An Interview with Tom Lord of Arch (O'ReillyNet)

Posted Sep 27, 2004 12:43 UTC (Mon) by hppnq (guest, #14462) [Link]

The interesting thing, of course, is that Larry's problem is usually solved by patenting the invention. If the patent system hadn't been so horribly abused, this could have been a good, pragmatic -- but not perfect -- solution to the problem.

So let's consider the chances of someone building a "better BitKeeper" with patents out of the way.

First, the scenario in which BitKeeper is released under the GPL. The way I see it, all Open Source developers who merely care about a SCM system would probably contribute to it, rather than taking the source and cloning it, or using vital parts of it in a new system. Really, most people have got better things to do than to write another major piece of software. Apart from that, it is not that simple to rip off an Open Source project. Why doesn't Hans Reiser fork the Linux kernel and be done with the constant hassle over ReiserFS? Because it just would not work. The Linux kernel has plenty of momentum, and that is without any corporate backing. The pure power of Linux is *exactly* the fact that the development process involves many, many people contributing to it (partly because of BitKeeper ;-). Given the current state of affairs in the SCM world, BitKeeper would not have a lot of trouble attracting committed, enthusiastic community of developers and users.

That leaves the people who have got the intention and the means to grab BitKeeper, slap the name BitSneaker on it and sell it as if it were their own. Quite possibly the intention is there, but are the means too? Because BitKeeper is released under the GPL in this scenario, BitSneaker would have to be Open Source too -- so at least the competition would be fair in that respect. But as long as BitKeeper is backed by enough Open Source developers, the BitSneaker project is doomed: BitKeeper will continue to become better and better and therefore alone more interesting to customers.

The second scenario is, of course, the one where BitKeeper is released under a closed license. Ignore the fact that people can use it free of charge -- Bitmover has made sure that anyone with an interest in SCM systems can never contribute anything to BitKeeper. Now, all the burden of developing it is on Bitmover's shoulders. There are plenty of developers with a need for a good SCM system, but they cannot use BitKeeper. Inevitably, this problem will be solved by those people, be it by contributing to arch, cvs or starting from scratch. It may take a couple of years, but they will get there. At that point, BitKeeper has an Open Source competitor for sure. People were laughting at OpenOffice two or three years ago, and now we see Microsoft releasing the code to Office to selected parties.

Apart from the Open Source developers, who just want a working SCM system, we still have those who might have the intention of making money off of BitKeeper technology. The fact that the license says "Thou shalt not poke" will not stop them from trying to see how BitKeeper ticks, and it will probably not hold in court anyway (IANAL ;-) if there is no evidence of line-by-line copying and something like a patent that protects BitKeeper's internals. It will, of course, be more difficult to extract the interesting bits from BitKeeper, but it will not be impossible. Even worse, competitors might get away with an inferior product that looks like it does the same thing (think VHS versus Betamax) because no one will be able to look at the internals of both programs.

That's a long story, and it's not even complete. ;-) Of course it is biased, but in a fair way, I would hope -- the bias is away from common notions, like the one that Open Source is easy to rip off and closed source is not.