Thirty years of Unix

Genetically, the way to revitalise a breed of animals, say horses, you use inbreeding to weed out lethally (or just annoyingly) recessive genes. The method, simplified, is quite straightforward (with respects to Niven, one of whose short stories I borrow this example from) - take a hundred horses of a specific breed, or mixed breeds. Separate them into twenty-five herds of Four each, two males and two females, and let them breed for as many generations as you have patience for. Colts that have lethally recessive genes reinforced, simply die or get removed by the stock breeder. The "good" traits, the ones that enforce the right to breed as dictated by lack of death and the stock breeder, simply leave the drastically reduced gene pool and leave dominant, hopefully more stable genes behind.

This drastic inbreeding, if taken for several generations, will create horses in each of the tiny corrals that are very different from any of the others, and quickly remove certain traits (and a few extra that you didn't know were there will come out) by removing lethally reinforcing recessives from the genetic pool.

Once this stage is complete (or you're starting to get a bit impatient, or further inbreeding won't effectively remove the traits you don't want), the method proceeds simply: all the horses are removed from their individual corrals and run together into a great orgiastic breeding cage together for a couple of generations. This accomplishes a number of things; further recessive genes are eliminated as breeds that do not effectively combine die out, and the key, the dominant genes are spread across the new population. In short, the result is a much more vigorous, healthy herd of horses - well bred and fit for their environment. Whether that environment is a corral, the plains of Eastern Europe, or a zoo, the principle is the same - when you take lots of small inbred groups of horses and run them together, the resulting genetic pool is far healthier and more adaptable than the group that came before. This evolutionary process (literally!) is fundamental to greatly accelerating the process of breeding stock.

Now, consider Unix. Thirty years ago, it was in one location: Bell Labs, with well-documented people creating and maintaining it; a relatively small group. It then split off into the Berkeley unix, the AT&T Unix, and a couple of others. Over the late eighties and early nineties, Unix existed in a freakish variety of flavours, all tiny and inbred and existing only for very, very specific niches - and all very tightly under non-disclosure agreements.

Idea flow between unices was very nearly zero as companies tightly held their code to protect those niches from potential competitors - in short, as inbred as it's possible for a program to be and still survive. The "genes", or the ideas behind the projects, were tested on small scale all over the computing industry; unices that were fit for their niche died off because they had particularly good ideas and could not rely on market domination to maintain their numbers; they had to have performance, flexibility, and power as a given, simply to survive in the inbred tiny niches that they were destined for. When, finally, they reached their peak of technology, the near-perfect adaptation to their niches, here comes along a double hit from the world of technology - the Internet, and the free software revolution.

The Internet was more than enough to completely change Unix forever; all of a sudden, the corrals had nice, wide highways between them - what was once seperate and disjointed suddenly became free, open avenues of cultural and idea exchange. Certainly the Internet existed before, yet the sudden geometrical change in the scale involved, and the sudden growth of the Internet culture led to a degree of interaction and idea interchange that was previously near-impossible given the lack of infrastructure.

Now, all the various fragmented breeds of Unix exist as part of a larger structure - this alone would have been enough to create a new breed of unix. One of those breeds, however, was the now-famous Linux variant, and the rapidly growing number of projects surrounding it in order to create a completely free unix variant.

In this new age of communication, and with the sudden availability of free interbreeding between the unices, this breed of unix quickly became dominant in a wide variety of niches; Linux was particularly receptive to the kind of environment that was available, and it has and is growing like wildfire, taking in the best, dominant traits of every other unix breed and becoming a source of ideas in itself.

This is the strength of Linux; its ability to take good ideas and use them, to use Unix's thirty year history to its best advantage and fit in a whole variety of niches. Like a breed of horses, Linux is far more flexible and adaptable than the first unix; it is more useful than any of the niche competitors that came before, and it's quickly spreading out in another diaspora of specialisation for the next attempt at it.

Even if Linux does fork into numerous incompatible flavours, so what? That basic interbreeding will still continue, and make Linux thrive.

Rob Bos, rbos@linux.com, live from the Linux World Conference and Expo show floor.