Some people do modify the times, apparently, but in my limited understanding the effects should be fairly marginal.
The idea with two-bath developers is that the first bath soaks the film in developing agents, then the second raises the pH so the absorbed agents begin actually developing at a reasonable rate. The first bath by itself is a SLOW developer, so if you leave the film in for a long time you get some additional development there, but I don't know how much difference it's realistic to expect in a matter of minutes. In the second bath, more time should have essentially no effect---once the amount of developing agents absorbed from bath A has been used up, it's been used up, and there you are. I guess the most effective control for reduced development would be a shorter time in bath A, followed by a normal bath B (absorb less developer, then use up what you absorbed).
I use it occasionally for the speed boost---HP5+@800 and Tri-X@1250, in my experience---but mainly for its compensating effects. Works well with high-contrast films as jbrubaker mentioned; I get contrasty mids but with a well-controlled toe and shoulder. Something similar happens with the late lamented Efke IR820, making the IR effects a little less extreme; and I've had good luck with it for long (tens of minutes) nighttime exposures on Fomapan 100, of all the strange things. It's not clear to me how much of the compensation is special to Diafine and how much is a general property of the two-bath approach.
Also, the A solution goes through various attractive colors from the antihalation dyes. The B solution seems to converge on a kind of honey color.
-NT