Get documents about "The Atomizing Hand
Document Sample
The Atomizing Hand
The strategy and economics of
peer production
Umair Haque
http://www.bubblegeneration.com
Spring 2005
Intro
What is Peer Production?
• Communities of prosumers or amateurs who
collaborate to make things
– Wikipedia
– Metafilter
– Flickr
– OhMyNews
• By…
– Sharing information about inputs and outputs…
– …which creates pooled knowledge
– ...which increases the efficiency of future production
– And having open access to the means of production
Why Share?
• Altruism
– Because we care
• Reputation
– Because there are spillover signalling effects
• Reciprocity
– Because of implicit or explicit social control
• Money
– Because there are explicit incentives
• Fun
– Because we derive social utility from doing so
• Etc…
– ….No definitive answer (yet).
• Empirical effects are unambiguous
– File-sharing, Wikipedia, threadless, etc demonstrate that peers do
share
Why Join a PP Community?
• Simple answer:
– Because you can do stuff more efficiently than by yourself
• By taking advantage of others‟ sharing?
– No!
– PP is not free riding – in fact, the opposite happens: increasing
returns
• When you do & make stuff, new knowledge is a side-effect
• Knowledge in a peer production community is a public good
• New users are attracted in proportion to the size of the
knowledge pool
– Because it‟s the strongest signal of potential marginal productivity
gains from joining that community
• And new users increase everyone else‟s productivity, growing
the knowledge pool
Why is Peer Production
Important?
• Peer production enjoys very special economics
– A new kind of scale economy which enables massively distributed
and ultraspecialized microproduction…
• Distributed economies of scale
– Which lets knowledge about production pool within the
community…
– Which creates sustainable supply-side increasing returns
• Under certain conditions, peer production communities will
be more efficient than firms or markets
– To understand why, I‟ll try and explain why peer production is really
about cheap coordination atomizing production…
– …which makes it a fundamentally new mode of production
• Or at least one analogous to modes of production that disappeared a
very long time ago, like the putting-out system
– …which makes it a great opportunity (or a great threat)
Network Economics
1.0 and 2.0 Primer
Network Economics 1.0
• Demand-side viral effects
– Massively decreases user acquisition cost
– Low-cost imitation barrier to protect early-mover advantage
• Demand-side network effects
– One key resource: consumption network
– “Metcalfe‟s law” – marginal utility increases in network size
• No upper bound factored in – misreading of econ lit.
– Switching costs rise nonlinearly with network size
– Increasing returns to adoption
• Strategic intent: scale – grow adoption
– Rapid growth of network size
– “Demand-side economies of scale”
• Search economies dominate strategy
– How costly is it for you to find stuff on the network?
– Cheapest solution wins: utility of network bounded by search
mechanism
Network Economics 2.0
• Supply-side viral effects
– Massively reduces prosumer acquisition costs
– Low cost imitation barrier to protect early-mover advantage
• Supply-side network effects
– Two resources: producer network and knowledge pool
– “Haque‟s Law” – marginal productivity increases in network size
• Upper bound: diseconomies of scale – large user network size creates
knowledge network coordination costs
– Switching costs rise nonlinearly with both
– Increasing returns to knowledge
• Strategic intent: speed – accumulate knowledge
– Rapid growth of knowledge pool
– Doing & making things faster than competitors
• Coordination economies dominate strategy
– How costly is it for you to make & do stuff on the network?
– Cheapest solution wins: network utility bounded by coordination
Markets and Firms:
Production 1.0
Apologies
• Five somewhat painful slides to follow
– Bear with me, because they‟ll help us explain why and when
firms are less efficient than peer communities
– This is the most complicated part of the presentation
• OK, I‟m lying…
– …skip to the Summary slide if (when) you can‟t take it
anymore
Property Rights
Organize Production
1. Right to use economic resources
2. Right to modify form and substance of resources
3. Right to benefit from use of resources
4. Right to transfer resources
• We will call 1 and 2 rights of coordination because they
organize production
– Liquid markets for 3 and 4 are prevalent
• Equities, fixed income, options, futures
– Markets to assign coordination rights often fail because of relatively
high transaction costs
• Copyright is the canonical example
• Firms coalesce to economize on transaction costs of
coordination rights assignment when markets fail
Understanding Firms
• What do firms do?
– Coordinate:
• The most efficient way to turn inputs into outputs…
• …By internalizing the uncertainty inherent in risky transactions...
• But economizing on transactions isn’t costless…
– Firms replace transaction costs with coordination costs
• How are coordination rights distributed within the firm?
• Within the firm, who can access the means of production and key
inputs?
– Costs of administrative overhead
– Internal transaction costs: negotiation, bargaining, and enforcement of
coordination rights
• These are essentially costs of information sharing and processing –
firms internalize information costs, where they can be „managed‟
– As long as coordination costs stay less costly than transaction
costs of inputs, firm is an efficient way to produce
– But they don‟t, because….
