselves time to put ICANN in ed in spring of 1997, as the NSF intend- ty the game is still closed. 33 Governance by Lawrence Lessig Lessig Demonstrates How the Would Be "Self Regulators" Took Control - Part Two of How ICANN Came to Be Highlights Editor's Note: Larry Lessig in the docu- governance is important; the facts are ment could be made. And so we resign ment that follows gives the best overview important. For they should drive us — ourselves to the religion of antigovern- that we have seen of the details under this history, and the facts it tells — to re- ment — to this absurdly naïve thought girding ICANN’s construction in the year learn something our grandparents that if we just privatize everything, all of 1998. In the talk that we republish with learned half a century ago. They should our problems will go away. his permission, he shows how the GIP drive us, that is, to understand what gov- ISOC Clique found in Joe Sims an attor- ernment is for. To understand govern- "The single unifying force is that we ney who enabled them to take advantage ment's role not as some unnecessary ap- don't want the government running of libertarian distrust of government to pendage — the appendix of society, wait- things." create an ICANN that they could use for ing to be excised by an overeager sur- their own narrow ends and brought on geon — but as an institution that makes Now to my surprise, as I've talked about four years feuding and distrust. ICANN possible a certain perspective on social this subject of the new corporation that from the very beginning was broken. life. will govern domain names, I've discov- Such was the distrust of government that ered that not everyone is following the no one would own up to seeing the bro- We have lost this idea, we inheritors of details of this story. Apparently the puri - kenness. Lessig saw it however and his the 21st century. We have lost the ideal tans enacted the Impeachment Clause of analysis of what could be expected from that there is a role for government here. the United States constitution, so until ICANN from the position of hindsight We — especially we who spend too that public flogging is finished, no seri- more than four years later reads like much of our life using electrons to inter- ous attention in the nation can be spared prophecy. act; we — especially we, who still stand for other public issues. amazed at the potential of this wired Governance world; we — especially we, who can't re- So let me review some of the facts, and member a time when there wasn't an un- retell some of the story, of this process CPSR Conference on Internet Gover- derbelly to every story about a hero. We that has led us to the place we are today. nance, October 15, 1998 http:cyber.har- — children of David Lynch, who can't vard.edu/works/lessig/cpsr.pdf help but believe that, just underneath the As I said at the start, for about a year now surface of the sensible, there is a decay the government has been shopping in "The single unifying force is that we that can't be avoided. We listen to the earnest for a way to remove itself from don't want the government running promises of our governors no differently Internet governance. It had at the start, things."1 than Soviet citizens listened to the prom- through the funding of NSF and DARPA, ises of their governors. We, like Soviet supported the work of sometime god of For the past year or so in earnest, and for citizens, have heard it before. "Hope" is the Internet Jon Postel in California, and some time before that, the government not a place; "Hope" was a television then approved the contract with (what has been scurrying to find a way to pass commercial. has become) NSI of Virginia to govern off its role in running the domain name between them the allocation of top level system to a private, nonprofit corpora- In the few minutes that I can ask for your domains. NSI holds the keys to four of tion. It has been scurrying because its attention this morning, I want to think these TLDs — .com, .org, .net, .edu; the contracts with Network Solutions and about this fact about us. I want to think balance was coordinated by Postel at Jon Postel's IANAwere about to run, and about this reality that all of us know — USC. because the theme of the day for both whether Republican or Democrat, Democrats and Republicans seems to be whether political or not. I want to think But as the net grew internationally, and that government cannot run things. about its meaning. For we are at a mo- as questions were raised about this state- ment of history where hope would come supported monopoly, the government de- "The single unifying force is that we only if we could get beyond this despair. cided that it was better simply to step don't want the government running We are at the cusp of a moment when aside. And so just a few months ago, the things." collective judgment should matter. But Commerce Department released a we are disabled from making that judg- "White Paper" that called for the creation The history of this recent privatization of ment; we are convinced no such judg- of a private, nonprofit corporation dedi- 34 The COOK Report on Internet March 2003 cated to the interest of the net as a whole. working groups uniformly rejected the to transform the set of principles around The government, according to the White invitation. Agreement on principles which consensus had been formed into a Paper, was to "ramp down" its involve- would precede debates about lawyers' document—a document that would form ment in the domain name system, and language. And the participants were in- the basis of a new corporation, consistent pass governing functions over to the pri- terested in the principles, not the lan- with the principles of the White Paper. vate sector. guage. And so the IFWPinvited Harvard's Berk- man Center to host a final drafting ses- "The single unifying force is that we So IANA went its own way — partially. sion, where the work of the international don't want the government running It continued to participate in the IFWP meeting could be transformed into a final things." process, as if it was participating as an document. The aim of IFWPwas that this equal in that process, but it also contin- final meeting draw together a representa- The IFWP ued to develop its own draft bylaws. In a tive group from the previous process, as classically Internet-like way, its drafts well as representatives from IANA. Immediately after the White Paper's re- were made public on the IANA web site, IFWP's presupposition was that it would lease, an organization called the "Interna- and comments were solicited from the proceed only if these representatives tional Forum on the White Paper" formed Internet community as a whole. And as a would attend. itself. The organization was sponsored by comment pinged at the right tone, the a gaggle of Internet related interests, and draft was changed. It evolved as ideas But here, the process stalled. For as was committed to convening a series of from the net struck its authors as good. Berkman scurried to see whether a meet- public meetings at which ideas about this ing was possible, and spent hours on the new corporation could be debated. IFWP IANAthus proceeded as IFWPdid to de- phone with potential representatives, held its first meeting in Virginia, and then velop its own view in the way that it IANA made it increasingly clear that subsequent meetings in Geneva, Singa- thought such views should be developed. they would work to subvert any such pore and finally Buenos Aires. And IANA in a process of comments and final meeting. At first they were willing through this movable feast of constitu- drafts that it ultimately controlled; IFWP to talk about the idea of a meeting, so tion-making, as my friend Tara Lemmey in an extraordinarily messy but public long as the meeting was not announced; described it, the IFWP tried to hammer process, with meetings that its directors and then, as time passed, and hence as out a consensus on a set of principles for could not control. Both processes had a time grew short, they became increasing- this new corporation. claim to legitimacy; but each represented ly insistent that no meeting be held. Then the views of the net in a different way. just at the moment when Berkman could At the same time that IFWP was born, wait no longer, IANA's representative however, there were others who were The IFWP process, however, had more [Editor: Joe Sims] announced that he thinking about what the new organization opportunity than IANAto say something had secured from the major interests on should be. Primary among these others significant about what this change repre- the net — the corporate interests, the was the old organizations that were gov- sented. I remember a particularly signifi- technical community, and some segment erning domain names. And here there cant moment, when Ira Magaziner flew of the Steering Committee of IFWP— an were essentially two — IANA, located in to Geneva to give a 10-minute talk, and agreement to resist any such final meet- California, headed by Jon Postel, and then turned around and returned to the ing. With his triumph, IANA's lawyer again, NSI. NSI joined the IFWPprocess United States. Launching the Geneva [Editor: Joe Sims] announced that Berk- at the start; IANA at first was uncertain. meeting, Magaziner said, "I'm going to man and IFWP could hold their final But after the surprising success of this welcome you, and then I'm going to drafting meeting if they wanted, but no first IFWPmeeting, IANAdecided that it leave. Not to insult you by withdrawing one would come. No one would come should participate. And so in Geneva, my attention, but to symbolize just how because through private negotiations, IANAwas full participant in the debates, the United States government conceives the content of which no one really and came to the sessions with draft by- of this process. Our job is to begin these knows, IANA had cut a deal with laws in hand. Representatives stood on discussions, and then get out of the enough people to stop the IFWP the floor of the working groups in Gene- room." His words were met with strong process. va as equals; they argued their ideas as applause, and once finished, he did just equals, and they tried as equals to per- what he said. Jet lagged, and a bit rum- Now, I don't mean to idolize what the suade others that their vision of this new pled, Magaziner left the stage and re- IFWP had produced. And I don't mean to corporation was the best. turned to the airport. pretend that the consensus it had generat- ed was perfect, or thick, or even consis- IFWP was eager for IANA's participa- "The single unifying force is that we tent. There were problems with its tion. But the participants were not eager don't want the government running process, not the least of which the eco- to hand the process over to IANA. When things." nomic problem of assuring representa- Postel offered IFWPhis draft bylaws as a tiveness. Only those with money could basis for their discussion and debate, the When the IFWP process was over, the afford to fly to the most expensive cities steering committee of the IFWP wanted in the world; and many with too much 35 COOK Network Consultants, 431 Greenway Ave. Ewing, NJ 08618 USA money, and too little to say, seemed eager ests; as if private interests don't have ends fronts—familiar problems in a new to fly. they will then pursue. To push the anti- place. Some things are different; the tar- government button is not to teleport us to get of governance is different. But the But I do think there is something signifi- Eden. When the interests of government difficulty doesn't come from this differ- cant about this difference in process, es- are gone, other interests take their place. ent target; the difficulty comes from our pecially as we become enamored of Do we know what those interests are? problem with governance. stakeholder government again. For while And are we so certain they are anything ideas in the IFWP process gained curren- better? Here's what I mean. cy through public debate and public recognition, ideas in the IANA process *** At a conference in Georgia — former The net is governed already. It is gov- gained currency in part at least through Soviet Georgia, that is — sponsored by erned in places by people — by people deals. some western agency of Democracy, an who set the protocols of the space, people Irish lawyer was trying to explain to the who enforce rules on the space; and it is This was not IETF. It was not the product Georgians just what was so great about a governed everywhere by code — by the of a disparate collection of genius geeks, system of judicial review. "Judicial re- software and hardware that sets the archi- thinking through the engineering prob- view," this lawyer explained, "is wonder- tecture of the place, and sets the terms on lem that Internet faced. It was a draft ful. Whenever the court strikes down an which access to the space is granted. crafted by a lawyer — hired to represent act of parliament, the people naturally These governors — these rulers both his interests, and serving a complex set of align themselves with the court, against human and code — impose values on the interests, he