18:08 <Eliel> Does it actually make any sense to have PoW on a chain with large enough blocks that basically only larger companies will ever run full nodes?

18:09 <Eliel> Wouldn't identity based consensus algorithms be more effective with those constraints?

18:10 <Eliel> Or to put this another way, what benefits would PoW provide for a system that is unable avoid regulatory capture?

18:25 <gmaxwell> I can't think of any, if the pow isn't competative it won't be large, and so it won't actually amass enough work to make reorg expensive.

20:40 <Eliel> gmaxwell: why would you assume PoW is not competitive? That wasn't part of the scenario I was thinking of.

23:46 <bsm117532> Eliel: the alternative scenario you present (identity based consensus) is "costless" generation of coins. The only advantage PoW has in that situation is that PoW is costly, forcing the miners to sell some of their coins, and achieving a larger distribution of coins than an identity based consensus would (which economically incentivizes consensus nodes to hoard coins they are allocated).

23:47 <bsm117532> But I'd argue you don't want PoW in such a case anyway. A federation-based chain doesn't need a coin in the first place.

23:47 <gmaxwell> Eliel: I think thats kind of implicit in your thinking, if mining is limited to just a few parties than clearly there won't be a lot of POW behind it. .... if not, for some reason, then at least the POW makes the history costly to forge.

23:47 <bsm117532> and is very much not censorship resistant.

23:50 <gmaxwell> I'd generally say that POW is a terrible way to make a centeralized system. If you're already accepting control from some consortium it's better to just have them sign the results.. that has very strong security (subject to the limit that you're trusting the consortium in the first place).