New: Book Report: The Process of Creating Life

The Process of Creating Life is the second book of Christopher Alexander's Nature of Order tetralogy. That is, this is a book that is Alexander's theory of the universe and how this nature should guide us in making buildings.

Summary: The first half of the book tells you that an incremental plan/build cycle is good. This is good advice, but it gets lost in the mumbo-jumbo. With this good advice in mind, skip ahead to section 10, part 9. Now start reading--the rest of the book is still about 50% mumbo-jumbo, but the rest is good--there are case studies, and you can learn from them. If you read nothing else in this book, read the appendix. It is one long case study, and has some good stories.

Enough with the summary, already.

This book feels like an update on his The Timeless Way of Building in that the message is: Don't try to plan everything out ahead of time. Instead, alternate between building and planning stages. Maybe this means that you won't know what your second floor will look like as you work on the first floor--but this uncertainty is worth it since it allows you to make decisions while standing on the site of the second floor.

He points out that this is not the normal way of doing things. Most architects try to plan out everything in advance. The results are often ill-fitted to the building site and don't work together well.

OK, so that's pretty much the same message as The Timeless Way of Building. What is new?

The first part of the book combines this incremental approach with the life-giving structural properties of The Phenomenon of Life. Alexander creates a napkin doodle by applying these features incrementally, always preserving symmetry. This is very pretty. It seems doubtful that one could apply this process towards making a building/town/whatever. Or, rather, I can not see how one could use this process to steer a building plan towards something good versus towards something bad.

Oh, wait, before that he talked about how growth works in biological systems. This section is also very pretty. I think the lesson we're supposed to take away from this is that things don't grow as if they were being squirted into an injection mold. They grow--they grow larger, and they grow more complicated. They grow at different rates in different parts. This is very pretty. It seems doubtful that one could apply this process towards making a building/town/whatever.

Then there is a lot of complaining about designs by other architects. Of course, Alexander can not just say, "I didn't like it." Instead, these structures could not have been created by a sequence of Wholeness-preserving application of the Properties. Except that some of them could have been. How can we explain these away? Well, these could have been created by iterated transformations based on the properties--but only if those transformations happened in the wrong order. And who can say what this wrong order is? Christopher Alexander can, of course. But he can not explain it.

He tried to explain it to a student. He told the student to first choose the most important part of the design, to start with that and get it just right. And not to start working on the next part of the design until the first part was right. So that determines the order, right? Maybe. Alexander describes stepping through this process with the student. Alexander asked the student which was the most important part, and the student answered, and Alexander said, OK, work on that part. Then Alexander asked, what is the next most important part of the design? And the student answered, but Alexander did not agree with this answer. And so Alexander said "Really? Is that really the most important part?" And he was eventually able to bully the student into backing down. The student rattled off some other guesses until he found one that pleased Alexander. And thus the process continued.

Alexander does not phrase it that way, of course.

I do not doubt that the resulting design was good, but obviously it's not so easy to determine what the right order of transformations should be.

There are some photos and descriptions of good buildings and places. Alexander points out some modern ones. I guess if I read more architecture theory books, I'd know that Alexander's critics accuse him of being a Luddite or something. Fortunately, I have no plans to read much more architecture theory.

That's the gist of the first half of the book. It had a pretty low signal-to-noise ratio.

Fortunately, it gets better. In the second half of the book, there are some good case studies. One thing I learned from them: if Alexander uses those 15 structural properties that he laid out in The Phenomenon of Life, he does so only unconsciously. Which suggests that one should not read The Phenomenon of Life.

There is a case study of the Julian Street Inn in San Jose. Here we see a series of sketches in which the plan takes shape. You can see the centers form, and Alexander talks about how he made decisions. It's by feel: he creates "strong centers". He creates a place with a "living feel". That is, he does not describe his process in a way that helps the reader to make these decisions. But the sketches help a lot--showing which parts of the plan took shape in which order.

There are some summaries of pattern languages for different places. Though these summaries mostly only provide lists of pattern names, it is interesting to see what variations of pattern language are useful in different places and cultures.

Chapter Fifteen "Emergence of Formal Geometry" has more site sketches and pretty pictures. Here, we are not taken through so many steps of the initial planning, but there are some. Again, the fifteen properties do not appear. Actually, he does rattle off some of them when he talks about creating a grid, but in a way that suggests that one can use those properties to justify anything.

A scary story for the computer geeks in the audience:

When I described the Pasadena apartmnet-building sequence and other examples to a prominent computer scientist in Silicon valley he said to me, with an astonished look on his face: "You mean the generator problem, for architecture is solvable?"

I told him that I did mean that, and that my colleagues and I were on the way to solving it for a large number of particular cases, and believed it to be solvable in general.

Can we claim that a problem is solved if we still believe it is AI-complete? Bah. Did Alexander ask what was meant by the generator problem?

He presents a list of elements of society in which changes must be made:

Distressingly, he makes a good case for these. Our laws and customs of land use are pretty screwed up. To make them good would require some pretty fundamental changes to our society.

The best part of the book is the Appendix "A small example of a living process", a case study of planning/building a house in the Berkeley hills. Here we see way that the plans changed as the building was built. We see how to trick the Berkeley building inspectors. We see the advantages and frustrations of the incremental building/planning approach. And there are plenty of pretty photos. Check it out.

Tags:  ;  ;

Labels: , ,

Posted 2005-03-05