hacked out his deal. It was a the parliament. The parliament, people space. Their actions reflect the values of deal, a deal done by a very good lawyer, believe, is just political; the Supreme the space. Their rules are expressed pri- meeting in with many interests, and ne- Court, they think, represents principle." marily through code, but their rules are gotiating, to find an agreement. Doors A Georgian friend was puzzled by this re- expressed also as rules. They give the closed. This was the process that pro- sponse, puppy-democrat that he is. "So space the character it has. The most fa- duced the ICANN draft. It produced why," he asked, "is it that in a Democra- mous of these governors are bodies such something else as well. For this lawyer cy, the people are loyal to the court, a as IETF — rulers with humility, who ex- who succeeded in striking the deal that nondemocratic institution, and repulsed press their law in requests for comments was the ICANN draft—this Washington by parliament, a democratic institution?" — RFCs. These governors, by the way lawyer, skilled, one presumes, in making Said the lawyer: "You just don't under- they act, by their humility, by their re- such deals—is also the author of some- stand democracy." *** spect for excellence—these governors thing else of note to us. He is also the au- give their rules, and the spaces that they thor of this phrase that I have recurred to There is much talk these days about constitute, a certain value. A collective so often in my talk. It was this lawyer something called governance in cyber- value, that has earned it respect. who said, "The single unifying force is space — much talk, followed by obscure that we don't want the government run- questions, and puzzles. It is said that this One would have thought that the values ning things." And in light of the process, idea — this idea of governing cyberspace of this space were values that we should and the freedom it allowed from the sorts — is anathema to our tradition. Who is have some say about — we, people who of constraints that government might ef- cyberspace? Where would it vote? And it populate the net, we whose lives are af- fect, we can begin to see why. is said that this idea — this idea of gov- fected, or taken over by the net, we who erning cyberspace — is abhorrent to cy- depend more upon the net than we do And so this should lead us to ask: When berspace itself. As John Perry Barlow put upon local government. One would have we don't have government running it, in his (maybe our?) Declaration of the thought that these were values that we things; when we unite behind this mantra Independence of Cyberspace: Govern- would have something to say about. But of anti-statism; when we erupt with this ments of the Industrial World, you weary then one wonders, how? How is it that scream of what we don't want — do we giants of flesh and steel, I come from Cy- these values would be values we choose. know what we will have in exchange. berspace, the new home of Mind. On be- How is it that we could choose? How is it When we don't have government, what half of the future, I ask you of the past to that we could have a role when the "sin- will we have? leave us alone. You are not welcome gle unifying force" is that the only mech- among us. You have no sovereignty anism that we've discovered to date for When Government where we gather. imposing collective values on a social space — we call that government — is Steps Aside But our problem is not the problem of the institution that we are all apparently For here's the obvious point: When gov- governance in cyberspace. Our problem devoted to rejecting. ernment steps aside, it is not as if nothing is a problem with governance. There isn't takes its place. When government disap- a special set of dilemmas that cyberspace It is as if the laws of nature were being pears, it is not as if paradise prevails. It's will present; there's just the familiar written; it is as if they were being written not as if private interests have no inter- dilemmas that modern governance con- while we stood by and watched; and as if 36 The COOK Report on Internet March 2003 we could see how these laws will affect sible for the absence of liberty. lation that I have seen I abhor. But what us — affect us more completely than any I find interesting — and the point I think laws of man — yet we still stand won- This is a very old thought. John Stuart we should focus — is why we have such dering, should we have a role in this writ- Mill, for example, was keenly concerned skepticism. What is its nature; what ac- ing? *** One would think the answer with liberty in Great Britain. But his pri- counts for its source? Why are we, like was obviously yes. But the fact is that mary concern was not the liberty threat- the Irish, exhausted by government? most of us would say that here the gov- ened by government. Mill's concern was Why does government seem like the so- ernment should stay away. We modern the threat posed by social norms, or stig- lution to no problem that we now have? Democrats from our well-developed rep- ma, to personal liberty. His book, On resentative Democracy —we, you and I, Liberty, was a corrective — not just to I don't believe that our skepticism about we and the Irish I spoke of — we who excessive government censorship of governance is a point about principle. We otherwise sing of the virtues of Democ- ideas and speech but to excessive private are not, most of us, really Libertarians. racy and freedom and control by the citi- censorship of ideas and speech. He ar- We may be anti-government, but for the zen, we have no faith in what we might gued for a world where liberty was pro- most part we do believe that there are do. We are at a time when the most im- tected from the threats of both private collective values that ought to regulate portant judgments about how this new and public action — from both laws and private action. Our problem is that we world will be made are being made. And from norms. For him the value was liber- don't know by whom, or how. We are yet, we are strangely disabled — immo- ty and his method directed him against weary with governments. We are pro- bilized by ourselves — from making threats against liberty, whatever the foundly skeptical about the product of choices about that new world. Laws are sources. democratic processes. We believe, being written in the code that that space whether rightly or not, that democratic will be, yet we have no idea how we Mill's method should be our own. We processes have been captured by special might participate in the writing of those should ask whether freedom is protected, interests more concerned with individual laws, and little desire to do so. not whether government threatens free- rather than collective value. While we dom. We should ask whether the archi- believe that there is a role for collective We are disabled for two very different tectures of cyberspace protect traditional judgments, we are repulsed by the idea of reasons. One is very lawyerly and, al- values of liberty, and speech, and priva- placing the design of something as im- most by definition, the less interesting of cy, and access — not whether govern- portant as the Internet into hands of gov- the two. That reason goes like this: In the ment is interfering with liberty, and ernments. main the net is private — thankfully so, speech, and privacy, and access. The pri- thankfully built (i.e., not funded) by mary good here is a set of values, not ab- The battle over domain names is a per- someone other than government — but sence of governmental interference inde- fect example. The White Paper called for whether thankfully or not, formally the pendent of those values. And quite often creation of a non-profit corporation, de- net is not government's creation. And so — more than the Libertarians seem keen voted to the collective interest of the net because the Internet is not government's to admit — these values are only protect- as an international whole, with a board to creation, constitutional values that re- ed by a government acting — acting be composed of representatives of stake- strict government need not restrict actors against tyrannies imposed by individuals, holders on the net, and charged with on the net. and by groups. making essentially the policy judgments that IANA had been making. In ex- This limitation in our thought — given to But I said that there were two reasons change, the government was to give up us by lawyers — drives me nuts in its that we were disabled from imposing col- continuing control over the domain name silliness. We are building the most im- lective values on this space — public val- system, and support its transition to an portant jurisdiction since the Louisiana ues that we would otherwise think natu- autonomous, separate entity. Purchase, yet we are building it wholly ral for a government to sustain. The first outside of our constitutional tradition. we can blame lawyers for; but it is the But think for a second about the kinds of There's no reason for this limit — no rea- second that is the most significant. For questions my Georgian friend might ask. son compelled by our history, or com- this is the reason of the Irish. It is this A "non-profit corporation devoted to the pelled by reason itself. skepticism that we all bring to the ques- collective interest"? Isn't that, he might tion of collective governance. It is our ask, just what government is supposed to We Should Focus on unwillingness to think about how "we" be? A board composed of representatives should influence this space; our prefer- of stakeholders? Isn't that what a Con- Liberty ence just to let the space take care of it- gress is? Indeed, if he thought about it, self; because we have so little faith in any my Georgian friend might observe that If our objectives, as a society, are to pro- structure of collective control. I share this corporate structure differs from tect ideals such as liberty, then my claim this skepticism; I am not a naïve New government in only one salient way — is that we should focus on liberty, and not Dealer; I don't have a 100 day plan for that there is no on-going requirement so much on these obsessively legalistic regulating the Internet; most of the regu- of elections. This is policy making, distinctions about who or what is respon- 37 COOK Network Consultants, 431 Greenway Ave. Ewing, NJ 08618 USA vested in what is in effect an independ- ues of free speech, or privacy, values of values of due process, openness, and free ent agency, but an agency outside of due process, or equality, values that de- speech. the democratic process. fine who we are, and which should lead us to ask — if there is not government to Gilmore, however reluctantly, however This is extremely odd behavior for de- insist upon them, then who? So think sadly, has seen the reality of when the net mocrats. That the idea that a governmen- again about ICANN — about the product hits earth. The code of the net will no tal body, whether American or interna- of this domain name debate, and about longer guarantee the values that he, and I, tional, should set this governing policy what we should do now. think fundamental. And so however re- was not even considered is profoundly luctantly, he chose. And he chose to re- interesting about us. It says something I spent a lot of time at the start complain- ject this corporation. I like Gilmore's about us — about where we have come ing about the process that gave rise to method; I like the values he teaches. But in this experiment with Democracy. It re- this new corporation. I cannot help that. I I don't yet share his response. For in this flects a pathetic resignation that most of am a constitutionalist; I am also a demo- grand experiment in "self-government" us feel about the product of ordinary gov- crat; democracy within a constitutional — this pathological urge to rid self-rule ernment. And while I completely share system is all about process. of anything called "government" — there the skepticism, and even disgust, I think is a third way out. We are not limited to it is important to notice how infectious it But I don't think we should reject the choices of the Commerce Department, or has become. We have lost faith in the ICANN proposal merely because of ICANN. We are not constrained to accept idea that the product of representative process. We should not forget that our or reject what has been proposed. In- government might be something more own constitution was erected upon ac- stead, there is a role here for the govern- than mere interest. To steal the opening tions themselves plainly unconstitution- ment in deciding whether this new cor- line from Justice Marshall's last opinion al.3 Madison's counsel then is still true poration lives up to the values that are on the Supreme Court, we believe that now: The test is what was produced, for our tradition. power, not reason is the currency of de- only it can forgive how it was produced. liberative democracy.2 We have lost the Rather than history, it is the future that The role is to insist. The government idea that ordinary government might should be the test for this new corpora- need not simply accept the