open source is/isn’t

Til the weekend, I may not have enough time to respond to the good comments on the Red Penguin series distinguishing the thoughts of Marx from the 20th century implementers of Marxism.
But wanted to add a few thoughts about the motivation behind the series. I first wrote Love and Money and Red Penguin as reflections on Benkler — and added the narrative link to Marxism because it was the Hunter article that led me down that path.
Peer production — open source software, wikipedia, weblogs and the rest — is a new way of organizing creation and people.
What’s most interesting is understanding how this wonderful thing works — what are the principles, properties, processes and passions that make the whole thing go, and that make particular things thrive.
So, the interesting avenues to follow, after the classics by Raymond and Benkler and the FSF crew, are the practitioners and observers who are doing it and figuring it out. Good references welcome.
Analogies to existing frameworks like Marxism are interesting inasmuch as they shed light on this new thing.

Atom Aggregator Mixmaster

Benjamin Reitzammer wrote in about another aggregator cuisinart project using Atom to gain more control of feed processing.

If you dont already know about it, you may find dbagg3 from Leslie Orchard interesting too. Right now it seems like “yet another aggregator”, but from what he outlined in his early posts about dbagg3, the aggregating feature is only one of many, and easy searching/processing of Atom feeds/entries is another. You can find Leslie’s latest post about dbagg3 at

The comments feature on this blog is now fixed, which is a mixed blessing. Good comments as above posted directly to the site. The morning comment despam routine.

Love and Money

Yochai Benkler’s classic essay, Coases Penguin, explains peer production as a third classic means of organizing economic activity, parallel to the marketplace and the firm. Peer production has a distinct set of advantages for information products, where costs of communication and distribution are low.
Benkler notes that in the real world, of course, market and firm aren’t mutually exclusive. There are many variants and hybrids between pure market and pure managerial forms. Industry ecosystems typically include a mix of big companies, small suppliers and service providers, and market-based commodity purchasing.
Similarly, we would expect to see hybrid forms combining aspects of peer production and the other classic market and firm-based forms. Following up to Benker’s analysis, here are some thoughts on some emerging combinations.
At Socialtext, we see companies using wikis to generate peer-created internal knowledgebases to share information, replacing centralized, structured collection and publication processes. Companies are finding that removing barriers to contribution by a larger number of contributors improves the speed, quality, and amount of content.
Another hybrid of organization form is for-profit customization and service on an open source base. This isn’t just “rent-seeking” — the economist’s slur for exploitation of freely contributed work. It’s a way to fill a gap in the peer production model.
Benkler describes the traditional method of prioritization in a firm as “managerial.” Resources are allocated by a manager’s dictate. This description ignores a key factor in managerial production — the customer. Managers gather customer needs, and ensure that production is done to customer schedule.
The advantages of peer production are greatest where the developers are the customers. When customers are separated from developers, by skill set or time priorities, there needs to be some way to communicate the customer’s needs to the developers. Customers also bring deadlines, which are foreign to pure peer production. Benkler writes that peer production is more efficient over time because of its ability to marshall vast resources. “Peer production relies on making an unbounded set of resources available to an unbounded set of agents, who can apply themselves toward an unbounded set of projects.” But if a customer needs functionality by a specific moment, it won’t help that a given free software project will probably develop that capability eventually.
Agile development methods — where customers and developers work collaboratively to set priorities against deadlines — take advantage of lowered communication costs to increase customer input and reduce risk. Using these methods on an open source base allows producers and customers to take advantage of peer production low cost and high-quality for things that are generic or non-time-sensitive.
With agile development, money is a measure of practical empathy — you get paid when you understand and meet customers needs and priorities over time.
Another interesting and puzzling question is the mix of incentives in a hybrid world. Where financial and nonfinancial motivations co-exist, how can money be introduced without discouraging people who contribute for free, for fame, satisfaction, personal need, and other nonmonetary incentives.
As noted in the previous post, some activities, like sports and music, where participating for money isn’t seen as mutually exclusive to participating for social rewards and personal satisfaction. There are fields like academia and scientific research, where participants make a living, but choose to earn substantially less than their peers who work in law practice and industry. There are also fuzzy lines in artistic and academic communities between being successful and “selling out” — tilting the balance all the way toward money, and away from values respected by peers, like unbiased research.
Benkler notes that there are some areas where perceptions change over time. In Shakespeare’s time, professional actors and musicians were looked down on; in the 19th century, professional athletes were disrespected. These values have reversed.
Given the substantial economic activity surrounding open source (IBM, anyone?), there are many people negotiating the “love and money” boundary every day. Personally, I like serving customers, and think it is fortunate that it is customer focus that makes money in an open source ecosystem. Much of the innovation in social software is coming from peer production — the way to innovate is to participate in the game. And, like anyone in a business based on “commons” resources, I want those resources to thrive, for reasons of inherent value, and for self-interest.
What do y’all think?

