@tyj8tim: Share difficulty has nothing to do with reject percentage. Or with overall profits, or with anything that matters.
I actually have my doubts on that with a switching pool.
I agree that IF we mine at a pool that only mines one coin, then the variance will average out and you will see a steady payout after a few hours or 24 hours.
However, at CM we get a lot of stale rejects. And I'm assuming these are not all counting towards our hashrate (correct me if I'm wrong). From my understanding, stale rejects are the ones that's too old because the pool already pointed to the new block. So if someone has a ~10% reject on average, that means ~10% of the time when he finished his share, the pool already moved on and his calculated shares are useless. This is especially a problem for a pool that's hopping around and getting into block times that's less than a minute or even 30 seconds.
So IMO, by lowering the share difficulties, you can shorten individual miner's share round time and minimize the probability of submitting stale shares.
Your thought process regarding lower diffs for a miner has been argued to death over at MC, the consensus is no, diff does not have any long term effect on profitability or otherwise.
Anywho, I see a lot of posts whining about high rejects and swapping pools because of it. But no mention of profits had during said period. If you swapped away from a pool just because of rejects without considering your btc/mhs, you're stupid.