Thought provoking take from
@alpacasw. Throwing out some shower thoughts:
The math is a bit off but the idea is if you evenly divide the current hash pools into equivalence of keys, then you have 2 pools that can collude and steal because they can vote to withdraw to themselves over ~3 months. It's unclear if people during this time would switch their hash rate away from these pools if they knew their pool was going to get a pay-day from the DC theft. Smaller pools wouldn't necessarily benefit from the theft. In return it could temporary cause more mining centralization where miners would want to be part of one of the benefiting theft pool. After the theft is complete, I imagine the pools would go back to some sort of homeostasis.
There are 2 things DC game theory relies on to combat this:
1. Fees on sidechain, which would make it unattractive to destroy the sidechain in the first place.
2. If the sidechain has enough economic activity then there is the potential idea of a UASF in response, to combat this threat.
I think of UASF's as the "nuclear option". Still imo never really been tested despite some wanting to say this was how segwit v0 was activated. Unlike theft from a spend any segwit transaction, theft is part of the protocol with DC. It wouldn't defaultly cause a chainsplit.. similar if DC were activated it wouldn't cause a chainsplit unless users wanted to actively look for suspicious DC activity and do a URSF or support minority miners stealing before the 3 month period (imo the cases for URSF/UASF are non-sensical and seem silly in both directions for DC - made more sense for segwit).
For hashrate escrows the interesting thing, (considering there isn't other issues with MEV!), is this is a permissionless federation. The term "decentralization" in the Bitcoin/Blockchain land was never about geographic distribution or division of pool entities, instead it was about whether or not the protocol was permissionless to allow newcomers to be first class citizens of the network without formal agreements and bureaucracy. DC is a true decentralized protocol where miners can (ignore, participate, vote, attempt theft) at will. Unlike Liquid, which uses signatories that potentially need to revoke keys, sign business contracts, be known (or semi-known) entities, and coordinate off-band to form the federation. (This is not to rag on the security model of Liquid btw, as
@LukeDashjr believes this is superior to DC and historically
@Excellion has said the only real attack on Liquid is an Oceans Elven style attack).
Decentralized idea of PoW blockchain governance is fascinating to me and why I believe most of us have accepted Bitcoin as something to pay attention to. It's unclear to me if this idea to secure L1 is also a good/bad idea to secure sidechains via a hash-escrow (given so many strong opinions, I've come across recently). The unfortunate part, imo, it's unclear if it's possible to prove any complex game theory without actually testing on the mainnet. I believe it's unlikely but also unfortunate this couldn't be "stopped" if supported by miners. Zealous bitcoiners that truly want to protect the longevity of Bitcoin would attempt to pressure miners to abandon or steal from DC's (i guess?). For me personally I would like more core maintainers/contributors to feel comfortable with DC before I personally felt warm and fuzzy with it. If done without core, via a public MASF, it will be the first one I've known about done this way (Unsure but my guess is there's a bunch of insignificant secret MASF's that aren't bothering people), but this SF with current implementation would be identifiable a couple ways including the spend-any escrow accounts.
Drivechains make Liquid look decentralized. 21350 keys but 10,000 of them held by two people.