A 2006 Christopher Alexander interview:
 excerpts relevant to computing

From "Early and Unpublished writings of Christoper Alexander" by Howard Davis
p. 207 "Conversation with Rem Koolhas"
beautiful.software

... there are six, seven billion people now on Earth. It's quite impossible to create a humane environment in which all those people can live well, unless they (not we) are controlling their environment. Only then will people be comfortable, and at one with the land and towns around them. But to do that, one needs an extraordinary and sophisticated tool, some kind of wonderful generative process that can interact with exquisite sensitivity to people, and their desires. The current machinery of architectural production and developers' production is far too crude and far too cumbersome to do it right. In most cases it does more harm than good.

This is where the use of generative processes comes in. By generative processes I mean new kinds of processes that include pattern languages, but that also include other vital elements -- above all, an infrastructure of power-sharing among people, so that people get some measure of control over the processes that produce buildings and the public environment in cities. It also needs a revolution in the construction industry, so that construction itself is directly harnessed to the production of organic and highly-adapted organic structures which allow the variation needed for things to fit together in their unique circumstances, and allows builders to make all the needed adaptations without using or needing blueprints in their 20th-century form -- all without increasing cost.

So, generative systems are capable of doing for the environment what DNA does for an embryo or a plant. I think the Pattern Language was an interesting first attempt to make such a thing, but I don't think it's good enough to do the job. First of all, there's the whole thing about the technics of this. What might a developed form of this look like, how might it be propagated, how would the construction arise. Those are all important questions. The last few decades have been dominated by the idea of DNA. Biologists are now beginning to realize that the shape of plants actually does not come from DNA. This is really remarkable, the DNA guides the process but the actual shape comes from the literal unfolding of the geometrical object that is the growing plant. So, these morphogenetic ideas are much more powerful than what was in the Pattern Language. I'm trying to formalize them now ...

... in a computer, in part.

Just to give you an idea of what I mean by generative system: you see one of the most interesting building projects that I ever did, a very primitive one in Mexico [the Mexicali project] ...

I achieved a unique building system which I have not yet ever been able to replicate again in other contexts. The system was unique because it enabled buildings to unfold step by step, thus allowing each individual house to be different according to its family, and yet cheap, simple, and ordered in its process, so that money was not wasted, and the steps could be simply and directly carried out, for all these different houses, without the use of drawings.

That system consisted of first of all, some special corner blocks. The procedure was really simple. You drive a steel stake to form corners of rooms and houses. You have freedom -- you can put them where you like as long as they fit some very loose conditions. Then there's a special block that we manufactured, that was put down over that bar. That block establishes the corners of the rooms and the buildings in a more definite way. Then other blocks fit into a slot in each face of the corner blocks. The blocks are all cuttable, so you don't have to be restricted by the dimension of blocks. Then the walls go up. You lay in the windows as the walls are going up. Then you put a ring beam around the walls. The ring-beam form was very special, having two pieces of wood, which were kept apart. So then I had the families weave baskets, over each room, using lathing strips -- long strips of wood with a cross-section of about 30 mm by 5 mm so they are quite flexible and very easy to bend. These strips made a basket with a weave about one foot square, and the strips were trapped by the ring beam. Then, when the concrete was poured into the ring beam, it was firmly fixed. One could crawl around on these things. And finally a concrete shell was poured over the basket. Of course each one took its own shape, because if the room had a funny angle here, then the vault that was formed by it took on a different shape. So in the end, the process alone actually was responsible for the form of the house.

...

... in a computer, of course you can set a number of parameters and churn out endless combinations and variations, but if they don't have meaning they are really just trivial games. For the people that live in a world that is created like that, it is actually frightening. It's not joyful, because it isn't coming from anyting actual. You can read the insincerity of it. It's trying to fake the variation between one tulip and another, but it's the wrong kind of variation.

Morphogenesis has to do with drawing the real variation from the situation. That's a much more difficult trick from a computer point of view. It's not playing games with computers, it's a question of having representations of the configurations which are profound enough so that when you draw on them you'll get authentic building variation, room variation, roof variation, etc., for really good reasons which makes sense, and which are related to the world around them, and the whole to which they belong. And then human beings will feel more at ease and relaxed, because that's what we're all used to in the world. It is this kind of deep variation which is generated by the whole that gives us satisfaction, and meaning.