Coordination & Hierarchy
• Negotiating for coordination rights becomes
combinatorially costly
• I want a top engineer on my team
• I negotiate with 4 VPs
• Each VP negotiates with each other
– Coordination cost explosion in firm size – negotiation doesn‟t scale
– How do we deal with this? Hierarchy – centralized decision-making
• Hierarchy is designed to minimize coordination costs
– In firms, coordination rights are governed by authority embedded in
hierarchical implicit or explicit contracts
• You‟re my boss, you choose the „best‟ products and features
• I‟m the product manager, so I get the top engineers
– Hierarchy minimizes coordination costs by economizing on
information
• Managers and administrators collect and process information from
subordinates
Coordination Diseconomies
• Minimizing coordination costs by economizing on
information creates a big problem…
– Private info is never made completely public
• Bosses, middle managers, bureaucracy, politics, bullsh*t, inertia all
make revealing private info costly
• Memos, reports, and meetings lose tacit info
• Disutility: authority sucks
– In firms, we can‟t know if efficient solutions to assigning
coordination rights are reached
• Because information, expectations, and preferences about inputs and
outputs are never fully revealed
• Coordination diseconomies (1):
– The efficiency loss of hierarchy in assigning the best coordination
rights, due to information loss
– This is an opportunity cost of hierarchy – we may forego a better
coordination rights assignment
Coordination & Specialization
• And…hierarchy does not eliminate coordination costs
– It minimizes them
– Coordination costs still exist
• Direct costs
– Admininstration, middle managers, bureaucracy
• Indirect costs
– Agency and influence costs – politics! Incentives dulled
• Coordination bounds specialization
– It‟s economical to coordinate activities until marginal cost of
coordination exceeds marginal gain from specialization
• Intuition: even in firms, coordinating 10,000 microactivities isn‟t
economical
• Or: 10 managers can coordinate 1000 activities, 100 managers can‟t
coordinate 10,000 activities
• Marginal cost of coordination increases exponentially in firm size
• Coordination diseconomies (2):
– Specialization gains hits a coordination cost wall
– Implication: eliminating coordination costs can realize specialization
gains
Summary: Modes of Production
• Markets fail
– Because of high transaction costs
• Firms coalesce but value creation is bounded by
– High coordination costs, which limit any specialization gains
• Hierarchies coalesce but value creation is bounded by
– Coordination diseconomies
• Total cost of hierarchy is high
– Coordination costs bound specialization gains…
– …and opportunity cost is optimal coordination righst assignment
• Or: a trade-off between specialization gains and information loss
• The more you minimize coordination costs, the more information you lose…
• …but the more you minimize coordination costs, the greater specialization gains
you realize
• A more efficient form of organization…
– …escapes the information loss vs specialization gains tradeoff, and realizes
advantages in both
• Zero coordination costs wenable huge specialization gains
• Not having to economize on information means optimal assignment of
coordination rights
Coordination & Specialization
Coordination costs
Value
Specialization gains
Coordination costs exceed
specialization gains
Size
Coordination costs bound specialization gains
Coordination Diseconomies
Hierarchy costs
Value
Specialization gains
Total costs exceed
specialization gains
Opportunity cost – information loss
– increases costs of hierarchy
Size
Coordination diseconomy: total costs of hierarchy are coordination costs and opportunity
cost: not choosing the best coordination rights assignment. Specialization is effectively
bounded at a small maximum efficient scale. Can trade off by either coordinating more, and
losing less information…or coordinating less, and realizing greater specialization gains.
Peer Communities:
Production 2.0
Peer Production
• PP vaporizes the real-world costs of assigning coordination
rights
– Information sharing mechanisms aggregate and reveal
private information and preferences frictionlessly
• About inputs
– What are the most efficient ways to produce our good?
• About outputs
– What are our most successful goods likely to be?
• Using explicit mechanisms
– Voting, ranking, pricing
• Or Implicit mechanisms
– Tags, playlists, reputation
• Without coordination costs…
• …so there‟s no efficiency loss of centralized decision-making
like in firms
– How…?
Coordination Economies
– Because information and preferences are public, most
efficient coordination rights assignment is transparent
• Public information regulates which inputs…
– code, people, designs
• …are granted access to means of production…
– platform, community, architecture
• …to create the outputs predicted to be most successful.
– T-shirts, games, OS kernels, music
– Frictionless, hyperefficient assignment of coordination rights:
coordination economies
• PP can assign more efficient coordination rights more cheaply
than firms or markets
• And so can realize gains from ultraspecialized production
– Coordination economies are a significant source of strategic
cost advantage
Coordination Economies
• How does PP achieve hyperefficient assignment of
coordination rights?
