When I was very young, and I was just becoming aware of a thing called "science", I thought that I could do something really worthwhile if I dedicated my professional life to doing research into fusion power. Even back then (around the time Neil Armstrong was making his "one small step") the prediction was that it would take several decades to get to the point where we could build a commercial nuclear fusion power station. That suited me because it meant that I would be involved in the main phase of the work, and I would retire at about the time these power stations came on line, and they would then keep me nice and warm in my retirement. The idea of infinitely extensible project time scales did not occur to me when I was very young.
Now I read the article Dream machine in the latest issue of New Scientist that the projected year for operation of the first commercial power plant is around 2050, and that is an optimistic projection. 2050 is almost certainly after I am dead. I am so glad that I didn't follow up my idealistic childhood dream of doing fusion research.
I feel much the same way about where I could have ended up after doing my PhD on QCD. I would probably have eventually been seduced by the beauty of string theory, which had a resurgence at about the time that I finished my PhD, and I would have ended up doing research on a long road to ... well, it might be somewhere, but it also might be nowhere (see Peter Woit's blog for lots on this). Ha! I couldn't resist a gratuitous swipe at string theory!
I like the sort of research that can be done successfully using simple equipment (e.g. a standard PC running Mathematica), that achieves useful results with the investment of only days or weeks of my time. The information processing research that I currently do admirably satisfies these criteria.
Endless projects? Not for impatient me.