COMMENTARY--According to Eric Raymond, every good work of software starts by scratching a developer's personal itch. But is it also the developers' interests that get served?
It's just about 4 years now since Netscape, in the pre-AOL days, released their source code to the new Mozilla project and cast their lot with the open source world.
At the time there was a lot of talk about how Netscape officials had been influenced by Eric Raymond's essay The Cathedral and the Bazaar, which argues the merits of open source software development against that of isolated companies, for obvious example Microsoft.
Raymond has been getting a lot of ink again lately for reasons which mostly escape me. In his interview he argues the inevitability of Linux popularity on the desktop, a courageous position for the complete lack of evidence in support of it. But perhaps we should be re-examining Raymond (or "ESR" as he is known) and The Cathedral and the Bazaar again in light of the last four years and what they say about open source development.
ESR argues for the romantic notions behind the open source movement and claims all things good come out of them. In fact, over the last few years Linux development has been led by people who have commercial affiliations with companies that (attempt to) make money on Linux. They aren't necessarily paid directly to work on Linux, but some of them get what amounts to grant money from companies like Red Hat. Now mind you, this doesn't bother me at all. In fact, it amounts to a semi-conscious pooling of R&D costs by companies in the Linux business, which is a good idea for them.
Alex Plant of SWSoft, another company involved in the current development of Linux, says, "While these companies do contribute monies and these developers are indeed on payrolls, there is a clear line of delineation between commercial and open source interests." But I have a hard time believing this completely.
IBM is a little more explicit. "IBM is one of many companies that have developers who are working to contribute code to the Linux kernel. Other companies include Red Hat, SuSE, Conectiva, HP, Intel, NEC, Fujitsu, and many others."
The part of The Cathedral and The Bazaar that stuck with me most was: "Every good work of software starts by scratching a developer's personal itch." This makes a lot of sense at one level. If you want something done, do it yourself. The problem with this concept is that only developers' itches get scratched. If a regular user wants something in their program they can't just decide to write the code. The end result is that it's programmers' interests that get served.
The evidence for this is splattered all over the history of open source projects. The only serious attempts at end user-oriented open source programs are Mozilla and maybe OpenOffice, and both exist only because of large subsidies by commercial software companies that released large existing code bases to the projects. True, the Netscape code base was so awful that the Mozilla folks chose to throw it away and start all over, but AOL/Netscape did provide many programmers on their own payroll to work on the project. I have no doubt that OpenOffice faces similar prospects; without the active support of Sun I don't expect it to go anywhere soon.
ESR makes many other arguments which make varying degrees of sense. "Given enough eyeballs, all bugs are shallow," the argument that the large number of programmers in a large open source project makes it more likely that bugs will be found is somewhat appealing, but only if people are looking for bugs. It's my experience that looking for other people's bugs is not the sexiest work for a programmer, and even if some perfect stranger from some suspicious country is doing free work for your project, should you trust them? In a well-organized project like Mozilla, one with commercial backing, perhaps bugs get better treatment and supervision is better.
But Mozilla is hardly a flattering example for open source. As good as their excuses may be, it's been 4 years and they're still not at version 1.0. For the same "enough eyeball" reasons, open source development should produce a finished product in at least a reasonable amount of time, if not quicker than the cathedral. History has not borne this out.
In fact, those cathedrals have a lot to teach these bazaars. I find it reassuring for the future of Linux that there are large, well-funded commercial outfits with an interest in it. Whatever problems Linux has, they are also the problem of these companies. For instance, I have my doubts about the future of pure Linux companies like Red Hat, but as bad as things would ever get for them, companies like IBM are into Linux too deep to let it languish.