I just spent a few frustrating days trying to locate the fault in an experiment I've been programming and running in SuperCollider. Finally, at 6:40 this evening I found the problem and fixed it. It was a relatively simple fault in the way one of my components worked.
It reminded me really strongly of Richard Feyman's Personal observations on the reliability of the Shuttle, which someone recommended to me once. Anyone who has ever run a project or tried to write more than two lines of computer code should read this article and take it seriously. There's some interesting stuff about the space shuttle, but even more there's a lot of wisdom about how to make things go right and how to make things go wrong.