Okay, somebody needs to say it: Eric Raymond just said something really stupid:
That’s wrong. Open systems are better, always. Cisco has just provided us with a perfect lesson in why that sentence is completely backwards, and why we can never trust closed-source software vendors not to do evil under the cover of their code secrecy.
For those of you who have missed the news…
And then he launches into a longwinded discussion about the Cisco router firmware update which has been hashed and re-hashed. Let’s accept, for a minute, that Cisco has done something evil (which is not, for the record, a position that I hold, but let’s run with it). Let’s really run with it: suppose they use their closed source software to murder little children.
Now, how is this going to demonstrate that open systems are better always? You can’t. To demonstrate that, you’d have to
1) Provide enough examples that we can form a theoretical model which explains why open systems are always better
2) Deal with any counterexamples. For example, I write some closed source software, some people in fact buy it. Presumably they buy it on the basis that they believe it is better than open-source alternatives. How do you explain this? Are they wrong? Am I lying?
Really what ers has demonstrated is that open systems are better than closed systems in one particular case (all of this assuming, of course, that we agree on the facts, and on “better”, which we do not.)
I think what Eric means to say is that, if we have a choice between particular proprietary product A and precisely the same product A as open-source software, that in this situation the open source one is better. Who wouldn’t? Unfortunately we do not have that choice. Some software will not be written if all software must be open-source. We cannot pick between open-source Cisco and closed-source Cisco. I can state this categorically, because I am writing software now that I would not write at all if I were required to open-source it. (I also start, maintain, and contribute to many open-source projects, so I’m not completely evil. Probably.)
Somehow out of one contrived example, Eric arrives at the further conclusion that:
This is also why people who make excuses for or actively advocate closed-source OSs and network software… are not merely harmlessly misguided cultists. They are enemies of liberty
I mean, there are a lot of ontological, philosophical, and epistemological assumptions buried in here. But the core question is “what is liberty?” If liberty means “the user has the maximum amount of power”, than liberty means designing maximally effective user interfaces, which allow the user to effect that power, which has nothing to do with open source. If liberty means the author has the maximum amount of power, then we simply say the author’s choice to open source (or not) is sovereign, and a software author who decides not to open source his software is exercising a sovereign right, not an “enemy” of “liberty”.
I think what Eric means when he says “liberty” is “giving the maximum amount of power to software developers who aren’t the author.” And this is a fine, logically consistent, thing to maximize. But it is arbitrary. The subset of “people who are affected by software” who happen to be software developers is an arbitrary set, and we’ve arbitrarily removed one of them. I think that this is a lot more arbitrary than the other ways in which we use “liberty” (for example, when we talk about “liberty” in post-colonial America, we don’t mean “liberty for silversmiths who aren’t Paul Revere”). The position is a fine one, and can stand on its own. There is no need to color the discussion with unrelated words that paint an inaccurate, hard, universal dichotomy, when the reality of the situation is we’ve just drawn a line in the sand and said “everybody on this side of the line is on my team”.
But returning to the main point, Cisco sold at least one router before today, because at least one person read the updated terms of service. I fully expect that they will sell at least one router after today. As a result, we have already constructed a non-empty set of customers who prefer Cisco routers. Thus we have constructed some set of customers for which Cisco routers are better. That set does not include Eric Raymond. Did you expect it to before this story? Why is this news?