corporation as work, and so deep is this thought that tion: does it, we should ask, protect the it has been designed; it is not constrained even the government doesn't consider the values that we think important. There are simply to roll over in the face of a set of idea that government might actually have some who say that it does not. John well-typed by-laws. It can say, we will a role in governing cyberspace. Gilmore is a perfect example here. acknowledge you only if you make the Gilmore has earned the respect of the net; following modifications to your struc- I say all this not to excuse. I am explain- his work, and his values, have earned ture. It can insist on changes that would ing how we got here, not justifying it. I him that respect. So consider his struggle make the organization ours. And if the understand the resignation, and the impa- here. No person better embraces the val- drafters accept these modifications, then tience, with governance. But it is an im- ues of the early net — the values of open- the sins of its past notwithstanding, I be- patience that we must overcome. We ness and freedom that seemed wired into lieve this body could be a start. must isolate its cause, and separate it that early net. And much of Gilmore's from its effect. If we hate government, work has been a celebration of the values What are its flaws? I count three, but it's we hate it not because the idea of collec- of that early net — "the net treats censor- only the first that I want to describe tive values is anathema; if we hate Gov- ship as damage, and routes around it," is here.6 The first is accountability. The ernment, we hate it because we have one example; the power of the net to greatest danger of this emerging structure grown tired of the corruption of our gov- crack laws on crypto a second.4 is the insulation it erects against influ- ernment. We have grown weary of its be- ences from the outside. Not all influ- trayal, of its games, of the interests that Gilmore's natural inclination, I suspect, ences; just those influences that don't ex- control. We have grown weary, but we would be to embrace the creation of bod- press themselves in a technical organiza- must find a way to get over it. ies such as IANA. His respect, and affec- tion. The corporation is a closed corpora- tion, for people such as Postel would tion; the board is potentially self-perpet- For we stand on the edge of an era when make him naturally open to the product uating. fundamental choices about what life in of such friends. But last week, Gilmore this space, and therefore, life in real had to make a choice, about whether to This could be changed. As the Boston space, will be like. These choices will be support old friends, or fundamental val- Working Group rightly insisted,7 the cor- made; there is no nature here to discover. ues. And Gilmore, no friend of govern- poration could be constituted with a re- And when these choices are made, they ment generally, chose values.5 In a bal- quirement entrenched in the articles of will be made either with the values that anced but fundamentally correct state- incorporation, pushing it into a member- we hold sacred held influencing the ment, Gilmore declared that we should ship organization. The government choices that are made, or they will be reject these new bylaws of the new should insist upon this change. made ignoring these values. There are ICANN. And we should reject them be- values that we have in this space — val- cause they don't embrace in terms the Here is a role for government to play. Not 38 The COOK Report on Internet March 2003 necessarily in the building of this self- Maine. Over time, of course, all this do this, we will only do it well if we have governing body; not necessarily in the changed. In the struggle leading up to the abandoned this selfindulgent anti-gov- regulation of this self-governing body; civil war; in the battles over reconstruc- ernmentalism. We will only do it well if but at least in the values of this self-gov- tion after that war; in the revolution of in- we develop again a capacity to choose. erning body. And whether or not these dustry that followed that — in all this, the We will need the capacity to say what values reach as broadly as Gilmore rec- sense of individual citizens as Americans values this space is to have. And we will ommends, it is crucially important that grew. In all this, in all the exchanges and need to govern ourselves there. The sin- the government play at least this role. We struggles which were really national, a gle unifying force should be that we gov- may not want government running national identity was born. When citizens ern ourselves there. Whether government things; but government must assure that were engaged with citizens from other runs things or not, we should govern our- the running runs according to the values states, only then was a nation created. selves. Right now, we cannot. This much that are ours. *** When this question about us must change. about domain names is resolved, howev- We stand today just a few years before er, this problem won't go away. For the where Webster stood in 1851. We stand Footnotes: domain name dispute is but the first of a just on this side of being able to say, "I series. And it is the series that pushes us speak as a citizen of the world," without 1 Bloomberg Business News, "Internet to resolve the more general problem. the ordinary person thinking "What a Control Compromise Could Keep US nut." We stand just on the cusp of an ex- Regulators At Bay," September 30, In his rightly famous book,8 Senator istence where ordinary citizens come to 1998. John F. Kennedy tells the story of Daniel know how the world regulates them. Webster, who in the midst of fighting a Where ordinary citizens begin to feel the 2 See Payne v. Tennessee, pact that he thought would divide the na - effects of the regulations of other govern- http://supct.law.cornell.edu/supct/html/9 tion, said on the floor of the Senate, "Mr. ments, as the citizens in Massachusetts 0- 5721.ZD1.html , Marshall, dissenting. President, I wish to speak today, not as a came to feel the effects of slavery, and Massachusetts man, nor as a Northern the citizens in Virginia came to feel the 3 See discussion in Akhil Amar, man, but as an American." When Webster effects of a drive for freedom. Philadelphia Revisited: Amending the said this — in 1851 — the words "not as Constitution Outside Article V, 55 U. a Massachusetts man" had a significance As we, citizens of the United States, Chi. L. Rev. 1043 (1988). that we are likely to miss. To us, Web- spend more of our time, and spend more ster's statement seems perfectly ordinary. of our money, in this space that's not re- 4 See Electronic Freedom Foundation , Who else would Webster be, except an ally part of any particular jurisdiction, Cracking DES (1998). 5See< American? How else would he speak? but subject to the regulations of all juris- http://www.eff.org/pub/GII_NII/DNS_co But Webster's words come on the cusp of dictions — as we spend more time there, ntrol/HTML/19980929 _gil a new time in America. They come just at we will increasingly come to ask ques- more_new_iana.html> the moment when the attention of citi- tions about our status there. We will in- zens in the United States is shifting from creasingly feel the entitlement that Web- 6 I describe the three in their citizenship to a state, to the question ster felt, as an American, to speak about http://cyber.law.harvard.edu/works/lessig of citizenship for the nation. Webster is life in another part of America. But for /c.pdf . speaking just when it becomes possible us, it will be the entitlement to speak to identify oneself apart from one's state; about life in another part of the world. 7 See http://www.mama- as a member of a nation. What will we do then? What will we do tech.com/boston/merged-bwg- when we feel that we are part of a world, bylaws.html . For at the founding, citizens of the Unit- and that the world regulates us? What ed States (a contested concept itself) will we do when we need to make a 8 Profiles of Courage (Memorial Edi- were really citizens of particular states choice about how the world regulates us, tion, 1989). first. They had loyalty and connection to and how we regulate this space? their own states first, because that's where they lived, and their life was deter- My sense is that we will do is just what mined by where they lived. Other states we are beginning to do now. We will cre- were as remote to them as Tibet is to us ate private, nonprofit corporations dedi- — indeed, it is easier for us to go to Tibet cated to the public interest. We will, that that it was for a citizen of Georgia to visit is, create bodies to govern. And when we 39 A Further But Not Unexpected Disgrace on the Part of NTIA & the US Government Editor's Note Published by Rick ICANN and a spokesperson for the heavily suggests that the ICANN-DoC Wesson to the IETF mail list as we we DoC's National Telecommunications Memorandum of Understanding (MoU) going to press on Feb 3rd, 2003 and Information Administration both and the IANA contract are essentially confirmed the extension, although inseparable, and that ICANN is the only US Grants ICANN Extension of Global ICANN general counsel Louis Touton party fit to run IANA. Domain Powers added that no contract has yet been signed. IANA is responsible for main- The NTIA document said that ICANN, By Kevin Murphy taining the definitive list of which organ- having assumed "key resources and izations, individuals, and domain servers associated privatization responsibilities ICANN, which manages policy aspects are associated with approximately 240 under the MoU" is therefore the "only of the internet's domain name system, is country-code top-level domains responsible entity that can continue to to be granted a three-year extension of (ccTLDs), such as .uk, .us, and .fr. provide seamless performance of the its powers to manage the world's coun- IANA functions". As a further link, the try-code domain names, ComputerWire The decision will cause concern to some three-year IANA contract will come up has learned. The US Department of in the international community, particu- for renewal at periods of six months, one Commerce last week quietly published a larly those concerned in the policy year, one year, and six months - paced to document detailing its decision to "sole- aspects of the ccTLD industry. Some coincide exactly with the times the MoU source" the contract for the so-called ccTLD operators had considered a comes up for renewal, Touton and the IANAfunction to ICANN, as opposed to counter-bid for the IANA contract NTIA said. opening the contract for competitive bid- before its March expiration. A statement ding. buried six clicks into a Federal web site [Snip] the desk of every employee in a feder- In addition to his role in NSFs major Highlights of Don al agency. In 1987, he left NSF's Divi- infrastructure activities, he was person- Mitchell's Nearly 30 sion of Grants and Contracts to join ally responsible for the Global School- the newly formed division of Net- house Project, Cornell University's Years at NSF working and Communications Re- (CU-SeeMe) Conferencing Software Including Internet search and Infrastructure (NCRI) development and Cells-in-Frames proj- shortly before the beginning of the ects, the InterNIC, the Internet Scout Infrastructure NSFNET Program. Project (http://scout.cs.wisc.edu), Dave Hughes Wireless Field tests Projects After joining NCRI in 1987, he ap- (http://wireless.oldcolo.com), the Na- Don Mitchell entered civilian federal plied the experience gained from his tional Laboratory for Applied Net- service in 1970 as a management intern previous activities to the networking working Research at the Department of Agriculture. In infrastructure programs in that organi- (http://www.nlanr.net), the Internet 1972, he moved to NSF, and began a zation. His personal experience in- Caching Project career in contracting. In the mid-70s, cludes broad involvement in the pro- (http://ircache.nlanr.net), the Coopera- he developed a system for motion pic- grams and projects of that activity tive Association for Internet Data ture production which was adopted in (which many credit with changing data Analysis (http://www/caida.org), the 1977 for government-wide use, pio- networking from an arcane technology National Center for Data Mining neered the use of support contracting used by a small research community to (http://www.ncdm.uic.edu),the very (before) and cooperative agreements the global Internet we know today). high performance Backbone Network (after) the inception of the Grant and He is also coauthor, with Kimberley Service (http://www.vbns.net) and the Cooperative Agreement Act and man- Claffy and Scott Bradner, of "In whose High Performance Wireless Research aged the first civilian agency SBIR so- domain:name service in adolescence", and Education Network (http://hp- licitation, as well as the first procure- which may be found at http://ksg- wren.ucsd.edu). ment which put a microcomputer on www.harvard.edu/iip/cai/bradner.html 40 Interview, Discussion, and Article Highlights Tools for Edge Control the center that has taken place in the bal- the