– PP model is an implicit or explicit auction for access to means of
production
• Implicit or explicit price mechanisms combined with price or quantity
auctions
– Generally:
• Peers self-select into activities – ex-post coordination
• Distributed local information is aggregated
– Implicit price of access to means of production is established
• Peers are granted access to means of production (or not)
– If peer activity value as measured by aggregate utility is greater
than implicit price as measured by aggregate bid
– Production without coordination costs
• No need to economize on information – massively distributed info
processing, just like in real-world markets
– Costs incurred are market costs
Summary:
Firms, Markets, and Communities
• PP Models combine the microeconomic DNA of markets
and firms:
– Places where people can exchange decentralized
information about preferences and utility (markets)
– In order to coordinate centralized actions of consumption
and production (firms)
– Without the coordination costs of real-world firms
– …or transaction costs of real-world markets, which create firms
• Which scale combinatorially with firm size – cost explosions
• Which makes complete information revelation uneconomical in firms
• Which makes it impossible for firms to assign the most efficient
coordination rights
– Peer production is theoretically hyperefficient
• It can assign the most efficient coordination rights more cheaply
than real-world firms or markets
• Or: PP lets the most talented people make the coolest things
Costs Shape Modes of Production
Organization Failure point Type of failure
Markets Transaction costs Market failure
Firms Coordination costs Growth failure
Hierarchies Information loss Corpocracy
Peers Sharing costs Atomization
Coordination Arbitrage
• What’s the implication of coordination economies?
– Coordination arbitrage
• Entrepreneurs will begin to use coordination economies – PP‟s
natural cost advantage – to atomize value chains
– They will target industries with high coordination costs (and large
efficiency losses due to information economization)
– The coordination economy-diseconomy differential is the largest
here: coordination arbitrage
• Rents will be redistributed to peers as peer demand intensifies
and outstrips supply
– Prizes on Yahoo Labs‟ prediction market
• Firms will not disintegrate, they will move up and down the
value chain
– Peer management
– Platform supply
– Goods for which means of production/inputs access is limited or
protected
Peer Production Economics
…So What?!
• Where things get really interesting
– After all, firms can (to an extent) realize coordination economies by
using sharing/internal market mechanisms
…So what makes peer-production special?
• PP redefines two economic orthodoxies
– Diminishing marginal productivity
• Resources become less efficient at the margin
• Marginal productivity of variable inputs declines as quantity
increases
– Increasing returns
• Positive demand-side network externalities: Your joining the
network increases it‟s value to me
• Demand-side lock-in: My high up-front learning costs increase
my switching costs
PP Economics are Revolutionary
• PP models create increasing returns 2.0:
– Sustainable supply-side increasing returns
– How?
Increasing marginal productivity of labor
– 2 key mechanisms
• Distributed economies of scale create increasing marginal
productivity
– Cars, computers, electronics, books, songs….
– Many hands make light work at the margin
• Knowledge externalities create increasing returns to adoption
– My productivity increases in network size
– Your joining the network increases my value to it
– …because our knowledge in collaboration increases marginal
productivity for lumpy goods
Peer Production Economics
Distributed Economies of Scale
and the Atomization of Production
Ultraspecialization and
Microproduction
• Massively distributed ultraspecialization
– Doesn‟t work in the real world…
– Tiny chunks of ultraspecialized activities are valueless in isolation
• Lumpy goods take time and collaboration to make
• Who will pay for 1/1000th of one of the ten car production activities?
– Even less economical to assign coordination rights for
ultraspecialization
• Hierarchy or markets can‟t handle it
– How do you coordinate 10,000 tiny chunks of ultraspecialized activities in
the RW? How do you contract/pay for 10,000 tiny chunks of
ultraspecialization in the RW? Combinatorial cost explosions…
– Combining microchunks requires costly coordination, integration,
contracting, payment relative to every other microchunk
– If 10,000 microchunks, 10,000 x 9999 coordination+transaction costs
necessary
• Massive coordination diseconomies for ultraspecialized collaboration
– Result: market and firm failure for ultraspecialization in the real
world
Distributed Economies of Scale
• But…
– PP enables cheap, optimal assignment of coordination rights at any
level of divisibility
• Overcomes coordination diseconomies
– By leveraging cheap information sharing mechanisms to costlessly regulate
access to inputs and means of production
– Coordination economies make organizing microproducers for
ultraspecialized production viable
• Distributed economies of scale…
• …Yield specialization gains
– Finer and finer input divisibility lets us each specialize in our comparative
advantage
– Not simply „many hands make light work‟ – many specialized hands make
production more efficient
• PP community can be a more efficient producer than firm at large scale
– If enough microproducers contribute enough microchunks
Coordination & Specialization
Firm coordination costs
Value
PP coordination Costs
Specialization gains
PP coordination costs
exceed specialization gains
Size
Coordination costs bound specialization
gains. Since peer community
Firm coordination costs
coordination costs scale linearly, peer exceed specialization gains
communities realize more gains
Supply-Side Externalities
• What can an ultraspecialized network of microproducers
do?
