At his blog over at Free Software Magazine, Chris Holt negatively reviews Clay Shirky's discussion of open source in the Breakthrough Ideas for 2007 section in the recent issue of Harvard Business Review. Holt's main contention with Shirky is his emphasis on open source failure, such as in this key passage from the HBR text:

Yet when we look closely at the open source ecosystem, a very different picture emerges. For example, the world’s largest open source site, Sourceforge, hosts more than 100,000 projects, and its most popular software is downloaded tens of thousands of times daily. But most projects have never broken a hundred downloads, and more than half are simply inactive: A project was proposed, but nothing happened. If the vast majority of open source projects are failures, has the press been wrong to emphasize the movement’s few successes? The answer is–obviously and measurably–yes.

On first read, colored by Holt's assessment, I tended to think that maybe Shirky's text was "some kind of obscure attempt at undermining the value of FOSS." Shirky's repeated application of the term failure to open source does give the text a negative tone. For instance, he attributes success to open source because it "outfails" commercial software development.

As Shirky explains, the commercial software development process may indeed act as an inhibitor to innovation because it greatly rewards success and discourages failure. With open source,

The institutional barrier between thought and action – the need to convince someone that your idea is worth giving a whirl – doesn’t exist. The low cost of trying means that participants can fail like crazy as they continue to build on their successes.

After a second read, I found Shirky's analysis credible and a contribution to understanding why open source succeeds, just not pitched well to FOSS advocates who are looking for celebratory rhetoric of open source. But it is also incomplete. Shirky focuses only on the initial individual's creation of ideas and seems to have forgotten the benefit of the open source license which privileges future innovation. Open source succeeds not only because the cost of failure to experiment and innovate is low, but also because the return is higher on both success and failure. As Holt points out, the excellent, sort-of-good, and half-baked ideas are all shared with the hopes that any idea could stimulate something better because anyone else can build on those ideas. If only 1 out of 1,000 projects on SourceForge is a winner (and I'd bet the ratio is better than that), open source as a principle of knowledge production is a winner, too, because others can learn from bad examples, develop ideas that the original creator doesn't have time to, or mutate the half-baked ones into something better.

Consider this additional analysis in comparison to other intellectual property development. Suppose we evaluated the publishing industry's success over the last ten years from a similar perspective of understanding the influence of failure. We'd not only have to look at those books that have made the bestseller lists, but of course those which had marginal sales and those which didn't sell so well and ended up in the bargain book warehouse racks with the covers torn half off. Now if books and book ideas were open sourced, we'd also have to count the thousands of sample chapters pitched to publishers, and I'm sure just as many full manuscripts that went straight into the trash after the publisher read the first page or two. Then let's not forget the good manuscripts which, for whatever reason, are not accepted by a publisher. Maybe they are read. Maybe they are never looked at.

Success is often built on failure, and a system which encourages that to happen has merit. So which is the real failure? An intellectual property paradigm which only preserves what a commercial entity feels like they can develop or sell and discards all other--good and bad--where the effort and ideas of human production are often lost when judged unmarketable? I'd go with the one that conserves all ideas as potentially useful in efforts to increase human knowledge and also privileges collaboration in building knowledge over individual ownership and control. Seems like a no-brainer.