Red Penguin?

Dan Hunter’s article arguing that open source development is a form of contemporary Marxism led me to read Coase’s Penguin, the classic paper by Yochai Benkler that provides an economic explanation of open source software and other peer production endeavors like Wikipedia.
Marxism argues in favor of collective production and against monetary rewards out of political belief that capitalism is inherently exploitative. The way to ensure a just society is collective production where production is organized and rewards are distributed fairly through central planning. But centrally planned collective production proved inefficient and corrupt.
The first puzzle about open source peer production isn’t whether or not developers have marxist political beliefs, but why it works, especially since the Marxist collective model failed miserably.
This is what Benkler explains elegantly. Coase’s Penguin builds on the theory of Ronald Coase, who explained in the 30s that firms exist when the cost of separate transactions with many independent parties is greater than the price-efficiency of a competitive market. The problem Coase was trying to solve at the time was to explain the persistance and dramatic growth of centrally managed corporations, if a market is an ideal way to allocate economic resources.
Benkler solves today’s version of the same problem. If money is the ideal way to incent and co-ordinate production, why are we seeing the persistence and dramatic growth of production methods that don’t use money?
Benkler explains that commons-based peer production is more efficient than either firms or markets for information goods, where the costs of communication and distribution are low, and the difficult problem is allocating human creativity. When there are masses of potential contributors, and it’s easy to participate in little chunks like an open source plugin or a wikipedia article, the best way match skills and work is a million little decisions by independent contributors.
Mandatory, Marxist-style collective farming doesn’t benefit from these resource allocation efficiencies. Workers on collective farms have pre-defined work and can’t leave. Collective farms don’t gain the benefit of unique, voluntary contributions by thousands of distributed workers.
Another attribute of political marxism is an belief in mandatory equality. Peer production projects often have a meritocratic culture with dramatic inequality, where founding leaders and high-value contributors have greater prestige, influence, and sometimes financial reward. It’s not considered inherently unjust that leaders of open source projects like Perl and Python have received grant, foundation, and corporate funding to do their work (although visible leaders of peer projects can also become lightning rods for criticism).
Another marxist value is opposition to a money economy. Cash is seen as a symptom of the alienation of workers from the products that result from their labors.
Clearly, the motivation of many thousands of open source, wikipedia, livejournal, and other peer content producers is non-monetary. But is it anti-monetary?
Benkler deals with the incentive question in the excellent third section of Coases Penguin. Benkler makes an astute distinction between activities where money is commonly thought to be an inverse motivation (sex), and where it is seen as complementary (sports, music). Many people who like basketball would love to be NBA stars. By contrast, most people who like sex would not like to be prostitutes.
Some Free Software activists are in fact marxists, with beliefs that money is inherently exploitative, and visions of a world that is socially and economically organized without money.
The GPL license, a strict license that forbids the redistribution of modified code with a nonfree license, doesn’t forbid selling the software in a package, or customizing software for money, or selling services based on knowledge of open source tools.
For many people, software development is pretty clearly in the complementary category, where the rewards of prestige and satisfaction coexist with monetary rewards. There are Apache developers on corporate payrolls, and companies supporting open source technologies, ranging from IBM to MySQL, Zope, and Jabber. There are developers who make a living consulting based on free software expertise.
So, while some software developers are marxists, it doesn’t follow that peer production is inherently Marxist.