– Peers contribute ultraspecialized microchunks of value activities
• I can produce 1/100th of a DAC and 1/50th of a spreadsheet application
– These microchunks are combinatorial, not additive
– So…
• Network effect: positive supply-side externality
– My marginal social productivity > my marginal private productivity
– I can‟t make lumpy goods efficiently by myself…
– …Your joining the network increases my value to it in 2 ways
• Microchunk modularity explodes production set
– Combinatorial explosion in feasible activities
– n^n-1 possibilities: 4 activities produce 64 possible combinations
• Microchunk reusability yields increasing marginal productivity
– I can reuse microchunks you‟ve contributed in the past
• In econ-speak, microchunks are non-rival and non-excludable in
production
Distributed Economies of Scale
• In general:
• Your activity set means my marginal productivity is enhanced
• Productivity increases in number of ultraspecialized microproducers
• Distributed economies of scale
– 10 specialized car activities become 1000 ultraspecialized car
activities
• ……creating more ways to recombine activities to improve efficiency
…allowing greater reusability of activities to improve efficiency
• Distributed economies of scale
– Recombining and reusing microchunks incurs massive coordination
costs in the real world
• Not all activity combinations in exploded production set are valuable…
• …and dividing activities doesn‟t always yield marginal productivity gains
• Finding beneficial combinations – assigning coordination rights
efficiently – is costly
– But cheap information sharing makes searching for best
coordination rights assignment frictionless
PP and Firm Production Functions
PP community is more
efficient producer
Output
Firm coordination costs
PP production function
Firm production function
PP marginal productivity exceeds firm
marginal productivity
Labor/Network Size
Distributed economies of scale mean that Firm is more efficient producer
at large scale, peer communities enjoy a
significant cost advantage over firms
Microproducers &
Ultraspecialization
• Ultraspecialization examples
• Jane knows everything about ab exercises (consumption externality)
• Dave‟s built replicas of every shoe since 1500 (production externality)
• Al knows everything about 80s Italo Disco (hobby)
• Trekkies, game mods, blogging,
• Why do microproducers ultraspecialize?
• Production or consumption externalities
• Hobbies
• Local effects: family, friends
• But mostly because they can:
– the Net‟s slashed search costs – specialization is less costly than it used to
be
• How did ultraspecialization evolve?
– …
Traditional Value Creation
Traditionally, value activities were integrated
because market transaction costs were high
Production: creation, editing, finishing
Management mediates transactions between
primary value activities
Management coordinates and exploits
complementaries between supporting value
activities
Publishing: finance, marketing, procurement
Disintermediation:
Integration to Disintegration
Cheap information disintegrates traditional
value chains: value activities rearrange in
horizontal layers, yielding efficiency gains
Production: creation, editing, finishing
Disintermediation: Markets mediate
transactions between primary value activities
Management coordinates and exploits
complementaries between supporting value
activities
Publishing: finance, marketing, procurement
Microproduction:
Disintegration to Atomization
Cheap coordination atomizes disintegrated
value chains: value activities rearrange in
microchunks, yielding efficiency gains
Production: creation, editing, finishing
Community mediates transactions between
primary value activities
Community coordinates and exploits
complementaries between supporting value
activities
Publishing: finance, marketing, procurement
Microproduction:
Distributed Economies of Scale
Microchunks have a positive supply-side
network externality: a peer‟s private
productivity is less than his social productivity
Why?
User N contributes green microchunk G.
This microchunk is reusable. Microchunk
reuse increases marginal productivity: G‟s
complementarity with other microchunks
reduces the marginal cost of future
productivity.
This microchunk is modular. Microchunk
modularity explodes production set. G‟s
complementarity with other microchunks
creates explosion in product set possibilities.
Distributed Economies of Scale
- Summary
2 Mechanisms
1. Coordination economies enable ultraspecialized production
2. Ultraspecialized production realizes network FX from reusing
and recombining activity microchunks
1 Effect
Marginal productivity increases in network size
Tiny bits of ultraspecialized knowledge can, aggregated across
enough peers, make peer production economically advantageous,
because of supply-side network effects
Condition: Lumpy Goods
• Distributed economies of scale obtain under 2 conditions
– Lumpy goods and divisible inputs
• Lumpy goods supply functionality in discrete bundles
– Computers, cars, runways, sunglasses
• They require specialized knowledge in multiple activities
– Coordination, learning, and switching costs of production are high
• Because opportunity cost of specialization is not knowing other
activities
– 10 activities required to make a car:
• 10 workers/car - 1 activity each
– Each worker is specialized – gains to specialization
• 2 workers/car - 5 activities each
– Only 2 activities/worker are specialized, workers must still perform
remaining 8 – relative productivity loss
– Marginal productivity increases in workers
• More activities/worker lower productivity
Condition: Divisible Inputs
• Divisible inputs
– Inputs must be able to be unbundled in arbitrarily fine
increments by microproducers
• You can contribute a sentence, a paragraph, or an essay
• …a scene, a script, or a review
• …a playlist, a vocal track, or lyrics
– For some goods, this is not possible (yet)
• Inputs are indivisible – can‟t be unbundled
– Because of cost, technology, regulation, or specialization
– Examples: cars, runways, oil rigs, MRI scanners
• Which goods have divisible inputs?
– Digital media: music, films, books
– Designs for low-complexity physical goods: T-shirts
– Knowledge resources: Wikipedia, journals, blogs, genes
– Without input divisibility, microproduction isn‟t possible
Peer Production Economics
Knowledge Externalities
and Increasing Returns
Recap
• PP models create increasing returns 2.0:
– Sustainable supply-side increasing returns
– How?
Increasing marginal productivity of labor
– 2 key mechanisms
1. Lumpy goods and distributed economies of scale
– Cars, computers, electronics, books, songs….