OECD countries in terms of their re- ance of control within the internet. In- spective economic power who are in - pp. 1-4 deed we have begun a fairly in depth ex- favor of v6. But looking at the over all Full Article - Summary ploration. It is not yet really clear what situation, you must say that the US is not Microsoft will offer in order to make quite there. That Japan, from the gov- page 2 The stark fact is that the blades of edge based IP v6 applications plug and ernment perspective, is totally pushing the VoIP scissors are closing in on the play. Standards would help enormously. it. Europe is trying to push it and, in telco’s cash flow. On the one hand one Five to ten years ago the IETF would fact, there is an international alliance be- blade is the result of large corporations have been the place to turn. Or maybe tween the European Commission and withdrawing voice traffic from the the IEEE. Japan to endorse and promote IPv6. PSTN and running it over their corporate IPnetworks. On the other hand the other At this point the answer appears to pos- But looking at all this official support blade is derived from international VoIP sibly be the Consumer Electronic Asso- you need to ask what is going on here? wholesaling by companies like ITXC ciation. Is it not good enough to get adopted on and activities by thousands of phone the face of things? It is rather unusual to card middle-men hammering long dis- look at a protocol and proclaim that tance rates ever downward. The ability Hussain on IPv6, pp. 5 - somehow it is the key to some economic of the phone companies to charge more 10 power. Or that it will lead to some ter- for a minute of voice traffic than they Full Article rific economic advantage. It seems to could for a minute of data traffic is rap- me that this outlook is one that fights the idly diminishing. Recently the differ- page 5 Some critics now would say that last battle. It says that the US gained ence has been as high as seven to one. part of the problem is that the whole goal great advantage from IP v4 so let’s try to That is if a telco could ake a penny for a of expanded address space is just prop- gain comparable advantage from being minute of data transfer, it could make ping up the established concept that the first with a replacement for v4. seven cents for each minute of voice every device reachable from the Internet transmission. p. 6 In this context the most important needs at least one permanent layer-three address. paper that I would direct your readers to For the most part the seven-cent differ- is one that they really should read before ential is no longer there. Bits are bits. getting absorbed into the detail and finer One cannot really distinguish voice from Ten years ago this was actually not such issues of why v6 is in my view unlikely data bits. That any price difference ex- an unsound approach. We then had this to be anything more than a niche proto- ists at all is increasingly a regulatory ar- idea that the car would have its IP ad- col. This paper was published by Glo- tifact. In two or three years market and dress and that within the car maybe the com in January of 2002. Its title is “Is technology pressures will have driven air conditioning system and carburetor IPv6 Necessary?” It is by Nobuo Ikeda the differential to zero. When this point also needed their own IP addresses. Just and Hajime Yamada. See is reached, the telcos could find their as every house has a phone number, http://www.glocom.org/tech_reviews/te revenues slashed by two thirds. They everything was to have its own IP ad- ch_bulle/20020227_s2/ The paper is will than have all the relevance of Zap dress. But things have turned out rather well put together with a very balanced Mail. argument. But note also that it is from differently. We are much more sensitive to devices and uses being session orient- Japan! page. 4 The center is dead. Forward movement is at the edges. The major ed. And having, as a result, temporary p. 7 Projected pull that is 3 to five years focal point for this issue is IPv6. Farooq addresses. distant is something that is too uncertain Hussain shows why its chance for signif- to be a reason for us to commit to capital p. 6 In Europe it is very strange to see icant deployment in backbones at the expenditure now. In short I think it quite that the European Commission is hugely core of the Internet is effectively zero. safe to assert that currently, there is no in support of v6. They have quite a few However in a discussion with David reason to deploy v6 because of market initiatives, including a couple of major Reed, Bob Frankston, Francois Menard pull. ones, on-going to push forward the pro- and Farooq we are introduced to the con- tocol. Meanwhile Japan has long been cept of V6 at the edge of the network. There are ways to implement v6 as tun- in favor of v6 and indeed has become the We begin to understand how V6, in the neled within v4 within a backbone net- one government to actually mandate v6. hands of end users at the edges of the work. You might consider doing this as You have then a significant portion of network, could redress the shift toward 41 COOK Network Consultants, 431 Greenway Ave. Ewing, NJ 08618 USA a means of gaining experience with it as need an IP address. Right? net -- the backbone people are just wor- a protocol. Most players out there who rying about their internal issues and say they have v6 are implementing it in Hussain: Correct. there is no concept hereof actually using this sort of marginalized way. When you the network. look at what operational benefits are to p. 9 Hussain: Yes. In a couple of years be gained by turning a backbone net- you will see all the major core routers Here is what I wrote last summer: Edge work at the Internet core into an IPv6 with dual v6 and v4 stacks. You will be Protocol (EPv6) rather than IPv6 network, there are really precious few. able to serve customers by setting up v6 http://www.satn.org/archive/2002_06_3 To turn a backbone network into a v6 tunnels inside of v4 for those customers 0_archive.html#85208157 network, there are actually quite a few which for some reason or other just have levels of complexity to undergo. To ar- to have v6. You could also use MPLS to I recently (June 21st, 2002) spoke at the rive at v6 you will need to do serious set up a native v6 PVC. But the aggra- IPv6 (Internet Protocol version 6) sum- levels of protocol translation at the edges vation to do this is extensive and even if mit (http://www.ipv6summit.com/ipv6- because obviously all but a negligible it were cheap, the idea that your engi- program.html). I was invited to speak fraction of your traffic will be originat- neering team will be eager to rush out about the issues raised in my essay on ing and terminating as v4. From an op- and embrace v6 just isn’t likely. the Importance of Encrypted IPv6. In erational standpoint, as a large network, that essay I pointed out we need to as- saying v6 does this and that better than We may well be faced with quite an sure that every system connected to the v4 for me makes no sense because no irony if we are faced with the need to run Internet has its own (IP) address so that such a network lives in isolation from two versions of IP on the internet – v4 it can be a full peer participant. Encryp- the Internet. You have to be dealing with which will likely never go away and v6 tion is important because the separation v4 anyway and what you end up with for which there may be a few niche mar- of the application layer (TCP) and the therefore is in effect a dual direction that kets? IPv6 was intended to replace v4. transport layer (IP) has been weakened is now being pursued. It is unlikely that it will ever achieve this by providers who are second-guessing goal. But it certainly has its advocates the traffic on the network. p. 7 The other issue is why does every and its niche applications. Given the device need an IP address and the con- current direction in which we are going Despite the urgency there are many who clusion is that it probably doesn’t. So we will not have a permanent address for wonder if we'll ever be able to make the put these two things aside and look at every device as envisioned in v6. There transition from IPv4 to IPv6. what you have. You have networks that could be some circumstances that in- are carrying IPv6, IP SEC, MPLS and clude the possibility of a global 3GPP The answer is "no" because that is the Ipv4. My contention is that in the future network that insists on having fixed IP wrong question. The idea of transition- IPv6 will be the smallest niche compo- addresses for every device dependency. ing the entire Internet to a new protocol nent of this traffic. So instead of one IP protocol to be man- represents a failure to understand that the aged you will now have two. Internet has thrived because it is defined p. 8 COOK Report: But once upon a by its users rather than a central authori- time wireless devices were going to need One has to ask whether the purported ty. IPv6 has been designed as a protocol fixed addresses. Do we now have the benefits – address space, security and that tries to meet the needs of the user equivalent of DHCP for wireless? auto-configuration are worth it. Do (application) layer and the transport these benefits outweigh the aggravation layer at the same time. While IPv6 does Hussain: I think the situation about of having to manage two versions of IP a reasonable job at meeting both require- wireless is fundamentally unclear and in the network? In other words IP Sec ments, the deployment model is serious- quite contentious. 3GPP, which is the would be nice, but if the cost of getting ly flawed because it ignores the dynam- third generation mobile project, adopted it on a meaningful scale is a multi billion ics of the Internet as a marketplace driv- IPv6 as their protocol of choice in 1999 dollar global reconfiguration program, en by the needs of each user. [snip] In doing so they probably gave v6 the are there other less expensive ways of strongest endorsement that it has ever re- ensuring security? The answer is very IPv4 (or just "IP") represented the birth ceived. It claimed that each cell phone likely yes. of the Internet by shifting the power to would have its own IP address and that define the network to the users at the there would be billions of handsets. The edges. requirement for using IPv6 to handle Edge v6 p. 11 - 16 such addressing issues seemed to make a Full Article The Internet has thrived because supply lot of sense. But there were a couple of is driven by demand. New application Bob Frankston replied: It's very impor- problems. services are supported by simply provid - tant to distinguish between V6 at the ing more transport (or IP) capacity. edges and V6 in the backbone. The rea- COOK Report: For one until a cell Rather than wait for new capabilities to son that V6 is not currently available is phone becomes totally digital it doesn’t be defined, users will create their own that those who are the guardians of the 42 The COOK Report on Internet March 2003 solutions. (When I say "users" I don't p. 14 Francois Menard: Bob's dead on. p. 16 We should build EV6 on the speci- mean all users create applications. It only fications for BV6 to the extent we can. takes one motivated, creative individual I'm trying very hard to get [Canadian] Doing so will give us an extended ad- with some time on their hands to create municipalities to implement IPv6 open dress structure. In deployment the most an application that will be adopted by access across municipal FTTH networks important step will be to use the IPV4 millions of others. We just don't know so that MPLS doesn't squeeze-in and address as a routing prefix. We may need which user that will be.) end-users become required to run PEs. an additional option for a form of routing I'm seeing ISP's provide value added that is able to get past older NATs. p. 12 On January 20 Bob Frankston re- services by offering commercial access minded us in response to the IPv6-in-the- to tunneling servers on their premises p. 17 With V6 I'll have to tone this backbone focus of our interview with Fa- which bridge to the good old legacy In- down to be taken seriously. But think rooq: ternet. For as long as two service about being able to take your computer providers across two different municipal anywhere and it would just be connected. The purposes of V6 in the backbone and FTTH system would want to intercon- But why not? Especially if I could just V6 at the edges do not have any relation- nect with IPv6, there would then be a drop an access point anywhere and con- ship whatsoever. Period. No qualifica- parallel Internet. nect simply and securely. What might tions. This has lead to the tragedy of the not be obvious is that the kind of misperception of a commons. The back- This is in my view a (the) killer app for "Moore's Law" price/performance im- bone has indeed accommodated itself to IPv6 ... I'm not sure why it seems so dif- provements that have made email free V4 since trying to address each atom on ficult for people to believe in this ... And (once one has paid for a pipe to the rest the net individually is a very big problem if its true in North America, it'll be even