Peer Production

Peer-production of information

2:00 May 14, 2004

Why do people share?  What is the theory of motivation?  That allows us to understand phenomena like free software and shared/distributed computing.

What are the enabling characteristics?

• Information production and human creativity
• Shareable goods

A transaction costs theory of sharing and peer production as organizational alternatives to markets and hierarchies

The emergence of social production

Free Software

One role to play today – you can’t ignore this  -- shows Apache server software as compared to other brands

People use Apache because it works (price is not the determining factor here)

Free sw development is happening, making products that are effective, under a model where no one exercises exclusive property control.

Theories

1. The first generation of responses was that software was a special kind of property.
2. Software developers are a different tribe.

• Peer Production
• Various sized collections of individuals
• Effectively produce information goods
• Without price signals or managerial commands
• Sharing material resources
• Large scale practices of effective productive sharing
• Compinentns are matieral resrouces,

Peer Production (examples)

• Academic Research
• The Web
• Content (clickworkers, K-5, wilipedia, mmogs)
• Relevance/accreditation
• Commercial – amazon, google
• Volunteer – open dir project, Slashdot, Kur05hin
• Distribution
• Value added – distributed proofreading

Talks about NASA project to throw crater pictures on the web with training for volunteers to mark those craters – 85,000 signed up.

So, what about something more sophisticate – like op eds.

Kuro5hin

You vote what gets published (hmm.. interesting model for micro press).

Wikipedia

Online encyclopedia – thousands writing entries.  Online group editing (imagine a political site allowing policy to be created this way)

Google outsources its single most important function to peer production.  The reason people go to Google is that you find what you want on the first page.  It counts links (votes) across the net.  This judgment is peer produced.

But, this is not neutral (no peer system ever is).

So, we look at

Material Sharing

• SETI@Home
• SETI has twice the teraflops capability compared to commercial/govt systems
• 4.5 million people sharing nicely
• Other@Home projects
• Folding@Home – 572k cpus, 28,000 teams – producing raw computation
• Open Wireless networks
• Kazaa
• Skype
• Carpooling

Open wireless networks is another form of social production/sharing

Motivational Feedback

• Information about the amount of work
• That each user contributes
• That the collective effort achieved
• Information about the scientific context

Used to be supercomputing was large expensive processors

Then you solved the problem by breaking problem down and running on multiple processors

Course you then assumed all those processors were crunched together in a box.

What if we broke the processor model up – breaking the problem even further so no processor had to wait on another for the answer to come and be modified by the computer.

Motivational Theory

• Agonistic giving
• Non-agonistic giving
• Individualistic and solidaristic
• Reciprocity – P2P and OWLs
• Are these projects shooting in the dark or is there no crowding out among forms of sharing?

Looking at car pooling

SOV 75% travel alone

17% of commuters use carpools

• Money not generally reported
• Where money exits, it is as explicit cost sharing, not price
• Evenin dynamic  caroools, complete strangers pick up at known points, no tpatterns of discrimination.

Economic systems

• What are the motivations
• What are the feasibility conditions?
• Is it more efficient, and if so when?

Diverse Motivations

• OSS economic literature maps the diverse approach mechanics

Motivation crowding out theory

• Motivation by monetary crowd out those that would do it for the sake of doing good
• Frey: social psychology focuses on intrinsic and extrinsic motivations
• People feel rejected and un-trusted when offered money to do something that a well-respected person would do anyway.
• Benabou & Tirole : People take queues from others in authority – if offered money they lose confidence in their abilities

Social exchange and Social Capital

• Carpooling, p2p file sharing, includes an instrumental component not accounted for on the psychology-based theories.
• Anthropology of gift literature includes heavy emphasis on reciprocity, social hierarchy
• Social capital focuses on instrumentalism
• Empirics: fehr & Gechter 2002: reciprocity crowded out by money

Combined

• Humans are diversely motivated
• Reward function includes
• Material motivations expressed in money
• Social-psychological motivations
• Different motivators have a complex relationship to each other
• Dinner with friends, sex
• Writing a legal opinion

The culture of money is contextually crossed to either motivate or de-motivate (you pay a lawyer to write brief – good, pay a judge to write opinion – bad as it is seen as a bribe)

Organization not incentives

• Peer production limited not by total cost or complexity of project, but by
• Modularity (how many can participate, how varied is scope of investment)
• Granularity (minimal investment to participate)
• Cost of integration
• The more a project requires many small contributions, the greater the advantage of social production over market, priced-based production

Transaction costs of markets are higher than transaction costs of social exchanges

Enabling characteristics

• Human creative labor
• Innately possessed
• Implicit – hard to codify – experience, wisdom, judgment
• Hard to specify for complete contracting or management
• Varies widely across and within individuals, with respect to project, resources, and collaborating with others
• Shareable goods (descriptions are not absolute – depend upon context/culture)
• Lumpy
• Come in discrete packages of functionality – producing resources/goods, that do not align perfectly with demand for the functionality to flow.
• Mid-grained
• Packages can be provisioned to a substantial segment of a population, given wealth, cost, and demand for functionality flow over the lifetime of the good.
• The present technological context: the most important imputs, into the core economic activities, of the most advanced economies, are widely distributed in the population
• Human creativity
• Computation and communications resources

Is it efficient?

Information gains

• Human creative labor highly variable
• Difficult to specify for market or hierarchy control
• In peer production agents self-identify for and self-define task
• Have best information about ability in time
• Mechanisms for correcting misperceptions necessary : e.g. “peer review”

Firms and markets require crisp definitions on what is given and what is paid.  Its decision system requires crisp, clear agreements as to units of transaction.

Tradeoffs

• Social systems have similarly high set up costs, but lower per transaction costs
• Tradeoff formal computability for texture
• Social sharing particularly good where inputs or outputs hard to measure

Allocation gains

• Property and contract reduce uncertainty of agents and resrouces
• Individuals highly variable in fit to resources, projects, and each other
• Substantial increasing returns

The Commons Problem

• How avoid freeloading
• Different problems requires different solutions

Role of Culture

• Tech characteristics create threshold barriers
• Beyond threshold 000
• Prevailing social institutions, etc differ among societies
• Based on relative attractiveness of money and social-psych rewards
• Trust, experience with cooperative practices
• Presence or absence of formal background institutions supporting, and investments in, one or another system

Under certain tech conditions it becomes feasible for social production to play a role in economic conditions

Typified by decentralization of authority

Reliance on social information flows.

One time ideal market transactions – which means the individual is not investing in long term relationship – have relationship characteristics like markets, operating in a social system.  Carpooling and free software are similar in these