perhaps we should do a deliberate fork? Set up a pool that the diff hoppers don't know about, mine on that with the wallet unsynced - the diff should drop off so we get consistent (~1min) block times with a consistent hash rate, then we have the longest chain once we resync... Just bar any userid that mines for <10minutes?
Go ahead. Those coins would be useless though since all the exchanges and everything else would be on a different fork
I thought in the event of a fork the longest block chain won, so a chain that has been generating blocks every minute will win out over one where the difficulty gets spiked and so the block generation slows right down... Basis of the 51% attack - generate a 'false' block chain faster than the 'real' one...
Since nobody's replied yet, I'll try to answer this with my somewhat limited knowledge (so don't take my word for it).
As far as I understand, clients are hardwired to use a certain block chain for a certain set of blocks, which is why hard forks require a new client. A 51% attack generates false blocks on the current chain, rather than creating a whole new one.