Analyst: Open source growing up or selling out? (Comino et al.)

I think this will be the last of the academic papers for awhile. It has been great to step back from the open source community, and look at from the onlooker's perspective, rather than as a participant. Some of the myths we take for granted are clearly dispelled when you look at open source through the (somewhat) objective lens of academia/science.

None of its benefits disappear, really. Just some of the myths.

For instance, Comino, Manenti, and Parisi wrote a great paper called "From Planning to Mature: on the Determinants of Open Source Take-Off." In it they analyze the evolving nature of open source development. They find, among other things,

...[T]he less restrictive the licensing terms the larger the likelihood of [a project] reaching an advanced development status...[which] effect is even stronger for newer projects. [They] also find [no big surprise here] that projects geared towards system administrators appear to be the more successful ones. (Abstract)

This is somewhat interesting in light of speculation by many that more restrictive licenses engender bigger development populations, because developers want maximum code freedom (as in RMS) - i.e., they want to make sure they're code will not be taken by commercial interests without their permission.

However, I suspect that the trend toward more successful (i.e., Downloaded and used) and less restrictive projects has more to do with commercialization than anything else. (The authors hint at this, as well.) The more Alfrescos, SugarCRMs, etc. that come on the market, with more need to control the direction of their respective code bases, the greater the tendency will be toward less restrictive (in the open source sense) licenses (that accord less "RMS freedom" but more "project owner freedom/flexibility"). This isn't an organic trend - it's fed by these companies' marketing dollars.

The authors also find, as I mentioned from a separate paper, that "around 80% of the [SourceForge projects as of December 2004 do] not show any interaction within the community of developers, having recorded no bugs, patches nor feature requests since [the projects'] registration" (6). In other words, SourceForge is long on project starts, but short on project development. Young and restful...as shown below.



In addition, projects with at most two developers account for more than 80% of the SourceForge projects. (8)



This isn't a bad thing, per se, as it could simply mean that open source happens on a much smaller scale than originally mythologized (as I've blogged before). Or it could mean that open source follows the rule that the rest of the world follows: the 80/20 rule. (I.e., 80% of the work is done by 20% of the contributors.) The authors suggest as much. (12) So, small development "communities" are not necessarily a bad thing. But the phenomenon does call into question the size of the open source development community that we've generally assumed.

Again, I don't think this is a bad thing. Why? Because I care less about development than I do about use. I care how many people/companies derive value from a given project. Open source is only important (to me) to the extent that it's useful. And I measure "useful" by end-user uptake, not developer uptake (though, if you can find both, that's obviously a huge bonus). As for where it's being used (i.e., for whom it's being developed), we'll end with this data from Comino et al:



Mostly developers writing for other developers (including system administrators). Not much end-user code in there...though commercialization will change that. Thankfully.