>You're disciplined enough to keep your functions to 20 lines, so it's easy to find the initialized value. Others (like me) aren't so lucky.

Al,

I don't think it a matter of discipline, rather I think it's a by-product of the design methodology I use. While there's nothing magical about 20 lines of code, I use it as a flag to indicate that perhaps the method isn't as functionaly cohesive as it either could or needs to be. If I re-factor what the defined purpose of the method is and I find that, indeed, it could be done better, I do it. Otherwise, I don't.

One of the things that this approach yields is smaller, more easily maintained code. Afterall, it's always easier to find a problem in a smaller module than it is in a 50+ one.