Please correct me if I am wrong.
The difficulty constantly changes according to the current hash rate. It changes to ensure that a block is mined every 10 minutes. If the hash rate was 1000 TH/s and a pool contributed 1000 GH/s, then that pool should theoretically be getting 0.1% of the coins being mined. Now say, hypothetically, that 90% of the network hashrate were actually rerouted to mine another SHA256 coin. If every pool and solo miner was reduced to 10% contribution to bitcoin mining, the payout percentage (and ultimately absolute payout) should remain the same. (100 GH/s)/(100 TH/s) = 0.1%. I am of course assuming that payouts work in this linear manner; I am not sure if they do. Now the other 90% of the network hashrate could be essentially one big pool mining another coin.
Comparing bitcoin and peercoin: 28,780.806 GH/s 151.079 TH/s
(0.90*28781)/(.90*28781+151) = >99%
Why then couldn't bitcoin be rerouted to do 51% attacks on various SHA256 coins until people leave each of those communities and move onto other coins until bitcoin is the only SHA256 coin left? Likewise, this could be done with scrypt coins as well, but each in its respective sector. My first assumption is that this is not possible, or else it would be used already.