Anarchist in the Library

I loved Siva Vaidhyanathan’s last book, Copyrights and Copywrongs. It’s a superb excellent intellectual and cultural history and critique of copyright policy.
His new book, Anarchist in the Library isn’t up to his standards. “Anarchist” tries to draw a connection between the spread of peer to peer file-sharing, peer cultural creation, and political anarchy.
The analogy is thin, though. Anarchism is a deliberate political philosophy that opposes central government. My guess is that most file-sharers aren’t making a political statement, they just want access to music.
The bigger problem with defining file-sharing as anarchy is that it focuses on what’s absent — central control; rather than what is present — strong and shifting networks of cultural influence.
After a brief historical period dominated by mass media, we’re seeing a revival of folk culture, with new forms of peer cultural sharing and creation — file sharing, blogging, mashups. The trend has been growing since the advent of cheap photocopiers and cheap videocameras, and accelerating with cheap distribution and improved tools for sharing taste and collaborating.
The portrayal of culture as anarchy is a Romantic notion, shaped by the ideal of the artists as lone rebels or dissident cliques. That concept itself is the result of the mass media dominance. Artists see themselves as an embattled minority, then their work gets co-opted into mass media (Lennon’s Revolution selling sneakers).
With the rise of mainstream folk culture, though, the interesting structural observation isn’t the lack of central control. It’s the emergence of networks of influence that are shaped by taste, by opinion, by identity, by personal connection, by mentorship.
Vaidhyanathan laments the lack of community formed around Napster. But that was just immaturity. We’re just inventing tools for groupforming around shared preferences and collaborative creation. Flickr has cool tools for building groups around sharing pictures. If Napster was allowed to live, if music-sharing were legal, we’d see faster growth of social software around music.
“Anarchist” segues from Napster to chapters showing science and libraries under attack by increasing corporate and political control. I found those chapters more interesting and informative, probably because I knew less about those topics than then internet copyright wars.
But the anarchist argument still wasn’t all that persuasive. There’s a strong case to be made that a balance is shifting toward control. But the converse — that science and libraries are inherently anarchistic — just doesn’t hold up. David Weinberger recently published an interesting piece on cultural bias in the Dewey Decimal system. Western science and technology has always had an alliance with military and industrial forces.
Also, the book’s politics contain a bit of kneejerk Chomskyism. Lets get this straight — third world unlicensed DVD factories are good guys, fighting US intellectual property protectionism. Meanwhile, the Brazilian domestic aerospace industry are the good guys when they implement protectionism, fighting US free trade. IP production is bad when the US does it, and good when the 3rd world does it.
In summary, Anarchist in the Library has some interesting ideas and information, but is a disappointing book overall. Beyond this book, Vaidyanathan thinks and writes well about interesting and important topics, and I look forward to reading more good books from him in the future.

IP reform is environmentalism, not Marxism

Wharton professor Dan Hunter has written a provocative paper in which Lessig-style free culture activism, open source software, and open spectrum are examined for their relation to Marxism.
The paper concludes that Lessigist IP reform is more like social democracy, which tempers the exploitative excesses of monopoly capitalism with safety nets; whereas the open spectrum and open source movements are more like Marxism, in that they attempt to remove property from the exploitive hands of owners.
I think the paper gets its Marxism wrong. For example, “a commons of any sort is inherently Marxian, even if other types of private property rights still operate within the commons.” Nope. Marxism argues that all property is theft, and all property is to be held in common. And “Marxism isn

Atom Wiki and the Writeable Web

At Foo Camp over the weekend, Ingy paired with Ben Trott to enable Atom posting from Kwiki to Movable Type. Their work complements the work of Autrijus Tang, who added Atom feeds and Atom posting to Kwiki using Ben’s XML::Atom perl module.
We’re following Joe Gregorio’s experiments a year ago, adding Atom support to PikiPiki, a python based wiki that’s the parent of MoinMoin.
For newcomers to the world of blog and wiki, Atom is both an API that lets you read from and write to websites from other websites, and a web content syndication format that’s intended to be more tightly standardized than RSS, to support more aspects of web content, including images and videos.
The initial impulse to do at Atom to a wiki is to enable offline editing. It’s good to be able to use desktop clients like Ecto to post to a wiki. That’s helpful if you want to compose or edit when you’re offline, and synch when you’re back online.
The second drive is publishing collaborative work. Many people want to develop content in an an open and collaborative process, and once it’s ready, to publish it to a wider audience. Atom lets you develop collaboratively in a wiki and publish out to a weblog or public website.
Things get interesting when you consider combining with other applications. Autrijus added Atom support to RT, an open source trouble ticket system. Think about posting problem log information to a wiki, to use in developing an FAQ.
Ben Hammersley brainstorms about creating a web-based proxy that will post to the community website of your choice. “I want to be able to choose multiple endpoints for a post, and publish to all of them with a single button click.”
Grant Young brainstorms about using feeds to “present issue-based portals very quickly and cheaply, drawing from news sources both within the organisations themselves, but also from external sources like local, national and international newspapers, online news sites and other topical weblogs.”
Things get even more intertwingled when you add Atom feeds into the mix. RSS syndication already lets you pull wiki updates into feed readers and search engines. In the workplace, this lets the collaborative work and expertise of the organization be available, in native context, to many more people without drowning people in a flood of email.
Services like Feedburner and QuantumArts have tools that parse RSS feeds and recombine them by date, author, category. Because an Atom entry is defined as standalone, and the elements are more tightly defined, it should be easier to splice and recombine feeds. Diego Doval’s AtomFlow is a set of Java-based command line tools to receive, store, search and then output Atom based content (Hammersley’s description, unfortunately, Diego’s site is down at this posting.
It’s fun doing R&D experimentation open source. No service level commitments, no paperwork, lots of people to try things with. When we find things that work, we can package, and polish, improve, service and support. Participate in the ecosystem, and serve customers.

