Grady Booch is a very well known guy in the world of commercial application development. He sort of gets it right. He states that there is no real architectural vision for most software projects. But what does he mean by architecture, anyway? I always thought architecture was a set of mental models for understanding the structure of something - a building, or a computer program, or a government project, or anything. The problem with this is that there is (1) no real continuity principle in software, and (2) not much in the way of a theoretical basis for understanding software. For (1), a single bit failure in the wrong place can cause unbounded havoc with a program. There's no idea of a "small" deviation from perfection. For (2), I always ask "What's the software equivalent of Maxwell's equations?" The theoretical results all seem to be negative. For example, Gödel's theorem, the unsolvability of the halting problem and the undefinability all tell us what we can't compute. It seems like it would be good to have some sort of theoretical tools which we could use tell us "Do this and your programs will work reliably and robustly."