Monday, 29 August 2011


A newspaper article reminds us that it is 20 years since Linus Torvalds set about writing the Linux kernel:

As a big fan of Linux, who started with Ubuntu and has now switched to Mint (both of which are at least as user-friendly as Windows and the Mac OS, as well as being free), I feel inclined to celebrate. But there are also some questions to ponder.

How exactly does the evolutionary environment for Linux distributions (distros) work? Anyone can create a new distro, and plenty do. Many of these are direct developments of existing distributions, for example, Mint from Ubuntu. If an existing distro lacks something that enough people consider important, such as lightness, or changes course in a way that many do not like, as with Ubuntu's switch to Unity, a new distro will arise. If it is good enough, it will attract fans away from existing distros.

This may look like simple Darwinism. New variations arise, and the ones that are well-adapted to the environment (the demands of users), grow fat. But there is a feature that we do not find in nature. The creation of new variations is not random. Someone takes a conscious decision that there is a niche for a new distro, and further conscious decisions as to how best to fit into that niche.

It would be an interesting exercise to model that process as random variation, making the conscious thought drop out of the picture. One might identify the population of distro ideas in the brains of all Linux developers with the existing members of a species, then assume that members of the population bred new members (new ideas) at random, then assume that some new ideas were crushed straightaway by the immediate environment (the individual developer's brain identifying an idea as silly), but that others got out for discussion in the community, meme-like, and a few led to actual new distros.

Thus one would have to change the focus from distros to ideas, in order to model the process as one without conscious direction. That would also change one's identification of the payoffs for good adaptation to the environment. With distros, the payoff for the distro itself is reproduction on lots of computers. The payoffs for the human beings who take the conscious decisions are kudos and, sometimes, the opportunity to sell support services. (A distro itself can be sold, on CD, but it must also be available free because of the terms of the Linux licence.) With ideas, the payoff is simply reproduction, in the heads of all others who agree that the idea is a good one.

We can also ask about the effects on the software industry. Those who make money out of software may complain about free alternatives. Witness Steve Ballmer's infamous "Linux is a cancer that attaches itself in an intellectual property sense to everything it touches". A (secondary) source is here:

That remark was, however, made ten years ago, and I think the complainers now know that they have lost the battle. Anyone who still complains should be referred to Frédéric Bastiat's wonderful Pétition des fabricants de chandelles, bougies, ..., in which the candlemakers argued that curtains should be kept closed during the daytime in order to defeat the unfair competition of the Sun. The text is here: