A nice tutorial. It happens that I've experimented with the same process a few times in the past with moderate success.

Lately, however, I've been taking a slightly different approach. Instead of laying out the plates on my own, I've been letting something like FT Pro generate random land masses and oceans and then trying to work out sensible plates from those results. It's made things interesting so far and, in some instances, let me identify dozens upon dozens of potential plates both above and below sea level.