Dr. A. Back is quite active on the bitcoin reddit discussing the sidechain project. From that discussion I gathered that there is actually no limitation to the type of blockchain you can "peg" onto the bitcoin-mainchain, so for the sidechains ASICS are not obligatory. Also since the whole idea is to stop minting of new bitcoin-tokens and only transaction fees to keep these sidechains running, convincing mining-pools to merge mine, might be more difficult than going with a different model like NXT-forging or bitshares-DPOS.
I don't like mining centralization and also have issues with the high energy cost for just security, but I do think this is a pretty clever and clean way to use the available hashing-power for bitcoin. Now that scrypt has ASICS and ostensibly asic-resistance on pow-chains is futile, this is a way to increase efficiency a bit for that model (even if it increases centralization).
I think I mentioned this before, but I do think bitshares should be very alert to the possibility of being copied and Borg-ed into sidechains and in my opinion should supervise the Borg-assimilation itself. Not in the last place because these sidechains also seem to allow easy forking or transformation of bitcoin to a different blockchain-design. If a sidechain becomes more popular than the main-chain I would expect that sidechain would become the new mainchain and this seems to open up possibilities for bitshares and DPOS as well. In short if bitshares-projects are going to be assimilated anyway, why not let bitshares try assimilating bitcoin? Also the sidechain 2-way-peg-solution might also offer another way to move between the different DAC-chains of bitshares and open up another range of possibilities.