Biomorphs from the Blind Watchmaker program

So much for the genetics. The game starts to get interesting when we consider the 'embryology'. The embryology of a biomorph on the screen is the process by which its 'genes' - those numerical values -influence its shape. Many very different embryologies can be imagined, and I have tried out quite a few of them. My first program, called 'Blind Watchmaker', uses a tree-growing embryology. A main 'trunk' sprouts two 'branches', then each branch sprouts two branches of its own, and so on. The number of branches, and their angles and lengths, are all under genetic control, determined by the numerical values of the genes. An important feature of the branching tree embryology is that it is recursive. I won't expound that idea here, but it means that a single mutation typically has an effect all over the tree, rather than just in one corner of it.

Although the Blind Watchmaker program starts off with a simple branching tree, it rapidly wanders off into a wonderland of evolved forms, many with a strange beauty, and some - depending on the intentions of the human player - coming to resemble familiar creatures such as insects, spiders or starfish. On the left is a 'safari park' of creatures that just one player of the game (me) found in the byways and backwaters of this strange computer wonderland. In a later version of the program, I expanded the embryology to allow for genes controlling the colour and shape of the 'branches' of the tree.

A more elaborate program, called 'Arthromorphs', which I wrote jointly with Ted Kaehler, then working for the Apple Computer Company, embodies an 'embryology' with some interesting biological features specifically geared to breeding 'insects', 'spiders', 'centipedes' and other creatures resembling arthropods. I have explained the arthromorphs in detail, along with the biomorphs, 'conchomorphs' (computer molluscs) and other programs in this vein, in Climbing Mount Improbable.

As it happens, the mathematics of shell embryology are well understood, so artificial selection using my 'conchomorph' program is capable of generating extremely lifelike forms (see over). I shall refer back to these programs, to make a completely different point, in the final chapter. Here I have introduced them for the purpose of illustrating the power of artificial selection, even in an extremely over-simplified computer environment. In the real world of agriculture and horticulture, the world of the pigeon fancier or dog breeder, artificial selection can achieve so much more. Biomorphs, arthromorphs and conchomorphs just illustrate the principle, in something like the same way that artificial selection itself is going to illustrate the principle behind natural selection - in the next chapter.

0 0

Post a comment