of the Internet) would operate to make and unnecessary. The IP "address" is like more so once open access is properly these access points act as part of a com- the circuit ID in the phone network and it provided in Japan. mon good in the same way that we gen- encodes a routing though not necessarily erally allow others to benefit from porch a precise one. I guess that, rather than debating this light or a restaurant doesn't charge for philosophically once again, I'd rather tap water. These aren't free either but it V6 in the backbone has become a feed- throw everything I Have into regulatory would seem counter-productive to try to ing frenzy for those who miss the PSTN interventions, charge a passerby who uses that light to and want to bring back QoS (AKA dis- read a map. The key to driving this cycle crimination in favor of legacy traffic and p. 15 One way to think about EV6 is as a is simplicity. This is not the post-Internet to justify maintaining scarcity) and common subroutine library just like TCP. era. We haven't even started to explore MPLS (circuits are forever). There's also TCP gives us those despised circuits but the possibilities. the bad idea of providing mobility at the at the application level where they pro- IP layers. (Yes, saying this is in conflict vide some convenience but are still not Customer Owned with complaining about temporary IPad- intrinsic. The advantage of Ev6 is that it dresses but that's a longer discussion). leverages the intellectual energy that has Networks & ZapMail gone into the Internet protocols and Full Article pp. 12-13 David P Reed: Bob doesn't gives us a minimal commonality that need me to agree with him. His points happens to parallel the minimalness of p. 18 The business Fred Smith imagined above are dead on, and in many ways put the basic IPv4 Internet. being in -- build a network that's cheap to more succinctly than I would be able to.. run but charge customers as it if were ex- I compare Edge v6 with MIME which pensive -- is the business the telephone My only amplification would be that became the common way to extend companies are in today. They are selling those of us who see no need for BV6 or email rather than having to choose be- us a kind of ZapPhone service, where a twisty convoluted web of walled gar- tween a lot of different ways to transport they've digitized their entire network up dens with trolls at all the gates may need binary and multipart messages. We didn't to the last mile, but are still charging the to revolt, and do an end-to-end overlay transition email to MIME, we just made high and confusing rates established network of our own (just as the original it available first to those who understood when the network was analog. Internet was an end-to-end overlay net- the need and later to those who just like work). When is a revolution necessary? pretty stuff. The mistake is to try to tran- p. 19 Voice over IPis another area where When the current market leaders keep sition the existing Internet to V6. The a service is becoming a product. Cisco building instruments of control -- like real need is for enabling the applications now manufactures an analog telephone NATs, like usage policies that bar certain that don't work well though the existing adapter (ATA) with a phone jack in the kinds of uses, and like attempts to charge protocols. Depending on how deep their front and an ethernet jack in the back. merchants a percentage on every transac- V4 assumptions are, we will find that ex- The box couldn't be simpler, and does tion on the net -- rather than what the isting applications can be re-implement- exactly what you'd expect a box with a users really want. ed atop V6 with modest effort. phone jack in the front and an ethernet jack in the back to do. The big advantage 43 COOK Network Consultants, 431 Greenway Ave. Ewing, NJ 08618 USA is that unlike the earlier generation of the first mile, which dominate. WiFi po- it seems only "nerds" really picked up on VoIP products -- "Now you can use your tentially does (and I emphasize potential- it - probably due to an article by Ander- computer as a phone!" -- the ATA lets ly, because we don't know yet how this son which was flagged on SlashDot in you use your phone as a phone, allowing will play out), since it may allow us to December. It should have hit the head- new competitors to offer voice service avoid the costs of wiring up every house- lines big time, since it allows a street to over any high-speed internet connection. hold. share all their broadband nodes, at a huge cost saving. It would allow a vicar in a Vonage.com, for example, is giving COOK Report: What Andrew is calling small village to hire a leased line, and away ATAs and offering phone service for may already be on the verge of hap- share the costs with all his parishioners - for40 a month. Unlike the complex           pening. On January 20, 2003 Guy              without any technical expertise." Editor
billing structures of the existing tele-      Kewney, of Newswireless.net published        – Readers should without fail read to the
phone companies, Vonage prices the            an article, "Become a wireless ISP: for      rest of the article at http://www.theregis-
phone like an ISP subscription.               £300," at the Register web site. The arti-   ter.co.uk/content/59/28972.html
cle certainly gives a partial answer to      .
p. 20 And hardware symbiosis will fur-        Andrew's question.                           p. 24 David Isenberg: You've clearly
ther magnify the threat of WiFi and VoIP.                                                  struck a chord with your ZapMail essay -
The hardest part of setting up VoIP is        Keweny wites: “While the learned are         - about a dozen SMART People have for-
simply getting a network hub in place.        laughing at Negroponte's fantastic "fu-      warded it to me. [snip] Who's going to
Once a hub is installed, adding an analog     turistic" vision of a mesh of intercon-      run the connectivity network after te-
telephone adapter is literally a three-plug   nected wireless LANs "like lilypads          lephony-classic and cable-TV-classic are
set-up: power, network, phone. Mean-          which you hop from one to another" a         dis-intermediated? This is the big
while, one of the side-effects of in-         UK company has produced Mesh wire-           unasked question, as Andrew Odlyzko
stalling WiFi is getting a hub with open      less technology which you can buy and        pointed out. (Fortunately for Fedex, the
ethernet ports. The synergy is obvious:       install, today, for under £300. Fancy set-   runways are not owned by American,
Installing WiFi? You've done most of the      ting up as a rival to BT Openworld?          United, or the US Post Office.)
work towards adding VoIP. Want VoIP?          Even in a remote village? Easy: buy a
Since you need to add a hub, why not get      Locustworld MeshBox; half the price of       Googin: To address David's and An-
a WiFi-enabled hub? (There are obvious        a home PC. You're in business."              drew's question as to who runs what
opportunities here for bundling, and later                                                 when the incumbents get disintermediat-
for integration -- a single box with WiFi,    "The software is the key to Locustworld.     ed. I have gotten that question since
Ethernet ports, and phone jacks for           Written by text-message pioneer Jon An-      2000. The answer is "who knows"?
VoIP.)                                        derson, it configures a group of wireless    While technological trends are some-
access points into a coherent "mesh" and     what deterministic, human behavior is
p. 20 WiFi hubs and VoIP adapters allow       connects them to any broadband Internet      not. The best we can do is set "boundary
the users to build out the edges of the       node available."                             conditions", as trying to pinpoint a "solu-
network without needing to ask the                                                         tion" tends to blind you to alternatives
phone companies for either help or per-       "Most experts regard the mesh approach       outside of our collective limited imagi-
mission. Thanks to the move from ana-         as hugely complex, because of the effort     nation. That is, in my opinion, the most
log to digital networks, the telephone        needed to set up the mesh. The system        productive line of reasoning is along the
companies' most significant competition       used to be known as a "parasitic net-        lines of: the next paradigm will most
is now their customers, because if the        work" - although the fashionable term        likely have these characteristics: (a)peer-
customer can buy a simple device that         these days is "symbiotic" - the idea is      ing, (b)fiber backbones (c)low budgets
makes wireless connectivity or IP phone       that you turn a group of wireless nodes      (d) a way around greedy city govern-
calls possible, then anything the phone       loose, and tell them to introduce them-      ments, bla bla bla. How they are imple-
companies offer by way of competition         selves to each other. Then you set up        mented beyond that will probably be de-
is nothing more than the latest version of    routes through the mesh. It can be           termined by some wild-eyed 20 year-old
ZapMail.                                      fiendishly complex, but Locustworld's        with tatoos. CHAOS. GO FOR IT.
mesh does this for you. You just buy the
Discussion of Zap Mail                        node from them: the current model is         Farber Faulhaber
£250 plus VAT."
"The last legal obstacle, according to       Article
p. 21 VoIP is indeed destroying the old
founder Richard Lander, was the deci-
business models, and in particular is                                                      The core of the Faulhaber/Farber argu-
sion by Oftel, allowing people to share
leading to a flat-rate pricing regime.                                                     ment is that all spectrum should be pri-
their broadband with up to 20 others. The
However, VoIP by itself does not deal                                                      vatized using something like the FCC's
excitement in the UK hasn't been quite as
with the basic problem of providing con-                                                   Kwerel & Williams "Big Bang" auction.
high as it was in the US, but even there,
nectivity. It does not lower the costs of

44
The COOK Report on Internet March 2003
[They] add that the private property         narrow to be commercially viable for          cause it is promising.
would have "easements" that require the      underlay uses. This becomes an empiri-
property owners to allow for spectrum        cal question. It's a narrower one than        Theories tell you where to look. You've
"underlays" using wideband spread            whether spectrum owners will try to poi-      still got to test them. And in this case,
spectrum or Ultrawideband and maybe          son competition in general.                   our theories incorporate theories about
also allow for cognitive radios to utilize                                                 economic behavior by future partici-
local unused, but allocated spectrum.        p. 28 Ikeda: As usual, Prof. Faulhaber's      pants. Those can only be validated by
theory is very unique in economics.           tests in the real commercial world. (Re-
They claim that this would deliver the       What does "marginal price" mean? He           member 802.11 was a big surprise in
best of both worlds (the economist's de-     must have invented it, because there are      terms of the economic behavior it has
sire for market forces to allocate nar-      no such terms in any textbooks in eco-        stimulated - so those who had a theory
rowband, high power spectrum and the         nomics.                                       that 802.11 might be big news ended up
engineers desire for a spectrum com-                                                       proving it by putting it on the market.
mons.                                        And why does ownership make sense?            Prior to that it was NOT obvious what
Maybe because he believes that com-           would happen with short-range unreli-
They even claim with a straight face that    mons will bring about the "tragedy of         able unlicensed wireless).
because of the strength of the commons,      the commons". It is a wrong application
that will force the price of spectrum        of Hardin's famous article. He empha-         For example, if we could only open up
property to near zero. If this is true, it   sized that the "tragedy" takes place          the software radios in today's 802.11a/b
seems to me to ask the question as to        when there is no "technical solution".        combo chips from Atheros, we could
why bother making it private property in                                                   conduct lots of experiments at low cost.
the first place.                             http://dieoff.com/page95.htm                  Not all important experiments, of
course, but lots.
