Lf by langkunxg

VIEWS: 0 PAGES: 32

									      Paying for digital libraries
Why is this hard?
• New services, but no new money
• Cost of transition
• Users not in the community
A problem of quantization:
   Sell by item or sell in bulk?
   Sell to one reader or to many?
Should research be available free?
           Wide range of opinions
Steve Harnad: making the refereed journal literature in all
disciplines on-line and free for all, with no financial firewalls, is the
optimal and inevitable solution for science and scholarship.
Floyd Bloom (editor, Science): Neither the public nor the scientific
community benefits from the potentially no-holds-barred electronic
dissemination capability provided by today's Internet tools. Much
information on the Internet may be free, but quality information
worthy of appreciation requires more effort than most scientists
could muster, even if able.
Jack Valenti: .. famously proclaimed (1982) that the videocassette
recorder was as threatening to the movies as the Boston Strangler
was to a woman walking alone. [In 2002 video rentals were 46% of
studio revenues, box office receipts were 24%].
      Digital reading in libraries

 Library of Congress, calendar year 2000 statistics:

                   Items sent out      Approx. bytes

Website            900M                9 TB
Reading rooms      1.6M                1 TB

   Our technology has outrun our economics.
   As always.
              More and cheaper disk
 Disk TB Shipped per Year                                                      $/MB
1E+7
                  1998 Disk Trend (Jim Porter)
          http://www.disktrend.com/pdf/portrpkg.pdf.
                                                               1950 1960 1970 1980 1990 2000 2010
                                 ExaByte
1E+6
                                                         10000
                  disk TB
                                                           1000
                  growth:
                                                             100
                  112%/y
1E+5
                                                              10
                                                               1
                               Moore's Law:
1E+4                             58.7%/y
                                                             0.1
                                                            0.01
                                                          0.001
1E+3
   1988       1991        1994        1997        2000
                                                         0.0001


(From Jim Gray)
                      Size of Web

            1000

             100
Terabytes




              10

               1

             0.1

            0.01
               1994     1997   2000   2003
And yet more information…
       Journal price crisis
Year       CPI        Periodicals
1988       4.1        9.1
1989       4.8        9.5
1990       5.4        9.5
1991       4.2        11.7
1992       3.0        12.2
1993       3.0        5.5
1994       2.6        9.6
1995       2.8        10.4
1996       3.0        10.8
1997       2.3        9.9
1998       1.6        10.3
1999       2.2        10.4
2000       3.4        9.0
2001       2.8        8.3
2002       1.6        7.9
Avg.       3.1        9.5
             Journal prices
14                        US CPI
                          Periodicals
12
10
 8
 6
 4
 2
 0
     1988 1990 1992 1994 1996 1998 2000 2002
ARL statistics
For-profit vs. society journals
                           ACM transactions
                   18
                   16
Number published




                   14
                   12
                   10
                    8
                    6
                    4
                    2
                    0
                    1970     1980   1990   2000   2010
ACM transactions, members
18
16   Members/10K
14   Transactions
12
10
 8
 6
 4
 2
 0
   75

   78

   81

   84

   87

   90

   93

   96

   99

   02
19

19

19

19

19

19

19

19

19

20
     How do we pay for it all?
• Communities pay (what happens now)
• Readers pay by month or year
• Readers pay by item or minute
• Authors pay
• Advertising pays
• Other (PR, bounties, cost avoidance, …)
None of these seem to quite work…
Electronic purchases at SUNY
“freely available to everyone, twenty-four
hours a day, everywhere, forever.” (Harnad)

 Remember when electricity was going to be “too
    cheap to meter”?*
 80% of a library cost is not shelf space.
 If we all we cared about was minimizing storage
    cost, we’d have been reading microfilm for
    decades.
 Books on disk are cents each instead of dollars:
    but the staff still gets paid (perhaps more).
 * Lewis Strauss, AEC Chm, Sept. 16, 1954
Costs: libraries and disk farms


Per book-year: $3-8
Per staff position: 7,000-20,000 books




 Per megabyte-year: under a cent
 Per staff position: terabytes (millions of books)
      Why is digital a problem?
Most users are not in the community paying. For example,
  the BUBL site had only 15% of its users from within the
  UK, which pays for it.