– Many hands make light work at the margin
2. Supply-side knowledge externalities
– My productivity increases in network size
– Your joining the network increases my value to it
– …because our knowledge in collaboration increases marginal
productivity for lumpy goods
Knowledge Pools
• Knowledge pools are repositories
– …for information sharing mechanisms about inputs and/or outputs
– They store the community‟s aggregated preferences and
expectations…
– To act as a collective memory for the PP community
• Examples:
– Product/feature reviews (Yahoo! Finance msg boards)
– Tutorials, guides, and walkthroughs (GameSpy walkthrus)
– User reputation systems (eBay seller feedback)
– Stored product rankings (Amazon product rankings)
– Archived playlists, tags, links (Blogdex archives)
– Metainfo about community production (Metatalk.metafilter)
– …Other metainfo
Knowledge Resources
• Why is knowledge important?
– Knowledge enhances productivity by more efficient use of inputs
• Intuition: algebra vs arithmetic
– Knowledge is an experience effect: a result of learning-by-doing
• Knowledge pool is the key resource for PP communities
– Users will select PP communities where they their marginal
productivity is maximized
– Knowledge pool is strongest signal of potential productivity gains
• Signals potential productivity gains of joining community independent of
network size
• Or: signals current productivity per peer
• And since microchunks are non-rival in production
– Knowledge resources don‟t get exhausted – they grow
• Knowledge pool is a resource stock
– That accumulates according to flow of knowledge externalities
Knowledge Externalities
• Recap: private info and preferences about inputs and
outputs is aggregated and made public
• This info sharing has a macro supply-side externality:
– Storing aggregated info creates a knowledge pool for the
community to learn from/access/utilize
• Inputs become more efficient
• Output quality increases
• Knowledge pool is a public good which acts as an input to
increase the efficiency of future inputs
• Initial knowledge pool is additive…
– My productivity increases when you join the network
• …Ongoing knowledge pool is cumulative
– Every time we collaborate to produce a good, new knowledge is
produced
Knowledge Pool
Knowledge is an
externality of value
activities: learning-by-
doing
Knowledge pools
when information and
preferences shared
during peer production
are stored
This knowledge
enhances future
productivity by
informing peers about
how to most efficiently
modify inputs in order
Value activities (brown) have a to produce the most-
knowledge externality, which flows desired outputs
into knowledge pool (silver).
Knowledge & Increasing Returns
• Leveraging this externality creates increasing returns
– Positive feedback:
• Network size growth increases marginal productivity
• Increased productivity generates new knowledge faster
• …New knowledge attracts more users
• Or…
• Your adding knowledge to the network increases my value
to it
– Positive feedback:
• Largest public knowledge pool attracts new users
– My marginal productivity increases in network size
– Public knowledge increases in my marginal productivity
– Your marginal productivity increases in public knowledge
– You join the network - my marginal productivity increases in network size
Doing and making cool stuff makes doing and making cooler stuff
more efficient
Increasing Returns Overview
New users ultraspecialized in different activities
Increase
Productivity at the margin (distributed economies of scale)
Which accelerates
Information sharing about inputs and outputs
Which grows
Public knowledge about the most efficient coordination rights
which attracts
…New users ultraspecialized in different activities
Knowledge & Community
• How is knowledge related to network size?
– Knowledge is an experience effect: learning by doing
– The more you do, the more you learn
– Productivity of users determines of knowledge growth
– Since marginal productivity increases in network size…
• Knowledge pool should explode in network size
– Identifying and managing this relationship is the key strategic
challenge in peer production
– Because it‟s the indicator (and driver) of increasing returns
• Speed economies become crucial
– Production throughput accelerates knowledge externalities…
– …which build critical mass in pool and accelerate network growth:
increasing returns
Increasing Returns
• Increasing returns means:
– PP communities will be winner-take-all markets
• The resource to which there are increasing returns is
knowledge, not just network size
• The biggest network won‟t win by default (like dot com 1.0) –
the most efficient knowledge generator will.
• Knowledge and network size are necessary to create natural
monopoly dynamics
– Dominant strategies leverage speed
• To build larger knowledge pools fast…
• Which attract new prosumers by offering discontinuous
marginal productivity gains
• …And build scale
Increasing Returns –
A Short History
• Where else do we see supply-side increasing returns?
– Accelerating down the learning curve – BCG
– Which came from…
– Disembodied learning-by-doing knowledge as public good in production
functions. Cumulative knowledge investment of all firms exponentially
related to capital stock (Arrow)
– Capital growth and knowledge externalities: (Romer)
– AKA endogenous growth theory
• How do these two theories of supply side increasing
returns differ from PP models?
– EGT assumes constant supply-side returns at the firm level, and increasing
returns only at the industry/national level
– Increasing returns in the BCG model aren‟t sustainable – the learning curve
flattens out
• Supply side increasing returns in PP models are sustainable as
long as knowledge pool grows.
