Let me preface my remarks by stating that I personally am fixated on keeping activity on the 'mainchain' as minimal as possible. This to keep it super lite and the entire support infrastructure distributable mostly in terms of messenging bandwidth. I want to minimize transactions.
I've claimed to you before that this is not necessary because the consensus design that solves the dilemma facing us now has none of the tradeoffs that cause you to wish for that. Your wish does not solve the dilemma. Please review the following post I just made today:
https://bitcointalk.org/index.php?topic=68655.msg11713418#msg11713418Economically, I feel that if an on-chain transaction represents thousands of off-chain transactions, very generous fees can be paid to core infrastructure supporters while end-users continue to pay a tiny fee (if any) for their coffee purchases.
The problem is you've implicitly asked some entity to aggregate fees which is the antithesis of decentralization, i.e. you would end up with few entities controlling the bulk of the transactions fees to paid to Core. The only way to solve the dilemma is with my consensus network design.
The movement of coin assets from one chain to other can't realistically be done piecemeal at a moment's notice via the SPV (as you point out far too much risk to allow that), instead atomic swaps should be used.
The SPV proofs should only be used for those willing to accede to very long contest period, say 30 days or so.
I don't really understand your atomic swap ideas with out more context. Specifically, swap what for what? I suppose BTC for sidecoin?
This is covered in Appendix C of the Blockstream whitepaper (which derives from an thread where TierNolan invented the concept for decentralized exchanges). Swaps between BTC to or from scBTC. Hopefully the light bulb goes on for you now?
If so, this is still activity on the mainchain which I wish to avoid.
The main reason you need SCs is because it is the only viable means to transition BTC to a new experiment which solves the dilemma. Otherwise, my idea would be released as an altcoin which could possibly destroy BTC (believe it or not).
With enough 'lossyness', I can imagine some efficiency, and I suppose that this is where your 30-day figures come from. Fine, but it still seems as kludgy as anything I could come up with in my early attempts.
Speculators are not going to do this for free. Thus rapidly moving assets between side chains is going to be lossy, i.e. you won't get 1 BTC for the 1 BTC you swapped (when demand exceeds supply) but going the other direction you will get more than 1 BTC (when supply exceeds demand) unless of course you accede to the long contest period.
Side chains which enforce a longer minimum contest period will have more lossy atomic swaps, but gain confidence.
Speculators are the last people I would give much of a shit about (although I am one myself and intend to continue to be.)
...
I have always had a high respect for the efficiencies that lossy systems can achieve. This from my experience in engineering. I would say that most parties playing in the reserve currency sphere are able and willing to play under a lossy paradigmn. It's just standard risk management which they are competent to do. The main thing they need is transparency.
...
I see nothing wrong with the lossy option. If you don't want it, then accede to the long contest period and get a 1-to-1 transfer. If you want speed, then pay the very small lossy fee at the market rate for an atomic swap. You will likely get it back going (atomic swapping) the other direction, so it will balance out and it shouldn't be large because speculators are efficient.