Mining at the beginning of a round is only better _if_ a block is found quick, because if not, the exponential rise of score per share submitted will make your early shares obsolete quite fast.
No, that's not true. Mining at the beginning of a round is better no matter what. It's better because if you find a block before anyone else submits a share, you get the same reward as if you were solo mining and if a block is found shortly after you submit a share, you get a non-zero payout. Thus mining at the very beginning of a round is better than solo mining and clearly it cannot always be better than solo mining.
All you must do to show that a pool is vulnerable to hopping is show that there is some point that an attacker can identify at which the expected return exceeds the expected return for solo mining. If I understand this payout method correctly, it has such a point -- before the first share is submitted in a round.
If you could predict how long a round takes, you could just mine in the short ones. But switching to solo or another pool after a certain time in the round will very likely result in no reward (or a very very small) for the round, since shares submitted by others after you leave will get a much higher score for them.
That's all true, but has nothing to do with anything. Sure, there are lots of ways one could hop pools that don't work. The question is whether there are any ways that do.
Besides, in your example, if someone else submits a share and you mine a block very shortly after, you only get part of it - thats not better than solo.
I agree. That's why the pool is vulnerable to hopping. There are good times to mine and worse times to mine. (Assuming I correctly understand how the pool pays out, which I hope I don't, since this is really an obvious flaw.)