Peer Production Strategy
Understanding Value Chain Evolution
Value Chain Evolution
Costly info and coordination integrates value activities – most economical way to produce
Inputs, Manufacturing, Marketing, Distribution, Retail
Cheap information disintegrates value chains, because segment consolidators realize specialization gains
and economies of scope – integrated production becomes costly
Inputs Manufacturer Marketer Distributor Retailer
Cheap coordination atomizes disintegrated value chains and explodes value activities within layers,
because microproducers realize distributed scale economies
Peers
Platform Peers Aggregator
Peers
Value Chains &
Value Appropriation
In disintegrated value chains, industry profitability migrates to the center (aka the ‘coordinator’)
5% 20% 50% 20% 5%
Inputs Manufacturer Marketer Distributor Retailer
Creatives Producer Publisher Distributor Retailer
Infrastructure Content Search ISP Telco
Why?
Market power in horizontalized landscapes – layer dominance – flows from relative scope (the center can
realize the greatest marketing scope economies of scope via marketing, branding, and advertising), or
relative specialization (the center can develop the most economically advantageous core competences)
EG: Nike, EA, SeanJohn, Google
Value Chains &
Value Appropriation
In atomized value chains, industry profitability will migrate to the edges
40% 20% 40%
Platform Peers Aggregator
35% 10% 10% 10% 35%
Inputs Platform Peers Aggregator Reconstructor
Why?
Market power in atomized landscapes is a function of relative scale: the edges can realize the greatest
relative economies of scale via demand and supply side network FX. The middle of the value chain
explodes and can’t realize scale economies. Core competences for disintegrated value chain coordinators
become core rigidities – value capture requires competences at both edges of the chain
The Explosion of the Middle
In atomized value chains, industry profitability will migrate to the edges
40% 20% 40%
Peers
Peers
Platform Aggregator
Peers
Peers
Why?
The center explodes and atomizes – hypercommoditization – and value shifts to adjacent segments,
because they can realize scale economies from the exploded center’s output.
Examples:
The Commoditization
of the Edge
In atomized value chains, the same technology that explodes the middle commoditizes the old edges
(example: cheap connectivity enables blogging). Industry profitability migrated to new edges.
40% 20% 40% 0-1%
Coordination Production Aggregation Connectivity
Blogger bloggers Bloglines ISP
AudBlog podcasters PodWorld BitTorrent
Community
reviewers Become
Owners
Value Appropriation Examples
In atomized value chains, industry profitability will migrate to newer (and newer) edges
35% 10% 5% 10% 35%
Coordination Production Publishing Aggregation Filtering
Platform Peers Publisher Aggregator Reconstructor
Laptop audio Artist Label Club DJ
Blogger bloggers hosting search Bloglines
MMOG Mods Community Gaming
gamers
Developer & plugins sites Open Market
PP Dominant Strategy:
Vertical Integration
• The point:
– Peer production communities are vertically integrated
atomized value chains…
• To capture the most value, integrate towards both edges
– …Plus knowledge pools
• Knowledge externalities don‟t pool without integration
40% 20% 40%
Coordination Production Aggregation
Platform Peers Aggregator
Peer Production Strategy
Competitive Dynamics
Peer Production Model
Increasing Returns
Knowledge pool grows New aggregated public info
Inputs Means of Outputs
(Coordination) Production (Aggregation)
Requirements Explicitly modifiable Open or community Efficient information
by community regulated access sharing
Revenue stream Sell access to users Advertising and % Sales revenue
sponsorship
Peer incentive Competition for % Ad revenue % Revenue from
access, prizes outputs
Example Wikipedia raw Wikipedia CMS Wikipedia entry
content
PP Competitive Dynamics
• PP model competitive dynamics
– You own a successful PP community facing new competitor
entry. What do you do?
– Positioning continuum
• Quality
– Operating model should be partly proprietary.
– Tight supplier relationships.
– New peer entry by invitation or referral.
• Quantity
– Operating model should be fully open.
– Loose supplier relationships.
– New peer entry open to public.
– What resources and capabilities drive competitive
advantage?
PP Strategy Building Blocks
• Key capability:
– Market-making
• Engineers increasing returns by exploding knowledge pool as
user base grows…
• …By creating efficient information-sharing mechanisms
– Which create coordination economies by vaporizing transaction
and coordination costs of production
• Results in:
– Economies of speed
• New users are attracted…
• …by accelerating knowledge pool growth
• …which increases productivity
• Or: the faster your throughput is, the faster knowledge pools…
• And reduces unit cost as network grows
How to Make a Market
• Markets don’t become automatically efficient
– Efficiency emerges from:
• Liquidity: assets are theoretically tradable
• Low transaction costs: assets are actually tradable
– Implications: graininess of information sharing mechanisms is key to
minimize information asymmetries between peer producers
• Profit maximizing traders
– Implications: must create incentives/gains for peer producers
– Peer producers must be able to internalize those gains
– Only way to avoid prisoner‟s dilemma – everyone expects someone else to
vote, Bush wins
• Goods traded can‟t be purely public goods
– Implication: access to means of production must be restricted to highest
valued peer producers: quasi-public goods
– What‟s the desired result?
• Information about which inputs used by which means of production
most efficiently produce the most desired outputs…
• …is traded and arbitraged, and the best solutions are revealed
Making Markets:
Efficient Information Sharing
• In the real world…
– How do you build efficient information sharing?
