I mentioned this back in an old thread, but it's worth hypothesizing that if the NCAA goes with your first idea -- the lowest overall 8 teams have 4 play-in games -- one of those winners will most likely beat a #1 seed in the not-too-distant future.
68 teams means 17 teams per region. This means:
The 17-seeds of 2011 are, essentially, the 16-seeds of 2010.
The 16-seeds of 2011 are, essentially, the 15-seeds of 2010.
Taking 2010 as an example, we had a 15-seed (Robert Morris) almost beat a 2-seed (Villanova). Villanova was very much in the mix for a 1 seed. This had legitimate upset potential; it was not a case of lousy seeding. The Selection Committee has gotten much, much better at evaluating low-seeded teams.
I realize that the best 15-seed in 2010 would stay a 15-seed in 2011, as the bottom 8 teams in this year's tournament were actually:
So yes, if Robert Morris were considered the best 15-seed in a 68-team field, they'd remain a 15-seed and avoid the play-in game. But it's no longer a stretch to imagine that one of four #1 seeds that happens to start a little rusty might fall to one of 4 play-in winners that has a little momentum. And the concept of a play-in game would no longer seem so humiliating.
Unrelated point: The second idea -- pitting the lowest at-large teams in play-in games -- would be disastrous for the fill-in brackets. The NCAA knows better than to kill the greatest promotional tool in all of sports.