Yochai: What I want to do is talk about free software and more generally, the possibilities on the Internet for new models of production. We see this emerging throughout the digital network environment. This offers some very interesting answers to questions regarding non-proprietary production.
Series of observations about free software and peer production. What are the characteristics? The framework within which it lives? How do we begin to understand how people are collaborating to create all sorts of interesting things? I will explore all sorts of business models, and tie this to what we've been talking about regarding political and cultural values.
So: free software. It works. Lots of it works. It's getting harder and harder to ignore its success. Apache is the dominant web server in the market. It can't plausibly be the cost, or lack thereof, that people use it.
What is free software? How do we define it in relation to proprietary software? Proprietary based on exclusion. You must pay for it, you cannot study it, customization is limited, and redistribution is not permitted.
Free software inverts these facts in many cases. It's not an absence of copyright. The license is a self-limitation of the owner's control. It permits adaptation. It enables learning. Allows redistribution. You may modify as you like, provided you notify others of changes and put it under copyleft.
Free speech/free beer distinction. It is free as in speech. You can have IE for free; it is not free software. Identifying as "free" through cluster of uses.
An anatomy of free software; Eric Raymond and [...] Moody describe it. Programmer writes, releases; others use, modify, extend; mechanism for communication; volunteers with different levels of commitment and influence focus on different aspects.
I use the term "free software" rather than "open source software." This is a religious distinction. Free software is about signifying a politics.
What is the institutional framework? Three models: proprietary, public domain, and copyleft. Property is the core of market and firm-based production. Parameters of exclusion permit charging a price and controlling output of employees.
At the other end is the public domain. All of the senses that we describe software as free would be available if the software was placed in the public domain.
Weakness of putting it into the public domain: ease of defection/reappropriation. Someone else can take it, modify it and sell it. The weakness is not in the lack of freedom; the weakness is that no one will be able to build upon it.
Copyleft--the mechanism used in GPL and the majority of free software. Dominant. Irrespective of political debates, the fact is that the majority of the projects are under the GPL. This is a cluster of provisions to ensure that no one will make the software proprietary.
Distribution is always accompanied by the source code. You cannot prohibit redistribution downstream. Freedom to modify is kept intact. You must give on the terms that you receive. If you want to benefit, you must benefit others. Covenants run with the program.
You may come into my living room as long as you behave yourself. You stop behaving yourself, you must leave.
The basic point: it goes from one generation of user to the next.
Then there are the questions: what counts as a modification? And an attack: IBM SCO case. Whether or not Linux is infringing. But it doesn't go to the basic question of the validity of the GPL.
What are the advantages/disadvantages of copyleft v. public domain? Advantages: 1/reduces incentives to make proprietary, 2/reduces opp. for "defection," 3/retains the integrity of contributions as a part of the peer-review process.
Why is this phenomenon described in such a way that people imply that it only works with software? It works elsewhere as well. Using a more abstract definition, 'peer-production," we see it everywhere on the Internet.
Definition: 1/collections of individuals 2/effectively produce information goods, 3/without price signals or managerial commands.
Human parallel to distributed computing: SETI @ Home.
This is not that but it's similar.
Let me tell you a few stories to give this more flesh.
Some common examples of peer production: 1/traditional academic research, 2/the Web itself, 3/content at sites like Kuro5hin, Wikipedia, MMOGs.
Mars clickworkers. NASA experimented; they had a few PhDs looking at maps of Mars. They identify craters. They decided to cut up the map and let people on the Net mark craters. 85,000 people volunteered, checking out each piece of the map. The results were indistinguishable from those of a fully trained PhD. Cutting this up into little modules. 5-minutes of effort.
Kuro5hin: as close as I've seen to peer-produced, op-ed level material. Produced by many, many people. Pre-publication queue. The software generates the number of votes one needs to publish the piece. 95 people have to say it is good enough to publish. You vote. And it is not only voting. There are also comments.
This is the basic structure of peer-review publications. And this is of a decent quality.