• By building incentives to leverage reputation and reciprocity
• By minimizing the risk of contribution
• By imposing punishment costs
– How?!
• By using an implicit or explicit currency
– Links are the currency for link aggregators
– Feedback points are the currency for eBay‟s reputation system
– Ratings are the currency for Amazon‟s book reviews
• Which is liquid and tradable
• Which creates incentives to profit
– To gain more links
– To get positive reputation feedback
• …and by eliminating transaction costs
A Quick Currency Case Study
• What is a currency?
– Karma/rating is not an implicit currency
• It‟s not tradable…
• …It has use value but no exchange value
– If I give you good karma, what does it cost me?
– Nothing – i have little incentive to reveal my true preferences
– Conversely, if I have good karma, what does it benefit me?
– Generally, I receive more privileges…
– But I can‟t trade it to get what I really want
– So I have little incentive to reveal my true preferences
• Your currency has to be more than a ratings system…
• …but not so complicated that peers won‟t use it
– What currencies have use and exchange values?
• Links – use value is linking from, exchange value is linking to
– Implication:
• Ratings and karma fail because exchange is unidirectional
• Real currencies succeed because exchange is bidirectional
Making Markets:
Sources of Transaction Costs
• Asset specificity
– Make information sharing objects not specific
• Example: tags
• Frequency and duration
– Make information sharing either frequent and small, or infrequent and large
• Example: playlists (infrequent and large) vs tags (frequent and small)
• Complexity and uncertainty
– Make info sharing mechanisms easy to navigate and predictable in results
• Difficulty of measuring performance
• Connectedness to other transactions in system
– Interdependent transactions can require complex coordination
PP Strategy Building Blocks
• Key resource:
– Peer network
• Owning the largest user network realizes increasing returns
– ….but only when knowledge pool explodes, attracting even
more new users, increasing productivity
– Knowledge pool explosion depends critically on market-
making capability creating efficient information sharing
• Results in…
– Economies of scale
• Speed and scale are interdependent
– If you can build both, you will achieve increasing returns
– … efficient information sharing mechanisms drive speed…
– ..and supply-side viral and network effects turn speed into scale
• Unit cost falls as marginal productivity increases in network size
– many specialized hands make production more efficient
PP Strategy Building Blocks
– And economies of scope:
• Scope is a secondary economy
• Production set explosion means unit costs can fall in network
size…
• …but only if means of production are flexible
• …and inputs are tradable
– It‟s no good having a community of diamond miners when
DeBeers owns all the mines
Drivers: Knowledge network size
Relative Advantages
Peer Network
Efficient Info Sharing Without a market making capability…
Public Knowledge …public knowledge doesn’t pool
New Users … new users aren’t attracted
Productivity …marginal productivity doesn’t grow
… increasing returns fail
Peer Production Revenue Sharing
• PP is not about getting others to make your stuff for ‘free’
– PP is way to create more value via coordination arbitrage
• …But there‟s no such thing as a free lunch
– Revenue sharing will emerge as supply-side price competition
• Peer producers will appropriate more value as communities grow
• Should you share revenues? Yes!
– Sharing creates incentives for productivity and knowledge growth
– But share strategically – not as margin-eroding price competition
– Strategic revenue sharing targets constant margins and revenue growth in
network size
• Strategic revenue sharing
– What determines the peer producer revenue share?
– Rivalry for peer producers within your industry
• Pre-empt margin erosion…
– By limiting rivalry for peer producers
– The more effectively you limit rivalry, the more you can share without
margin erosion
– What limits rivalry for peers?
Peer Production
Industry Structures & Rivalry
Suppliers & Buyers:
Control access to key technologies,
inputs, and distribution channels –
vertical integration
Control buyer/supplier scope economies
by controlling high-value complements
Fragment/commoditize key technologies
or inputs if you can‟t control them
Example: Amazon – Fedex deal
Peer Production
Industry Structures & Rivalry
Suppliers:
Complements: Buyers:
New Entrants:
Control access to key technologies or Control distribution channels – vertical
inputs – vertical integration integration
Create larger numbers of higher Vertical integration creates strong
Fragment/commoditize key
quality complements than competitors Fragment/commoditize distribution
entry barriers
technologies or inputs if you can‟t channels if you can‟t control them
control them EG:
Raises switching costs by lowering Scale and speed economies in
EG:
relative value of substitutes network markets create strong entry
barriers
Example: lightweight, open standards
(RSS) drive cheap complementarity Other sources: first-mover advantage,
preemption, deterrence, complements,
Anti-example: Apple closing off nonlinear pricing, differentiation
iPod/iTunes platform kills
complementarity Example: eBay‟s user network
Where PP is Inferior
• PP is an inferior way to organize the production &
design of goods
– For which access to means of production is unavailable
• Tightly controlled by suppliers
• Aggregate peer network learning costs higher than opportunity
cost of production
• Capital and connectivity intensive
– For which sharing isn‟t viable
• No learning externalities
• Connectivity issues
– Inherent properties of goods
• Goods not lumpy
