"If you see something that needs doing, do it. It is not easy for me to explain what I mean by this guideline, but I think it has been deeply ingrained in me for a long time. I rather believe that I was once promoted to Acting Corporal at the age of 19 because of this trait. I suppose that it entails a partial contradiction of the conventional injunction to get your priorities straight before acting. To my mind, the priorities are not so certain that one ought to pass up any opportunity to get something useful done. Perhaps it also reflects my belief that much more good is done by tinkering than by starting over from scratch. I claim for this approach that it fits in with the Hippocratic injunction to the doctor to "do no harm" and that gradient methods are a good all-purpose method for local optimization. What about global optimizations? Good point. I suppose I worry that enthusiastic seekers after global maxima run the risk of falling off steep cliffs. On the bad side, I know I sometimes find myself doing meaningless busywork when I could presumably spend my time at something more useful. My wife reminds me that once, when we discovered that the automatic wake-up mechanism in our hotel room was not working, I spent an hour and a half trying to fix it. (I got it to work. Once.) No recipe for coping is perfect."