Consensus Group Centralization: Why Are We All Here, Anyway?
Let’s imagine that David’s talk is more of an open question that he proposed; and then get to the end game he was imagingin may exist.
Say the Consensus Group is the group of miners, stakers, & req’s to select & censor transations. If the consensus group is one party, it’s $ (cash).
Iterations: PayPal (didn’t succeed in its initial goal for digital money for the Net b/c it’s best known for closing accounts, not doing business in many parts of the world).
Try different consensus groups: a group that can’t coordinate effectively?
Two research areas:
1) PoS … very centralized. Has always been; today ~50 corporations.
Has its uses. Libra as an example is a better model than others: explicit parties run the system, there are ways to regulate and sue.
2) PoW … even bleaker? mining centralized to 1-2 companies. Very sensitive to power cost. economies of scale are unclear. If we trust David’s analysis, maybe we give up here. If the # of entities in the consensus group will be fewer than in Libra Foundation, this might not be an improvement. But!
a) Separate miners and pools. Today consensus makers are pools, but we could make them [individual] miners. Not inherent that pools pool decision-making; that’s more an artefact of current protocol designs. Cf: p2pool, slushpool data.
b) Research needed into the power market structure. Interruptible power is ~free in small quantities; industries pay a premium.
Asides: research payment-channel network privacy! We don’t know if onion routing provides privacy. Balance disvovery… other correlation attacks… does probing protection work? Needs more formal work.