How Buildings Learn, by Stewart Brand

I finally read this book after stumbling across it, oversized, on a library shelf. I had it on my reading list, that I keep updated on my Palm Pilot, for a while. Twice, in fact, as it had been recommend from a couple different sources, and when I went to browse the list again, I realized I was removing it from the list for the second time. It was a very good book, achieving what so few nonfiction books do - explaining a subject so well that almost all of the lessons carried over to dozens of examples more of interest to me personally. The subject at hand was the evolution of buildings over time, a subject that had no specific interest to me, but since the book came so highly recommended, from so many sources, when I saw it I checked it out.

Essentially there are no permanent solutions for buildings that are worth having - If it's doesn't adapt, it's obsolete before it's complete. Given this, there are two opposite models that Brand proposes are worthwhile; temporary buildings, and permanent ones. Seriously, the two working paradigms are buildings built over generations, and building so obviously deficient that they are treated as permanent projects. The first category includes buildings like feudal manors, where each addition is the permanent addition of an individual to a family tradition. The second are buildings that was designed to be temporary, such as building 20 at MIT built during WWII as a temporary building that is still standing, in which the occupants feel free to make "improvements," such as nailing up shelves wherever they need them, or inserting extra staircases, etc.

Another theme in the book was that short term planning for an object that is intended for the long term is, well, stupid. And recommended against. More money is pent changing existing buildings than building new ones. I am pretty sure the same applies to computer code,
and many other fields. The solution to this type of problem is to separate and layer different aspects - the load bearing elements should not be in the way of any conceivable floor plan, since the floor plan will change more than the structure. If architecture is seen as art, modernist styles are disastrous because they tend to experiment, and like all experimentation, fail often. This is acceptable unless failures must be used - in architecture this is true.

And of course, like any good book, it makes fun of somebody. The lessons here? Flat roofs are stupid - the reason they slant roofs is to get the water off. Flat roofs fail to do this. Fashionability is a bad thing when the building outlasts the trend. And, of course, most importantly, geodesic domes were a miserable idea, but everyone was high enough while talking about it so that they didn't notice.