My claim is that their approach is based     So we should investigate whether there
on a false assumption that such a market     is a technical solution before we discuss     p. 30 Berger I did get to talk with Gerry
would be open, unbiased and transpar-        economic solutions. If we can supply          and Dave for a while and even though
ent. If history is any guide, the people     more capacity than demand by opening          there are obvious differences, primarily
who will get the spectrum as property        spectrum, it makes no sense to "fill that     Gerry's strong support for a significant
are the top monopolists / oligopolists /     need".                                        position for private property, I got the
lobbyists who will use their market, cap-                                                  feeling that their approach is more of a
ital and political power to eliminate eco-   Timothy X Brown (University of Col-           way to make something happen that
nomic interference under the guise of        orado) Consider what I'd call "The Dis-       breaks the "GOSPLAN" style of spec-
eliminating technical interference.          ney Argument". Property values may            trum allocation. My interpretation is that
not go anywhere close to zero even with       they also strongly support the commons
p. 27 Kevin Werbach: For Farber and          competing alternatives and having large       and they believe that their proposal is a
Faulhaber, easements are a hedge. I          swaths set aside as a commons. Consid-        way to make it politically possible.
don't think they believe that prices will    er Disney who bought a big chunk of           There is probably more we have in com-
go to zero in a commons, or that the         swampland and created value from              mon with them than with people who
commons will work at all, but they can't     nothing. Disney is able to extract high       want a pure property rights.
rule it out. As good economists and en-      rents and nearby property values in Or-
gineers, they recognize that if the as-      lando are relatively high. This is despite    NSF & ICANN, pp. 31-33
sumptions open spectrum advocates            a plethora of competeing theme parks
make are true, a commons is indeed a         and despite a big chunk of Florida being
better ordering mechanism than a mar-        set aside as a free park.                     p. 31 Although during the past year
ket for most spectrum.                                                                     ICANN has become almost universally
By analogy, if someone with deep pock-        reviled, it is unfortunately not yet dead.
The question at issue here is whether        ets locks consumers into a wireless serv-     One reason is that the fundamental ques-
government can create them in a way          ice in a particular band, the value of that   tions of Internet governance posed by
that ensures they can be viable. The as-     band could be held quite far from zero.       ICANN are not yet solved. Six years
sumption is that, by defining the owner's                                                  ago the primary question was the insti-
property rights as stopping at the edge of   p. 29 Reed: I am greatly in favor of ex-      tutionalization of the IANA function.
the easement, a win-win solution will        ploring alternative systems that might        Today IANA works but the question of
emerge. But that depends on real-world       be scalable, and testing them in the real     who pays for it and who controls it is
conditions. At the last open spectrum        world. My favorite concept "societies of      still critical. It is important to under-
meeting at Harvard, Andy Lippman             cognitive radios" is one such thing that      stand how ICANN was created, if we
pointed out that the FCC's "interference     is promising and ought to be tested - not     are to avoid similar miscarriages of jus-
temperature" might leave easements too       because it is the only option, but be-        tice in the future. ICANN was to be all

45
COOK Network Consultants, 431 Greenway Ave. Ewing, NJ 08618 USA
about the glories of industry self–regula-   formation of ARIN in order to handle the    son of Jon Postel as the sole outside au-
tion. Instead it has given us an excellent   distribution of IP and ASN numbers had      thority over Network Solutions Opera-
demonstration of what happens when           to be completed, the memo informed the      tion of the DNS. Either the movement
you let the fox regulate life in the hen     Board that no public announcement of        on the part of those behind ICANN oc-
house. Or as another observer com-           the planned withdrawal would be made.       curred as a complete coincidence or it
mented: ICANN isn't really about "in-        As the draft put it “it is important that   occurred because word of the coming
dustry self-regulation." Rather ICANN        our plan to withdraw from this activity     policy change had leaked. Circumstan-
is an example of large-industry govern-      not be prematurely divulged to the gen-     tial evidence points to a leak. Judging
ment collusion to safeguard mutual self-     eral public” in order that the necessary    by what happened it seems likely that
interests and lock out newer and small-      coordination between Network Solu-          the IBM, MCI, ATT, ISOC group decid-
er stakeholders.                             tions and “the appropriate Internet gov-    ed that they had to put a brake on the
erning bodies” be achieved.                 process approved by the Science Board
p. 31 ICANN was the aborted “child” of                                                   in order to give themselves time to put
the privatization process. Some ob-          p. 32 -33 Had the NSB disagreed with        ICANN in place.
servers looking at the 1996 period have      what was presented in the memo, it
recently wondered why the “feds” just        would have objected and NSF efforts to      p. 33 Since NSF wanted out, an interim
didn’t walk away and avoid further in-       disengage would have ended. It did not      manager had to be created. By April
volvement. I point out below that it was     object. Consequently participants car-      1997 Brian Kahin and Burr were moved
government policy to do just this until      ried away from that meeting news that       to NTIA in the Department of Com-
the ICANN clique found out and execut-       would have substantial impact on the        merce and put in charge of such a “man-
ed a veritable coup d’etat.                  continued development of the commer-        ager” the Interagency Task Force on Do-
cial Internet. Planning at the NSF went     main Names. The announced goal was
p. 31 I have railed against ICANN from       forward for some 90 days when sudden-       to develop a coordinated DNS policy for
the very beginning. I can report now         ly on March 2, 1997 it was cut off at the   the US government. The reality was that
that I did so because Don Mitchell had       knees. Policy control was effectively re-   Burr and her associates at the Depart-
been a key inside source on matters of       moved very suddenly from the hands of       ment of Commerce were making the de-
Internet governance since January of         NSF and taken over by Kahin, Burr and       cisions and giving orders to NSF (which
1995. Don has given me his permission        Magaziner. What happened?                   was still the legal point of authority over
to be publicly identified. Don should be                                                 NSI) that the Cooperative agreement not
justly proud of his career at NSF where      I suspect I know. James Duderstadt, the     allowed to expire. The chance to extri-
from 1987 on he had a hand in virtually      President of the University of Michigan     cate government from the process was
every important project on which the         was also President of the Science Board.    lost. We maintain that it was lost be-
commercial Internet was founded. [See        Doug van Howelling was the VP of IT at      cause the levers of ICANN control were
text box on page 39 below.] To the extent    the University and a fervent supporter of   not yet ready.
I have been “right” about ICANN, I owe       IBM. The NSFnet backbone had been
my rightness to the education that he        built at Michigan by MERIT with joint       Had the cooperative agreement conclud-
provided.                                    study partners IBM and MCI. Duder-          ed in spring of 1997, as the NSF intend-
stadt himself has an association with       ed, the problem of institutionalizing the
p. 32 In a January 6, 2003 phone con-        IBM going back to the days of project       IANA function would have been forced
versation Don reiterated what he had         Andrew at Carnegie Mellon. It would         out on an open table (or, possibly made
told me in February 1997. Namely, that       have been very easy for Duderstadt to       moot) by the demand for (and creation
The Science Board Committee on Pro-          leak word about the highly sensitive        of) additional TLDs. It might also have
grams and Plans accepted that approach.      plans at NSF through van Howelling to       been forced into the courts. It certainly
Cerf and Patrick who could then very        would have become more clear to many
Don notes that this memorandum should        quickly pull the necessary strings to get   more people that one of the most critical
now be available from the NSF public         to Magaziner. Right now this is only a      underpinnings of the Internet, the IANA
affairs office.                              hypothesis. I have no proof. I suggest      function, had no basis in law. Neither
though that since the existence of the      domestic nor international. If the play
An early draft of the memo now in my         NSB memo is now known those doing           had been open, the high stakes mania
possession states: “NSF has determined       formal research could use it to find out.   that festered into the Internet bubble
that the best course is to disengage from                                                might well have not reached such a fever
our involvement in and oversight of the      One thing is certain. The would be con-     pitch. The industry might not have rid-
registration activity. This action will be   trollers of the Internet moved with great   den so high and fallen so hard.
taken March 31, 1997, the end on the 4th     speed after the NSF determined that the
year of the present award, as the date for   DNS should be in effect declared fully      The over ruling of NSF plans for termi-
our formal disengagement . . .” Howev-       mature and commercialized. Such a de-       nation by Burr and her ISOC clique and
er because a number of things like the       cision would have left IANA in the per-     the resulting extension of that agreement

46
The COOK Report on Internet March 2003
allowed a small number of high stakes        When the IFWP process was over, the         Now, I don't mean to idolize what the
players to keep the game closed. The         steering committee of the IFWP wanted       IFWPhad produced. And I don't mean to
game was still closed in June of 1999        to transform the set of principles around   pretend that the consensus it had gener-
when in the ICANN board emails we            which consensus had been formed into a      ated was perfect, or thick, or even con-
published Esther Dyson, IBM, Vint            document—a document that would form         sistent. There were problems with its
Cerf and Mike Roberts hatched a strate-      the basis of a new corporation, consis-     process, not the least of which the eco-
gy to get money for ICANN from the           tent with the principles of the White       nomic problem of assuring representa-
venture capitalists of Sand Hill Road by     Paper. And so the IFWP invited Har-         tiveness. Only those with money could
warning them that their investment were      vard's Berkman Center to host a final       afford to fly to the most expensive cities
in danger if ICANN did not succeed and       drafting session, where the work of the     in the world; and many with too much
by meeting with Tom Kalil in the White       international meeting could be trans-       money, and too little to say, seemed
House to seek support. Today the in-         formed into a final document. The aim       eager to fly.
vestments of the Sand Hill VCs have          of IFWP was that this final meeting
largely vanished, the IANA function is       draw together a representative group        But I do think there is something signif-
still not institutionalized. Indeed today    from the previous process, as well as       icant about this difference in process, es-
February 3, 2003 the IANAfunction was        representatives from IANA. IFWP's pre-      pecially as we become enamored of
just handed back to the same closed          supposition was that it would proceed       stakeholder government again. For
group of high stakes players who profess     only if these representatives would at-     while ideas in the IFWP process gained
to operate ICANN with openness with          tend.                                       currency through public debate and pub-
authority. In reality the game is still                                                  lic recognition, ideas in the IANA
closed.                                      But here, the process stalled. For as       process gained currency in part at least
Berkman scurried to see whether a meet-     through deals.
