Why you should slack
A lovely paper entitled "The Effects of Moore's Law and Slacking on Large Computations" proves mathematically that "overall productivity can be increased for large enough computations by 'slacking' or waiting for some period of time before purchasing a computer and beginning the calculation."
The paper starts:
According to Moore's Law, the computational power available at a particular price doubles every 18 months. Therefore it is conceivable that for sufficiently large numerical calculations and fixed budgets, computing power will improve quickly enough that the calculation will finish faster if we wait until the available computing power is sufficiently better and start the calculation then.
It goes on to supply some math, a figure, and some observations:
[Y]ou could start a computation now, calculate for 40 months, and get a certain amount of work done. Alternately, you could go to the beach for 2 years, then come back and buy a new computer and compute for a year, and get the same amount of work done [in less total time than if you start now].
They do note that the slacking strategy is non-optimal for any calculation which will require less than 26 months' worth of computing time; if your calculation will require less than 26 months to complete, you should go ahead and start it now. Unfortunately, such calculations comprise the vast majority of calculations made. Still, it's comforting to know that for a big enough job, slacking pays off.
I'm reminded of someone I worked with many years ago, a brilliant programmer and cool guy, whose project was scheduled to take something like two years. He pointed out that the percentage of the project he could complete on any given day was infinitesimal, and thus that it didn't matter if he goofed off all day. He still managed to get the project done and polished on time (and it's one of the coolest pieces of software I've ever been associated with), so his strategy must have worked.
For that matter, the stuff that I was primarily working on back then involved animation of computer-generated 3D images of Platonic solids and other simple objects. The renderer took up to 2 minutes to generate each still image; at 24 frames of video per second, plus time to spin up the video recorder, this meant that recording a 10-second sequence required running the machines all night (and hoping you hadn't made any egregious errors that would require repeating the run). If we had slacked for a couple of years and then purchased a mid-range SGI workstation, we could've created the animation in realtime.
Thanks to Charles D. for the link to the paper.