"I really like the approach to expand the security of AMBs, especially since message bridges don’t seem interested in standardizing their interface layer (as discussed in this thread: Standardisation of cross-chain messaging interface).
I do have a couple of concerns, however. Firstly, this approach removes the need for any validation mechanism by AMBs since the Hashi oracle network is extensive. If message bridges implement Hashi oracle aggregators, then there is no need for validation at the transport layer for AMBs since it would be unnecessary. They can simply act as relayers since the bulk of validation can be done by validating the payload delivered by AMBs against the proof delivered by the Hashi oracle aggregators.
Secondly, in such a case, if the oracle network of Hashi decides to work against a particular AMB (let’s say they collude with a competitor AMB), is there a way for an AMB to challenge the decision of the oracle network?"
Can’t we achieve the same by using a single oracle, and multiple relayers (to reduce the cost) and use a definite quorum approach.
For eg., use 1 oracle that just deliver the proof and the rest of them are relayers delivering multiple fragments of the message, then we reconstruct the message and prove it against the proof. (This will reduce the additive cost). just a thought !