Lessig on Governance,                        ing was possible, and spent hours on the
phone with potential representatives,       This was not IETF. It was not the prod-
p. 34 - 39 Full Article                      IANA made it increasingly clear that        uct of a disparate collection of genius
they would work to subvert any such         geeks, thinking through the engineering
Editor's Note: Larry Lessig in the doc-
final meeting. At first they were willing   problem that Internet faced. It was a
ument that follows gives the best
to talk about the idea of a meeting, so     draft crafted by a lawyer — hired to rep-
overview that we have seen of the de-
long as the meeting was not announced;      resent his interests, and serving a com-
tails under girding ICANN’s construc-
and then, as time passed, and hence as      plex set of interests, he hacked out his
tion in the year 1998. In the talk that we
time grew short, they became increas-       deal. It was a deal, a deal done by a very
republish with his permission, he shows
ingly insistent that no meeting be held.    good lawyer, meeting in with many in-
how the GIP ISOC Clique found in Joe
Then just at the moment when Berkman        terests, and negotiating, to find an agree-
Sims an attorney who enabled them to
could wait no longer, IANA's represen-      ment. Doors closed. This was the
take advantage of libertarian distrust of
tative [Editor: Joe Sims] announced that    process that produced the ICANN
government to create an ICANN that
he had secured from the major interests     draft. It produced something else as
they could use for their own narrow ends
on the net — the corporate interests, the   well. For this lawyer who succeeded in
and brought on four years feuding and
technical community, and some segment       striking the deal that was the ICANN
distrust. ICANN from the very begin-
of the Steering Committee of IFWP —         draft—this Washington lawyer, skilled,
ning was broken. Such was the distrust
an agreement to resist any such final       one presumes, in making such deals—is
of government that no one would own
meeting. With his triumph, IANA's           also the author of something else of note
up to seeing the brokenness. Lessig saw
lawyer [Editor: Joe Sims] announced         to us. He is also the author of this phrase
it however and his analysis of what
that Berkman and IFWPcould hold their       that I have recurred to so often in my
could be expected from ICANN from
final drafting meeting if they wanted,      talk. It was this lawyer who said, "The
the position of hindsight more than four
but no one would come. No one would         single unifying force is that we don't
come because through private negoti-        want the government running things."
ations, the content of which no one re-     And in light of the process, and the free-
pp. 34- 35 "The single unifying force is
ally knows, IANA had cut a deal with        dom it allowed from the sorts of con-
that we don't want the government run-
enough people to stop the IFWP              straints that government might effect,
ning things."
process.                                    we can begin to see why.

47
Executive Summary :

The Action IS Taking                                                   stimulate the economy by encouraging         governments with be on the same fiber.
the largest phone companies and their ri-    This includes northern Quebec where
Place at the Edge, pp.                                                 vals to build more networks and spend        only the most remote villages will rely
Powell truly believes he is living in a      Robert Proulx President of XIT telecom
At the same time that AOL and the                                      dream world and ought to be removed          in Quebec told us in a February third
phone companies are trying to staunch                                  from his position by the Congress for in-    conversation that his small company has
the flow of blood (cash) from the center                               competence. The fact is the phone com-       all the business that it can handle includ-
innovation is taking place at the edges.                               panies cannot under any circumstances        ing major fiber community network
The US has built a bankrupt national                                   save those of government enforced mo-        builds in Hungary and in Jordan.
fiber system. Under Michael Powell                                     nopoly high prices use the networks they
who zigs and zags faster than a speeding                               already have. Given their debt they have     The Canadian CRTC is taking a very dif-
bullet between innovative spectrum pol-                                no money to buy new equipment for new        ferent tack from the American FCC.
icy and a retrograde insistence that if just                           networks. The equipment makes that           Telecom in Canada is understood as a
allowed the walking dead of last cen-                                  Powell rattles off as companies that         major national infrastructure resource in
turies telecom, the LECs and Cable Cos                                 would allegedly benefit further demon-       the same way the US understood the in-
will invest in building meaningful infra-                              strate his ignorance. Cisco yes, Intel       terstate highway system 50 years ago.
structure.                                                             perhaps. But Lucent has gone from 10         Now our dominant ideology permits in-
billion a year in revenue to two billion     vestment only in private corporate re-
Powell it seems isn't much interested in                               because it doesn't make equipment that       sources. Already 17th among the global
getting the details correct. Rather than                               sane management would buy were it too        users of telecom services according to a
taking the trouble to understand the dy-                               invest in a new network. With all the        recent OECD study the United States
namics of the technology on the market                                 fiber in the ground the only new market      economy will suffer in coming years be-
place as demonstrated by Clay Shirkey                                  in fiber for Corning is fiber to the home    cause of our current ideological short-
in his ZapMail essay republished in this                               and if Powell gets his way and gives the     sightedness.
issue Powell goes on to insist that it was                             telcos a monopoly on that no sane home-
unfair regulation imposed by his demo-                                 owner would buy it.                          The future is in asset-based and cus-
cratic predecessors that has bankrupted                                                                             tomer-owned networks. We have in-
the industry.                                                          But Powell, it seems, is interested much     stalled cable modem service in order to
more in ideology than in accurately fig-     move our long distance calls to Vonage.
According to an article in the February                                uring out where the technology is going.     Suddenly unlimited long distance in the
second New York Times http://www.ny-                                   Dave Hughes caught Powell giving Sen-        US is flat rate. All of Canada is 5 cents
t i m e s . c o m / 2 0 0 3 / 0 2 / 0 2 / b u s i n e s s / y o u r-   ator Brownback of Kansas in correct in-      a minute and most of Europe and much
money/02FCCC.html?ex=10451732                                          formation about Wi-Fi in his testimony       of Asia is not much more.
42&ei=1&en=c807a35b91f72fd1, Pow-                                      on January 20 as Powell stated that the
ell has asserted that deregulation "should                             radios used would transmit at best 300       Innovation at the edge is possible and as
not be like a dessert that you serve after                             feet on an 802.11b network.." Hughes         prices continue to fall the huge compa-
people have fed on their vegetables and                                skewered Powell in public as well he         nies that Powell want to serve will be-
is a reward for the creation of competi-                               should have. Within 24 hours Hughes          come more and more unwieldy. Fiber to
tion." Rather, he said, deregulation is "a                             heard back from a Powell assistant.          the home is worth having only if the
critical ingredient to facilitating compe-                             "Thank you for your comments. In the         homeowner can control it.
tition." Powell is talking the same naive                              passage you reference below the Chair-
faith in industry self-regulation that put                             man simply made a mistake." One won-         Meanwhile the edges continue to canni-
the ICANN fox in charge of the hen                                     ders how many "mistakes" Michael             balize the center - like a million termites
house. The phone companies would be                                    Powell is making these days?                 chewing on the soggy log of the PSTN.
acting against the interests of their stock-                                                                        BellSouth was the first ILEC to ac-
holders, if given a chance they did not                                The Canadians are not making mistakes.       knowledge the inevitable and at the end
charge the most extortionate rent for the                              They are building a working national         of January announced that it would
use of their monopoly possible.                                        fiber system. They are investing 200 mil-    begin to resell Vonage to its DSL cus-
lion dollars in linking all public schools   tomers. See http://news.com.com/2100-
The times writes: Mr. Powell and his                                   throughout Quebec with fiber and are         1033-982606.html And from a trusted
supporters say a change in the rules will                              doing it such away that all municipal        source we are told that in Japan NTT has

48
The COOK Report on Internet March 2003
effectively ceased development in its cir-     Japan. Farooq concludes that we will          Francois Menard commented: I'm trying
cuited switched landline network.              likely have the worst of all worlds with      very hard to get [Canadian] municipali-
most of the internet running v4 and few       ties to implement IPv6 open access
Backbone v6 going no                           isolated instances of v6.                     across municipal FTTH networks so that
MPLS doesn't squeeze-in and end-users
where pp. 5 - 11                               When we asked Farooq about the Janu-          become required to run PEs. I'm seeing
Highlights, Full Article                       ary 22 announcement by Telehouse of an        ISP's provide value added services by of-
Farooq Hussain explains how reasonable         IPv6 peering exchange in New York, he         fering commercial access to tunneling
uses for IPv6 in Internet backbones have       responded PAIX has had the ability to         servers on their premises which bridge to
evaporated. DHCP and Nats acting as            support v6 for at least two years, The        the good old legacy Internet. For as long
firewalls have gotten the need for v6 as a     more interesting announcement was the         as two service providers across two dif-
sources of extra address space well under      one for the exclusively v6 exchange set       ferent municipal FTTH system would
control. The idea of universal address-        up in France as part of the EC initiative.    want to interconnect with IPv6, there
ability across the internet for all devices    Also has hardly any takers except those       would then be a parallel Internet.
has receded in importance.                     who are compelled by politics to go there
by virtue of being participants in the EC     A current problem is the absence of a
He cites very interestingly that in the        initiative. It's fine for Telehouse to make   good v6 tool set for end users. Right now
United States almost all the support for       this kind of announcement but the capa-       it is not clear where one will come from.
v6 comes from the defense department.          bility is of little commercial interest ei-   Standards development would prove
In Europe it comes from the European           ther to enterprise or service provider net-   useful. But by whom? The IETF is very
Commission and in Japan from a man-            works. There's simply not enough traffic      unlikely. The IEEE perhaps. The Con-
date by the Japanese government.               volume with v6 and there are so few na-       sumer Electronics Association claims to
tive v6 networks that none of them need       be doing work in the area. Unfortunately,
Mandate or not there no longer seems to        to go to Telehouse or any other IX to ex-     we have not had enough contact to eval-
be any market pull. End to end applica-        change traffic. The IX's used for v6 have     uate them.
tions like voice over IP that once were        been established primarily to foster R&E
thought to be dependent on v6 are being        projects and are sustained primarily on       ZapMail pp.18 -20
re-engineered to work with NATs in the         non-commercial rationales.
v4 world.