Why demonstrations?

Demonstrations were among the main reasons why I was politically agnostic in high school and college. The student bodies were fairly liberal, and there were episodic demonstrations on a series of issues: anti-nuclear, anti-apartheid, against US Central American policy, pro-clerical-workers strike, busing to Washington for pro-abortion rallies.
The demonstrators had slogans but didn’t have particularly cogent explanations. I have stronger opinions on the issues now, having learned how to research, and having more perspective to weigh conflicting evidence. Students seemed to go to demonstrations the same way they went to parties — word spread about the cool place to mingle with friends. Instead of alcohol, slogan-chanting made people feel good and lowered inhibitions.
What was worse, it wasn’t clear to me how gathering in a sports field was going to have the least bit of difference on the policy issue. Ronald Reagan and international arms control negotiators weren’t going to pay any attention to a group of young people standing on the grass and chanting slogans.
Only on apartheid divestment did a group of students — with wealthy alumni parents and allies — have a direct impact on the people making decisions.
Demonstrations still puzzle me most of the time. When the faithful rally on the issue of the day, what are they trying to achieve?
The classic civil rights demonstrations revealed that there were massive numbers of people who cared about racial equality — enough people to effect elections. The demonstrations made it impossible to ignore segregation. Massive demonstrations provided confidence and unity to people who were isolated and downtrodden.
But it’s really hard to see what most demonstrations and rallies achieve these days, other than making the demonstrators feel good. Politicians pay attention to money and voters. Demonstrators want to get on TV. But television-watchers will just see a bunch of yelling people. Does that actually persuade any of the TV-watchers to change their minds?
The attempts in Boston and New York to pen up and stifle demonstrators were disturbing. The constitution protects the right to peaceably assemble. People should be free to gather and speak.
Most demonstrations and rallies seem like ritual re-enactments of forty-year old battles, with as much political impact as a Society for Creative Anachronism gathering.
Please let me know what I’m missing.

Virtual crimewatch at Wikipedia

One of the members of the Wikipedia technical team writes about the techniques Wikipedia uses to detect vandalism and fix mistakes.

Each contributor has a personal watchlist which will inform them of changes made to articles which they have registered an interest in. On average, each article is on the watchlist of two accounts. Relatively mainstream topics usually have more watchers, obscure trivia fewer or none (a hint to those who want to try circumventing this tool…). These watchlists are often checked several times a day, most within a week or so. Since these are usually topic experts, they are likely to detect any subtle changes which the RC patrol misses.

In a physical neighborhood, people watch out for their neigbhors houses and kids. A couple of weeks ago, a burglar taking electronics from my next-door neighbor’s home was foiled by a passerby, who nodded hello to a man walking down the street on the way out for breakfast tacos. On the way back, he saw the same man, leaving my neighbor’s home with a duffel-bag full of rectangular objects. He called the cops. My neighbor got her home electronics back.
Wikipedia’s alert system turns the 300,000+ article peer encyclopedia into a warren of virtual neighborhoods. People who have a common interest in a topic keep an eye out for problems and fix them as they happen. (The article also describes more hard-core techniques to block systematic vandalism)
Alex Halevais decided to test Wikipedia’s peer editing by inserting small errors into thirteen articles. The errors were caught and fixed within hours.