The original suggestion of using 15 witnesses to be holders of the private key of a multisig address seems simplest and best.
Although, I am still unconvinced that bytemaster's suggestion of the top 15 is the best option. bytemaster seems to think that it's not an issue, due to historical data, that a voting-out event could occur. That's fine; however, we should always recall that past results are not indicative of future results.
The better solution (and I refuse to call any possible solution the best) is to randomly select 15 of the witnesses to be private key holders, with possibly a slightly weighted distribution towards the top ones. This way, even if there is a switch between the 15th and 16th witness, we wouldn't have to constantly move to a new multi-sig address for the collateral.
Best block producers of BitShares are probably not the best signers of BTC address. Imo we need another voting algorithm for the BTC multi-signers, for better security, for example one stake can only vote for one signer, so BTC38 can't control all signers. We can probably ask the signers to put some collateral in BitShares, for example to an account under the control of the committee (the voting algorithm of committee should also be revised), to make sure they won't steal customers' funds.
Randomness is evil. Stake talks.
This extra transaction seems cumbersome and also is more costly, in terms of transaction fees on the bitcoin network. If you foresee this sidechain business as something that could happen with any altcoin (which I do), then mitigating additional and unnecessary transactions by properly creating a sidechain protocol for any alt on bitshares is the proper way forward.
This is good will. But I don't think BitShares is powerful enough to let other chains adapt BitShares's protocol.
tl;dr -- side.btc seems superfluous and has potential issues. Simpler solution is the original one offered, i.e. 15 of the witnesses hold the private keys to a multisignature address that contains the collateral of the bitcoin.
I understand the security concern regarding the witnesses. However I think if we sidestep them as our infrastructure of trust, we are losing the opportunity for them to grow.
At present the level of trust given to them is relative to the stakeholders regard for BTS block witnessing.
It's interesting that now that we would add BTC to the witnesses suddenly there is all this security concern... the same concerns that exist 'now'.
I think going through this process will cause the voting stake holders to become more active if not take the process of voting more seriously.
It will also force witnesses to step up their game or be voted out... I know we have similar witnesses to what we had as delegates in the past largely due to associations with known large stake holders, but with something like this that brings in whole new holders that are going to look at every witness with a far more objective eye and judge them on their presentation, credentials, and their infrastructure.. we are going to see that whole arm of our TBD (three branches of delegation) mature.
If we side step them with something like this though... they stay the same and don't improve.
There is the other issue of expense you mentioned though.
We can have a greater degree of security, however multisig transactions in the bitcoin network are more expensive.. and that will get passed on to the end users... I suppose that is not a bad thing.. it might actually encourage some to get people signing up to bitshares to get their BTC sent to them... in record time at that!
Great discussions over all from everyone.