Edge based v6, pp. 11 -
3GPP, which is the third generation mo-                                                      When does a service become just anoth-
bile project, adopted IPv6 as their proto-
er product that the phone company’s cus-
col of choice in 1999. In doing so it gave                                                   tomers can deliver best and at lowest cost
v6 in the backbone. Farooq agrees. It
v6 the strongest endorsement it has ever                                                     for themselves?
seems that v6 at the edge can be used by
had. Yet because at the height of the
end users to establish their own applica-
bubble the 3GPP people decided to build                                                      Clay Shirkey has written a powerful
tions and perhaps even routing by using
their own internet parallel to the global                                                    essay that likens the Local Exchange
v4 internet, their plans now seem rather                                                     Carriers’ world view to that of Fed-Ex
silly. With the slowdown in wireless                                                         when it though it needed to build a fax
Frankston has written: IPv4 (or just "IP")
growth has come a slowdown in wireless                                                       network to gain an advantage that other
represented the birth of the Internet by
demand for IP numbers. If we assume                                                          overnight carriers didn't have to offer
shifting the power to define the network
that with the arrival of software defined                                                    their customers only to find that the cus-
to the users at the edges.
radios over the next few years radios will                                                   tomer could deliver information by fax
communicate with each other on the                                                           much more cost effectively themselves.
The Internet has thrived because supply
basis of IP rather than geography this is
is driven by demand. New application
likely to delay indefinitely the need for                                                    The business Fred Smith imagined being
services are supported by simply provid-
v6 in wireless devices.                                                                      in -- build a network that's cheap to run
ing more transport (or IP) capacity.
Rather than wait for new capabilities to      but charge customers as it if were expen-
SONY has announced that all of its de-                                                       sive -- is the business the telephone com-
be defined, users will create their own
vices will speak v6. However in the ab-                                                      panies are in today. They are selling us a
solutions. (When I say "users" I don't
sence of widespread v6 deployment                                                            kind of ZapPhone service, where they've
mean all users create applications. It only
SONYs products will also have to com-                                                        digitized their entire network up to the
takes one motivated, creative individual
municate in a v4 world, The problem is                                                       last mile, but are still charging the high
with some time on their hands to create
that if the communicate well in v4 there                                                     and confusing rates established when the
an application that will be adopted by
is likely to be no use for their v6 capabil-                                                 network was analog.
millions of others. We just don't know
ities. It seems that even Jun Murai is no
which user that will be.)
longer promoting v6 wholeheartedly in

49
COOK Network Consultants, 431 Greenway Ave. Ewing, NJ 08618 USA
and their maintenance of property rights    vestments of the Sand Hill VCs have
Discussion of ZapMail,                       in spectrum while saying oh by the way      largely vanished, the IANA function is
you open spectrum folk may be given so      still not institutionalized. Indeed today
radio technologies will not get in the      just handed back to the same closed
Adrew Odlyzko agrees that VoIP will
way of our much more rational corporate     group of high stakes players who profess
lead to the flat rate commoditization of
approach.                                   to operate ICANN with openness with
long distance phone service. but he
authority. In reality the game is still
them wonders about wireless being able
How NSF Was                                 closed.
to create a substantial enough infrastruc-
ture for voice communication.                Prevented from                              Lessig on Governance
A product known as Locustworld may
Removing the govern-                        pp. 35-40 Highlights, Full
have the answer. "A UK company has           ment from domain                            Article
produced Mesh wireless technology            names? pp. 31 -34
which you can buy and install, today, for                                                Larry Lessig in the document that fol-
under £300. Fancy setting up as a rival to                                               lows gives the best overview that we
BT Openworld? Even in a remote vil-          Don Mitchell explains that a major poli-    have seen of the details under girding
lage? Easy: buy a Locustworld Mesh-          cy change by the National Science foun-     ICANN’s construction in the year 1998.
Box; half the price of a home PC. You're     dation was aborted. The change would        In the talk that we republish with his per-
in business."                                have ended government involvement in        mission, he shows how the GIP ISOC
the DNS. It could have nipped ICANN         Clique found in Joe Sims an attorney
"The software is the key to Locustworld.     in the bud. But this was not to be..        who enabled them to take advantage of
Written by text-message pioneer Jon An-                                                  libertarian distrust of government to cre-
derson, it configures a group of wireless    Had the cooperative agreement conclud-      ate an ICANN that they could use for
access points into a coherent "mesh" and     ed in spring of 1997, as the NSF intend-    their own narrow ends and brought on
connects them to any broadband Internet      ed, the problem of institutionalizing the   four years feuding and distrust. ICANN
node available."                             IANA function would have been forced        from the very beginning was broken.
out on an open table (or, possibly made     Such was the distrust of government that
"Most experts regard the mesh approach       moot) by the demand for (and creation       no one would own up to seeing the bro-
as hugely complex, because of the effort     of) additional TLDs. It might also have     kenness. Lessig saw it however and his
needed to set up the mesh. The system        been forced into the courts. It certainly   analysis of what could be expected from
used to be known as a "parasitic net-        would have become more clear to many        ICANN from the position of hindsight
work" - although the fashionable term        more people that one of the most critical   more than four years later reads like
these days is "symbiotic" - the idea is      underpinnings of the Internet, the IANA     prophecy.
that you turn a group of wireless nodes      function, had no basis in law. Neither
loose, and tell them to introduce them-      domestic nor international. If the play             Dave Hughes to
News Item
selves to each other. Then you set up        had been open, the high stakes mania
routes through the mesh. It can be
Chairman Powell -
that festered into the Internet bubble
fiendishly complex, but Locustworld's        might well have not reached such a fever    Jan 22, 2003
mesh does this for you. You just buy the     pitch. The industry might not have rid-
node from them: the current model is         den so high and fallen so hard.             FCC Chairman Powell in testimony to the
£250 plus VAT."                                                                          US Senate. "That's the way that current
technology is configured and deployed.
The over ruling of NSF plans for termi-
"The last legal obstacle, according to                                                   Right now the leading standard of 802.11 a
nation by Burr and her ISOC clique and
b and g in their very first have a limit in
founder Richard Lander, was the deci-        the resulting extension of that agreement
their range. At best 300 feet on an 802.11b
sion by Oftel, allowing people to share      allowed a small number of high stakes       network.. " THAT IS AN ABSOLUTELY
their broadband with up to 20 others.        players to keep the game closed. The        FALSE STATEMENT!!!!! 'at best 300 feet'
game was still closed in June of 1999       WHY DID HE MAKE IT?
Farber Faulhaber                             when in the ICANN board emails we
published Esther Dyson, IBM, Vint Cerf      1. There are over 10,000,000 Wi-Fi sys-
versus Open Spectrum                         and Mike Roberts hatched a strategy to      tems out there. 1.5 million more each
pp. 25 - 30 Highlights, Full                 get money for ICANN from the venture        month. 2. There are over 2,500 and proba-
Article                                      capitalists of Sand Hill Road by warning    bly over 4,000 Wireless ISPs using Wi-Fi
them that their investment were in dan-     radios doing business across the United
A discussion of the problems created by      ger if ICANN did not succeed and by         States as I speak. Largely RURAL. I will
the presentation of a paper that looks to    meeting with Tom Kalil in the White         wager not ONE of them is serving cus-
the past encourages spectrum auctions        House to seek support. Today the in-        tomers 300 feet or less. Most are from 1

50
The COOK Report on Internet March 2003

mile to 10 MILES using off the shelf                                                        The Customer Becomes the
equipment certified by the FCC and with-      I KNOW what I am talking about. Why
doesn't the Chairman of the FCC? Or his       Network - $395 Now avail- in the power limits - 36dBm EIRP - pre- scribed as the maximum for 802.11b ra- Staff, who prepared him for this Hearing? able.http://cookreport.com/a dios. 3. Cisco sells tens of thousands of Or is there a hidden agenda there? That ssetbased.shtml 802.11b 'Aironet' radios which are AD- kind of completely false and misleading VERTISED as reaching 18 miles at statement before Congress angers me! For Subscription Rates 11mbps or 25 miles at 2mbps! in effect he was telling Senator Brown- Choice of either ascii or Adobe Acrobat http://cisco.com/univercd/cc/td/doc/pcat/3 back, whose 'colleagues' and constituents (PDF) format 1. Individual; College or 50wlbr.htm#fea 4. Young Designs Inc CORRECTLY identify Wi-Fi as ONE University Department; or Library; or Small sells COMPLETE 'Wi-Pop In A Box' sys- technology which can bridge the 'last Corporation -$300 2. Corporate - (revenues
tem for communities advertised at 12          broadband mile' until whole new genera-       $10 to 200 million a year) -$395 3. Large
miles! Standard, certified, systems. 5.       tions of radios are invented, that Wi-Fi is   Corporate- Revenues of $200 million to$2
of NO REAL VALUE for Broadband. I             billion per year - $495 4. Very Large Well funded companies are ramping up to KNOW what I am talking about. Why Corporate- Revenues of more than$2 billion
deploy Wi-Fi across cities all over the US.                                                 per year - $595 One announced today it was targeting doesn't the Chairman of the FCC? Or his 80% of the Front Range Colorado popula- Staff, who prepared him for this Hearing? Site License: The right to distribute ascii tion. And yes, they will backhaul over Or is there a hidden agenda there? and PDF via email to all employees of cor- broadband wired networks, to answer your poration. 5. Small corporate:$495 6.
That kind of completely false and mis-
question accurately, Senator Brownback.                                                     Corporate: $775 7. Large Corporate:$1000
6. I have spent the last 3 YEARS buying,                                                    8. Very Large Corporate: $1250 . Site me! For in effect he was telling Senator deploying, testing Wi-Fi 2.4ghz as well as License Distribution via intranet web site Brownback, whose 'colleagues' and con- other Wi-Fi Bands (915mhz, 5.7Ghz) ra-$400 a year additional. See www.cookre-
stituents CORRECTLY identify Wi-Fi as
dios for 4 more years from half a mile to                                                   port.com for more detail
ONE technology which can bridge the
15 and more miles. And I AM a Wireless
'last broadband mile' until whole new gen-    Gordon Cook, President
ISP ALL of whose customers are Wi-Fi
erations of radios are invented, that Wi-Fi   COOK Network Consultants
2.4ghz at ranges from a third of a mile to
is of NO REAL VALUE for Broadband.            431 Greenway Ave
2 miles!
Ewing, NJ 08618, USA
Telephone & fax (609) 882-2572
Internet: cook@cookreport.com
The COOK Report on Internet                                                                 How to use interactive features of this
`