Sunday, February 08, 1998 at 9:20:40 PM Pacific

Eric Raymond Responds

Last week I wrote two pieces in response to Eric Raymond's Cathedral and Bazaar essay. This evening I got a response to the first piece, I pointed him to the second piece, which I feel is more constructive. Here is his response to the first one. DW

Eric Raymond Responds

Some friends directed me to your comments on "The Cathedral and the Bazaar", which are about the only public critique of it anybody has done yet. I thought I'd try to answer some of the questions you raised.

You question whether "commercial-quality apps" can be built around Linux or other GPLed software. To which I answer: Apache. GIMP. Postgres. And have you seen the KDE stuff lately? Astonishing, even to me.

The bazaar method *already* supports a community of millions of users (I'm one of them), and a software industry that is looking at a lot better growth prospects than the battered and hurting Macintosh world. You guys lost your war when Steve Jobs bent over and spread 'em for His Gatesness last year -- we, as witness the Netscape announcement, are beginning to win ours.

But your most basic misconception is that the bazaar method is somehow incompatible with paying great programmers a lot of money. This is not really your fault; the term "free software" has messed with a lot of peoples' heads, which is why many of us in the Linux community are moving to drop it in favor of "open-source software".

In fact, I have identified three major business models in which going the open-source route is a net economic gain for investors. Netscape is working one of them now, what I call the "loss-leader" theory. Another is "give away source, sell service", which seems to be working quite nicely for Red Hat and Caldera (among others). The third is for hardware companies that can run bazaars to support their interface software; I'm discussing that one with some major Silicon Valley players in the peripherals industry right now.

The "give away source, sell service" approach, in particular, is looking better and better today because the rising cost of programmer time makes it hard to stay competitive using only the number of programmers J. Random Corporation can afford to hire. By managing a bazaar-mode development, JRCorp gets to trade various non-monetary goods with outside programmers who are willing. JRCo still needs its core of paid programmers to do integration, testing, and design oversight, though; there are good jobs here. Without net gain for all parties this form of exchange would collapse -- instead it's thriving. Does this suggest anything to you?

You have an oddly narrow view of what `customers' are, as if you think money is the only tradable good. Hasn't it occurred to you that time and attention are also goods? But even making your assumption, you ignore the end users who pay for commercially-branded Linuxes like Red Hat.

I'm not surprised you "don't believe" in the bazaar model, because your understanding of the actual economics and social dynamics involved is (no offense meant) crude and deficient. This is OK; most of those involved don't consciously understand it themselves. They just do it and win.

I don't have an "emotional relationship" with the people who wrote my TCP/IP stack or X server. A generalized feeling of community, maybe, but nothing that would substitute for actual performance on the job. And as for the implication that my attitude to Linus Torvalds is "religious", he'd get a good long laugh out of that.

If the open-source culture only scaled to the population size of a preindustrial village, or the (somewhat larger) size of the average charismatic cult, you might have a strong point. But the most exciting lesson of the Linux world is that the model scales well to an Internet-linked population orders of magnitude larger, in which personal bonds *cannot* be a significant factor and the typical relationship is the "cool" and professional one you like. There is nearly a gigabyte of high-quality software written by thousands of individual authors on my Linux system. D'you suppose I'm intimate with all of them? :-)

More generally, I suggest you need to rethink some of the scarcity-economics assumptions you're carrying around. Corner an anthropologist sometime and ask him to explain what a "gift culture" is. Economies in which most players have plenty reward different kinds of behavior than those in which material want is the rule. People still compete, but on a different level and in ways that seem bizarre to an exchange-economy mindset. Does the word "potlach" mean anything to you?

Actually, the open-source community is a hybrid -- gift economy at the center, exchange economy at the edges. And no, I didn't believe it could work until I saw it in action, either.

Eric S. Raymond


Raymond replies to Linux on Mail Starting 2/8/98.

This page was last built on 2/9/98; 9:35:34 AM by Dave Winer. dave@scripting.com