Why shouldn't the federation secure its own deposit? It deposits via peg-ins, and that liquidity would attract users. I think Liquid has been implemented like that, by I don't cross my fingers.
Liquid is a centralized and static model; the incentive to lock capital in a massive initial peg-in here is simply that as the Liquid foundation manages the access to the sidechain forever (or at least for years), and thus it can collect all the fees.
Dynamic federations like those I mentioned are a bit more complicated. Of course you could start with a group of people bringing in their capital and start a sidechain in the Liquid fashion. But after that start phase is over you need rules to ensure that the peg-ins and peg-outs will be managed in a sustainable way and that the incentives for correct behaviour and availability/liveness always work. These rules have to crafted in a way that even if the sidechainBTC liquidity on the sidechain approaches zero, it continues to work.
You could say that a sidechain with no sidechainBTC would have failed, of course. But if the sidechain security depends on the sidechainBTC liquidity, there may be attack scenarios of an actor orchestrating massive peg-outs to generate doubts, driving out the current members of the sidechain federation (decreasing their incentives to participate), and replacing it with your own nodes and then attack it.
So you need a long "start phase" (or a massive publicity buzz) where you basically compete with all kinds of centralized and semi-centralized sidechains and IOU-type services and have to build up a lot of liquidity.
I guess what you say could work, but it would need a group with massive capital to start it if it really wants to be successful. Big mining pools who merge-mine the sidechain would be ideal. Instead, a sidechain with an auxiliary currency can be started by a "poor" developer group, even without a premine, because the auxiliary currency rewards is what ensures the security, not the availability of sidechainBTC.
I think thus the most realistic solution really is a strong existing altcoin (preferrably a non-premined and decentralized one) implementing a two-way-pegged Bitcoin-token to operate as Bitcoin sidechain.
It sounds really cool as concept. It's orders of magnitude more efficient than lightning, because it's multiple bidirectional channels opened by just one UTXO, which can communicate with other such UTXO (pools). The problem with this solution is interactivity:
It seems at least a bit better than Channel Factories, see
this beginner-friendly article (on the BitMex blog) - a single leaving member won't force the pool to "reset". On the other hand afaik channel factories could be implemented already now, albeit they would benefit greatly from SIGHASH_ANYPREVOUT. CoinPool needs SIGHASH_ANYPREVOUT and other changes to make it really useful.
As an already existing solution there's also "Lightning Account management" (see
LNbits). This seems to be a simple Lightning channel managed by a group instead of a single person, but the group members have to trust themselves so it's more a solution for families, friends or so. CoinPool and channel factories instead don't need so much trust between participants, only trust in their availability.
From all the solutions I've seen, all have their tradeoffs. In general I think the sidechain model is the one which allows you to operate closest to the traditional on-chain model (i.e. no availability is needed) and thus I think it has such large advantages I would accept an auxiliary currency as tradeoff. I'm not "maxi" enough to see that as a disadvantage
![Smiley](https://bitcointalk.org/Smileys/default/smiley.gif)