Adventures in Commuting, That Happened Edition:
I've had the theory since I switched to my (shorter) commute that the marginally later time I'm able to leave is somewhere after the tipping point at which traffic becomes really bad. After a few successive 1:40 40-mile commutes, I decided to do some empirical testing. Turns out I'm correct: if I leave 20 minutes earlier, I get to the office 60 minutes earlier.
...and don't really have enough class prep to do to fill up the extra time. So it's probably back to sleeping an extra 20 minutes.