The basic error of this article is that it ignores the prisoner's dilemma problem. The basic idea behind that problem, as a reminder, is that parties acting in their best interest for themselves will, in certain situations, select a suboptimal outcome.

Applied to open/closed source software, the prisoner's dilemma suggests that parties acting in their own interest will opt for closed-source software, because of the personal benefits (trade secrets, market power, etc.). Thus, even if open-source software is more efficient as a general matter, everyone will go for closed source, resulting in an inefficient outcome.

The solution to the prisoner's dilemma, of course, is contracts. By binding the hands of the parties, you can force people to choose the more optimal solution. The ideal contract adds a breach penalty that offsets whatever selfish gain may be reaped from defecting, and thereby encourages cooperation.

This is, in effect, what the GPL does: it adds a penalty to closed source software (namely, you can't use open-source with it). If open source is indeed more efficient, then developers will be willing to forego the benefits of closed source in order to take advantage of open source software.

I think that this generally summarizes what everyone else is saying ("nothing would deter an unscrupulous company from stealing open source code," "economic principle of free riders"). My own comment is offered as a general thought, not a precise cost-benefit analysis. There is much more careful thinking on this subject, much of it in the law review articles I read while in law school.

This article is written in a confusing manner but his point is basically that the GPL forces people who write software that derives from GPL licensed code to open source their derivative software. By doing so it discourages people from using GPL licensed software because some of those developers might want to close their source code.

I see the point but I’m not sure I agree with it. He claims the open source community “loses allies” when it enforces the GPL against companies who closed their source. But if those people are taking GPL licensed code and violating the terms than they are not allies at all. So his argument really boils down to having the GPL but not enforcing it so you can keep allies that are allies in name only.

There are more permissive open source licenses out there and if a developer wants someone to be able to use their open source code in closed source projects they can use one of those. But when people choose the GPL they are doing so to further the ideal of completely open source code. Meaning allies like the ones described above are counter-productive because they would compromise that ideal by doing exactly what the GPL users are fighting against.

we stand to lose not only Cisco as an ally, but any corporation that estimates (rightly or wrongly) that their own potential exposure to an SFLC lawsuit might be greater than their potential efficiency gains from open-sourcing.

Why should that matter if we live in universe B and their estimates are therefore wrong? The open source community, being more efficient, would just build an alternative product that works better.

OTOH, if we did not have the GPL or refused to enforce it, nothing would deter an unscrupulous company from stealing open source code and not contributing their own improvements to it to the community. Any improvements to the software from the open source community would quickly be incorporated into the proprietary version, while any improvements made by the hypothetical unscrupulous company would remain proprietary. Once a few of these companies started enjoying success, every other profit-maximizing company would be forced by the market to follow suit or perish, and we'd eventually have a clusterfuck of companies stealing open source code while the open source community dwindled to nothing because their products could never possibly compete with the proprietary alternatives.

I think we live in universe C, where, without an enforceable GPL, the most efficient method of software production would be to take what you can (from the open source community), and give nothing back. YARR!

Sorry, not a very good article, the arguments are simplistic and don't make much sense to me.

There is a good discussion to be had about the GPL but this is not it.

I find it somewhat amusing that none of the comments so far seem to acknowledge that this is written by Eric S Raymond. I guess that shows how far he's fallen into irrelevance.

Without wanting too engage with the actual content too much, he appears to be arguing, despite the headline, that there is no economic benefit or cost to the the GPL. The only impact is based on human (actually corporate) psychology of not liking being sued.

I think it's a well-written article, but he fails to acknowledge the economic principle of free-riders. I think it destroys his argument.

http://en.wikipedia.org/wiki/Free_rider_problem

The author fails to take into account that many(?) open source developers aren't dependent on or motivated by market forces. Regardless of its virtues in a business setting, GPL will still live on.