WikiPedia. Is it fantastic? No. Is it as good as Encyclopedia.com? Yes. Interesting thing about WikiPedia is that it depends of transparency and ease of correction. Mistakes are shallow. This is all based on social norms.
More detailed content. Why an encyclopedia? Why not a novel? Small chunks of information. Modular.
There's this question of relevance. We can produce information. How do we know it's worth anything? You can produce relevance and credibility not through the publisher, but through a community filtering mechanism.
Slashdot: here's a story, and a number of comments. I've filtered it a level already. Organized by highest score first. Adds relevance.
Other things for relevance. Google. It's a peer-production enterprise. It ranks according to what people on the Web think is relevant to link to. Innovation in Google was to use as a primary way of ranking how many linked to a site. It guesses accurately what is relevant. It gives us a snapshot of what is relevant, farming out the most important aspect to peer-production.
Overture is the opposite. [Demos what Google gives you when you search for "Barbie"; contrasts with Overture results.]
At Yahoo, they paid people to look at pages and categorize; open directory in many cases does better.
To round out claim about Barbie, let's look for "democracy" at Yahoo. [Contrasts with Open Directory entry; it is better.]
Not entirely unexpected that the politics of a non-market based system will provide "alternative" views.
Another example: distributed proofreading. Thousands of pages done this way.
First question on anyone's mind: Why do people do this? A puzzle to economists. Two bins: intrinsic and extrinsic rewards. Instrinsic: hedonic. It's fun! Community ethics: we do this to be a part of the community.
Extrinsic: Human capital, reputation enhancement. Service contracts and widget sales.
This is the model we get from the economists looking at this.
People are driven by diverse motivations; in other words, they care about things other than money. Social-psychological interactions. The presence or absence of money can actually decrease the social-pyschological rewards. Offered money as an academic to write an article--it is different than writing it as your own research.
Money matters, and it doesn't necessarily increase incentives to do something.
We have a lot of issues in society where money is correlated to negative things: transactions for sex, etc.
What this means is that as we begin to recognize peer production, companies will have to manage a peer production enterprise with sensitivity.
The important thing here is organization, not incentives. Important aspects: modularity, granularity, integration.
The value of this in economic terms. Human capital is hard to define. By comparison to firms and markets, peer production has 1/information gains and 2/capital gains.
The locus of decision making about the contribution is with the contributor, who has the best info about his or her availability. Some people are wrong about the value of their contribution; this is why the peer review is important.
As Larry said, the word "commons" is a problem. There are problems that property solves; that is not nonsense. We do have commons problems. Different kinds of solutions to these problems.
Primary concerns: 1/defection, 2/poor judgment of participants, 3/putting it all together.
Solutions: 1/formal rules like the GPL, 2/code like Slash, 3/social norms, and 4/redundancy and averaging out.
Primary approaches to improving integration: 1/iterative peer production of integration, 2/reintro of market and hierarchy w/low cost and no residual appropriation (Red Hat). Most of the cost of the service is iteratively peer-produced.
Business models: 1/surf on what's there (Google), 2/services (Red Hat), 3/toolmakers (SourceForge, MMPOGs).
Wrap up: we see 1/diverse motivations w/complex relationship to money, 2/peer-production, not OSS, 3/anti-defection mechanisms, 4/working businesses.
Participant: Many open source developers make their living working in companies who sell proprietary software. What will happen if free software kills proprietary software?
Yochai: Question is, how will they live? All you will have is hobbyists, yes? Answer: we have enormous slack. We have found ways to combine play and production. There is a sum of money in the US paid to developers through customization and services. They will do okay. Enough non-software derived revenue.
Participant: Do you see the two types of production as mutually exclusive? Are there changes you think are needed in terms of legal framework within which both are offered?
Yochai: I don't see the methods as mutually exclusive. The primary threat to peer production is that policies aimed to make the world more congenial to traditional production can undermine it. [...missed a bit, cites DVD/DeCSS cases...] The series of behaviors we see are ill-conceived even for proprietary production.
[Concluding remarks; end of session.]
Last updated February 19, 2008