I'm also aware... "I've tried various combinations of "restart suspect miners" and "detect orphaned miners", but I rarely see an effect on this problem. "

However, two times (out of ~100 tests) I have seen the process log state that mm restarted "frozen" miners, but this has been almost impossible to recreate. "Detect suspect miners" seems to happen immediately when you START mining, and when you have 30 miners this can lag for 3 mins or more before any hashing is reported. So I leave that unchecked now. So we've got frozen, suspect, and orphaned miners... maybe Nate could shed some light on this area.

I've read a few posts where people ask for mm to launch multiple instances of bfgminer to make restarting individual miners easier, that seems smart, but I honestly don't care how he gets there. I would love to see some more aggressive restarting of miners based on lack of accepted shares, without restarting all of them. Ideally, if we could rightclick a miner and restarting it manually, that would be awesome.