• Durable goods
– These properties limit rate of knowledge pool growth and kill
increasing returns…
– Or limit ultraspecialization gains and kill distributed economies of
scale
Where PP is Superior
• PP is a superior way to organize the production &
design of goods
– For which real-world production transaction costs are high
– For which real-world production coordination costs are high
• But for which , virtually, production and consumption
information can be efficiently shared, realizing coordination
economies
– For which ultraspecialized knowledge is dispersed
throughout the larger population
• Which can realize distributed scale economies
– For which inputs are divisible
– For which the means of production are accessible
Summary:
Peer Production Economics
• Assigning coordination rights is costly in the real world
– But virtually, coordination economies make atomized
microproduction viable
• By efficiently sharing information about about who and what should
have access the means of production
• Atomized microproduction…
– Yields ultraspecialization gains
– Reusing modular microchunks of production create increasing
marginal productivity
• Distributed economies of scale result
• Knowledge pools
– When goods are produced, and new shared information is stored
• Increasing the future productivity of the community
• Attracting new users, who make it more efficient to produce more
goods…
• …sustainable supply-side increasing returns
Summary:
Peer Production Economics
• Disintegrated value chains will atomize
– Cheap coordination will atomize and explode the center, where
production of goods generally happens
– …and value will migrate to newer and newer edges
• The dominant peer production strategy
– Is to vertically integrate across the entire value chain
• To capture the most value by providing infrastructure, inputs, and
marketing/aggregation/filtering of outputs
– This can be seen as a shift from goods to services in many
industries
• In order to achieve this strategy, you need…
– Key resources and capabilities in market-making…
• In order to enable efficient info sharing and engineer increasing returns
– …and a sufficiently large and productive peer network…
• In order to realize distributed economies of scale
Summary:
Peer Production Models
• Open or community-regulated access to the means of
production
• Open or community-regulated access to modify inputs
– Low level inputs require little technical knowledge
• content, designs, ideas, proposals
– High level inputs require complex technical knowledge
• product architectures, interfaces, features
• The catch:
– Supply-side productivity gains depend critically on
information sharing to create knowledge externalities.
• How do you know how to modify inputs?
– Without knowledge externalities, the PP model can‟t reach
hyperefficiency.
– Supply side effects alone are not enough
Summary:
Peer Production Models
• Demand-side sharing mechanisms make private info and
preferences about inputs and outputs public
– Playlists, tags, votes, ranks, values, reputations…
• Aggregating sharing mechanisms reveals:
– Utility functions & preferences – most efficient outputs
– Private technical expertise & info – most efficient inputs
– The cream will rise to the top - the most efficient inputs and
outputs become transparent
• Assuming search costs are low
• Efficent sharing pools knowledge
– Which creates sustainable supply-side increasing returns
– No sharing without market-marking
• Leverage implicit or explicit currencies to build incentives
– No productivity gains without sharing
• Knowledge must be a public good
Thank You
Heuristics
Peer Production Advantage
Coordination Economies
Low High
High
Dot Com 1.0: Search engines, portals, Peer Production
networking services
Search Economies
Low
Yellow Pages Fabs
Coordination Arbitrage Criteria
Coordination Diseconomies
Low High
High
Hedge Funds, VC Ford, GM
Input Indivisbility
Low
Lifestyle goods Peer Production Sweet Spot
Media…
Relative Advantages in…
Market Making Capability
No Yes
Amazon Peer Production Community
Yes
Community Size
No
Out of the Game eBay
Vital point
Who’s Attracted?
Knowledge Pool Quality
Low High
Slackers: Everyone:
High
Inefficient info-sharing Sweet spot
Restrict peer entry by invitation/referral
Knowledge Pool Size
Shift to part-closed model
Control quality of complements
No one: Geeks:
Low
Vital point Expand scope
Shift to fully open-access model
Leverage viral effects & complements
to win peers
Productivity Explosions
Knowledge Pool Size
Low High
High
Productivity Implosion Real-World Productivity
Community Size
Low
Real-World Productivity Productivity Explosion
Peer Production Cost Advantage
Gains to Specialization
Integrated Disintegrated Atomized
Size of Production Set
Notes
Peer Production -
Resource & Capability Dynamics
Output
User network
Productivity
Speed Effects
Scale & Scope Effects
Action
Coordination
Knowledge
Externalities
Information
Exchange
Increasing returns
Knowledge pool
Analyzing Network Economics 2.0
• 3 dimensions of next-gen b-models:
– Horizontal scope (products)
– Vertical scope (features)
– Peer productivity (knowledge network size vs user network
size).
• 3 economic variables:
– Scale, scope, and speed drive positioning on the above
dimensions.
2.0 Strategy Space
Vertical Scope (Product Depth)
Horizontal Scope(Portfolio Breadth)
Peer Productivity
Scale, Scope,
and Speed Economies
• Scale
– Reduction in unit costs due to increase in operational size
• Scope
– Reduction in unit costs due to increase in number of products
marketed
• Speed
– Reduction in unit costs due to increase in number of products put
through supply/demand chain
Coordination and Transaction
• Coordination
– Search
– Matching
• Transaction
– Negotiation
– Bargaining
– Enforcement
Related docs
