When last heard from, heavyweight commenter Nick Carr was flattening IT — dismissing it as no more important than facilities and treasury — interesting, even important corporate functions that provide no competitive advantage. His book, Does IT Matter?, prescribed “IT as utility,” a back-office function that should be managed as a cost center with a mandate to run at the lowest possible cost.
In Carr’s unique world perspective, competitive advantage is defined as something that provides a permanent lead over other market players. And, because IT can’t guarantee permanence, it should be relegated to the nether regions of G&A. In his book, he cites a pharmaceutical company that created a drug-ordering application and cleaned up in the pharmacy business because its systems were so much easier to use than the manual alternatives from other providers. After 10 or so years, however, the rise of the Internet made its clunky client/server app obsolete, so … IT is irrelevant.
This is a classic case of creating a straw man by defining a situation that in fact doesn’t exist, and then destroying it with arguments that point out its shortcomings. Carr put the straw man of “IT can provide permanent competitive advantage” into the ring and then delivered a quick right hook that left it unconscious on the mat. There was only one problem with this match — his knockout was a fixed fight.
There is no such thing as permanent competitive advantage, not in IT, marketing, manufacturing, not anywhere. And I can guarantee that most companies would kill for a 10 year competitive advantage, even knowing that it would not be permanent.
Carr has now brought a new straw man into the ring. This fight is with open source, and his argument is that open source isn’t really that open — that the supposed advantage of thousand of distributed contributors actually is a handicap, and that the real truth is that open source only works if a few real innovators hatch the product while the rest of the hangers-on provide some minimal testing capabilities.
In Carr’s view, real invention is the realm of a few super-creative types with the rest of the world relegated to awestruck admiration. Consequently, the “bazaar” notion of open source posited by Eric Raymond in The Cathedral and the Bazaar is naive or even pernicious, being as it lures people into thinking
that there’s a way for lots of people to contribute to open source projects, when in fact they should realize that innovation can only be performed by the few.
Carr cites Raymond himself that development can, in fact, only be done by small teams. Carr even smears open source by comparing it to Wikipedia, tut-tutting about the user-built reference work and judging as puerile because it devotes twice as much coverage to the Flintstones as to Homer.
The implication, of course, is clear. Software development, just like reference works, is the province of the elite and should be left to experts.
Carr is wrong. Worse, he is engaging in sophistry.
The notion that open source is dissembling or foolish when it posits that it provides a way for large numbers of people to contribute fundamentally misunderstands the organization and processes that go on in the open source world — so much so that one can only assume that Carr willfully created a superficial portrait of open source for the express purpose of flattening it in a sham fight.
Carr fails to distinguish the key difference between a small group controlling code contributions and a small group contributing code.
The genius of open source (and, I would argue, the burgeoning world of Web 2.0’s user-created content) is that it offers a way for people to contribute who were previously shut out from contributing their knowledge and expertise.
The old world of software development (which, by the way, I spent plenty of time working in) was limited to the few. There is a difference, of course, between the few and the elite, and most of the software development organizations I’ve seen were not elite — far from it.
Open source provides a process and an economic model for the many to participate. Is everyone capable of contributing? Of course not. Does that mean that the right model is the old one of isolated developers? Definitely not — we’ve seen the shortcomings of that model, not only in high prices, but, even worse, in unsatisfactory products created just because there was no involvement with the outside world.
There has to be an organization and process in place for large numbers of people to contribute, naturally enough. The method to accomplish that in the software world is bounded interfaces, which removes dependencies
and enables groups of developers to cooperatively engineer without imposing overwhelming coordination costs.
To see the distributed, decentralized approach of open source working successfully one has no further to look than Linux. There are hundreds, if not thousands, of developers working on Linux, ranging from corporate-sponsored individuals all the way to hobbyists. If they’ve got the chops, their code can make its way into the product. Of course their code gets vetted. Of course Linux is not built by the random submissions of anonymous developers. But it’s a long way from recognizing that Linux is not just a melange of code to concluding that the only way software can be written is by small groups of cognescenti.
And, by the way, that argument about how the elites do so much better at discerning the really important things in life, like the Flintstones vs. Homer? Having seen the commercial media’s coverage of Paris Hilton’s jail woes and comparing it to its coverage of the dismissal of the head of the Joint Chiefs of Staff, I’d say a two to one ratio of superficial to profound looks pretty good to me.