Bookmarked so I can give this a proper response, but let me see if we agree about some basic principles and terminology:
The set of participants who hold hard power over the protocol are the actors who receive coins in exchange for something else. They decide if a transaction is legitimate or not in the same way that the clerk in a shop decides if your banknote is real or not. In your terminology this set of participants is the social layer.
Block producers provide immutability as a service. This is what you term the consensus layer.
What emerges from the interplay between the two is a market where immutability is traded.
The adversary is an actor motivated to attack the market itself and prevent immutability from being traded, except perhaps under certain conditions (e.g. seigniorage, KYC, etc).
Login to reply
Replies (2)
Small note, miners do not decide whether a transaction is legitimate, they can only decide whether a transaction is illegitimate. They do this by not including it. This is a somewhat weak mechanism, limited by the degree of decentralization of mining.
Conversely, if they try to decide that a transaction is legitimate (that doesn’t follow historical rules) then everyone’s nodes will ignore it. This ultimately falls back to a social layer, but there is a strong bias toward not changing anything since everyone needs to physically update their nodes to work with the now-forked chain.
Social layer and consensus layer are oversimplifications, but my point is that the "social layer" cannot include a transaction that has been censored by the "consensus layer".
Thus, the attack vector that will be used is mining empty blocks.
It's not possible to use external capital to overpower the adversary if your consensus mechanism relies on stake. This is why Satoshi used PoW and not stake based consensus.