Lets break this down to the basics at the risk of creating a straw man design:
Step 1) In Bitcoin chain, suspend coins on main bitcoin block chain
Step 2) In Sidechain, verify that the coins were suspended in bitcoin chain using light weight validation.
Step 3) Time Passes
Step 4) In Sidechain, issue transaction that provably destroys coins.
Step 5) In Bitcoin chain, given transaction that provably destroys coins, restore bitcoins.
1) Bitcoin blockchain is secure and has insurmountable hashpower
2) Side chains only need the bitcoin block headers to validate incoming transfers
3) Bitcoin miners need headers of side chains to validate Step 5
- with proof of stake coins, headers alone may not be sufficient for light weight clients because full block contents is necessary to know who should be signing for each block.
- this means miners will have to rely on some kind of block explorer....
- all bitcoin miners must be able to validate all transactions.... which means that side chains require 51% support of the miners.
Bitcoin miners determine which side chains get supported.
Bitcoin miners are centralized which means experimentation will be limited.
Miners would have to voice support for a side chain and only allow transactions in/out of the side chain while 51% of blocks are produced by miners that claim to recognize the side chain.
So while steps 1 through 5 sound spiffy in theory, I believe that the only way to solve step 4 to 5 in a reasonable way is to assume that all side chains are also secured by proof-of-work which can be used as the simple, generic, validation measure. This means that side chains must be merge-mining based. I think this is a case of them making bold claims without actually walking through the details.