Why should a university provide a good digital library
 service, if most of the users are not on premises?
   Can you compete with free?
Does Gresham’s Law apply to the Internet?
 (Ian Irvine: the Internet is what got rejected
 by his journals).
Conventional answers: bottled water, cable
 television.
              Versioning: theory
Hal Varian is known for supporting the idea of “versioning”
  to enable sales to different people at different prices.

Suppose some report costs $7 to write; copies are free.
If A will pay $3 and B will pay $5,
• Neither can buy the report just alone
• If the price is set at $3, the author can’t afford to write it.
• If the price is set at $5, only B would buy it, again fails.
• But if you can charge A $3 and B $5, it works.
            Versioning: practice
This is actually extremely common

Airline tickets are the best known examples.
Paperback vs. hardback books are another.
Books in India vs. books in the US.
Movie pricing: theatre vs. rental vs. purchase.

If you think about it, even staples like food and gasoline
    vary more by neighborhood than they might.
 Called “zone pricing” for gasoline.

Everyone may be better off, but many of of them are angry.
Subscriptions
    Library subscriptions work
Elsevier reports regular growth in
  ScienceDirect, with $1.5B in online
  revenue.
Consumers spent relatively little for online
  content, however.
Enough is available free, and people resent
  paying for better: Gresham’s law
     Journal prices per page (1999)
Journal                        Publisher            Price    Price/page

J. of Economic Studies         MCB UK               $6999*   $13.40

Managerial & Decision          Wiley                $905     $3.10
Economics
J. of Forecasting              Wiley                $760     $1.47
Econometric Reviews            Dekker               $650     $1.45
…
J. of Economic Issues          Assn Evolutionary    $45      $0.04
                               Economics
Population & Development       Population Council   $36      $0.04
Review
Amer. Economic Review          Am. Econ Assn        $142     $0.03

    * Not a misprint. The current price is $9389.
            Journal prices per cite
Journal                        Publisher   Price   Price/cite

Public Finance Review          Sage        $394    $56.29
Economic Modeling              Elsevier    $535    $53.50
J. of Mathematical Economics   Elsevier    $1147   $47.96
Mathematical Social Sciences   Elsevier    $824    $29.43
…
J. of Political Economy        U Chicago   $159    $0.63
Quarterly J. of Economics      MIT Press   $148    $0.52
Amer. Economic Review          Am. Econ    $142    $0.14
                               Assn
              Page charges
Brain Research costs 19,000 euros/yr
PLOS will recover costs from authors at $1500 per
  article; then post free.

US legislation to prohibit copyright on government
  funded research
Warning: proportionally more authors are in
  universities and readers in industry
The answer has to be in profit and setup costs,
  since authors are not paid and paper is cheap.
            NIH-funded papers
NIH director Elias Zerhouni says researchers receiving NIH
grants should send their manuscripts to a free, Web-based
archive managed by the National Library of Medicine as
soon as they can, after first submitting them to medical or
scientific journals … Scientists can ask for a delay of up to
one year to protect the profitability of journals, Dr Zerhouni
said. (Reuters, Feb. 4, 2005).
Pat Schroeder, president and executive director of the
Association of American Publishers, … likened the report's
recommendations to the "federal government putting down
its thumb." (2004 discussion of same idea)
(NIH runs a repository called PubMed Central).
    Advertising unlikely to help
Most journals in past did not get that much
 advertising; most web advertising goes to
 a few sites and does not seem likely to be
 targeted at readers of journals.

Mostly, libraries advertise for donors; and
 computer equipment becomes obsolete
 too quickly to attract bequests.
Advertising
            Micropayments
Lots of resistance: people dislike
   Recordkeeping, metering
Hard to collect small amounts of money
Lots of administrative overhead
Bad social incentives


Users, librarians, publishers all risk-averse
So we’ll sell in bulk to groups
 Micropayments not so micro
Odlyzko: cost per article read perhaps $200; cost
per article published some $4000.
Modern publishing not adapted to small press
runs and small readership.


Odlyzko again: US mathematics publishing is as
much as: the entire NSF research budget in math;
or a new faculty member in each department.
                Other ideas
Foundations
Cost avoidance
Advertising print
Reputation
Bounties
Dedicated taxes
Government support
Pledge breaks
              Conclusions
It’s probably library subscriptions vs author
   page-charges (paid by institutions).

Both represent bulk sales to groups.

Inertia is winning: library subscriptions

								
To top