Today I was musing on the complaints that Bitcoin network is too energy greedy and centralized (let's not discuss if this is actually true) and the suggestions that memory bound POW would be improvement for both issues.
My thoughts are that:
- regardless of POW mechanism the main cost of mining is operational cost, and operational cost is mostly energy; this is becomes even closer to true assessment as time goes by;
- given the above, for the same price (which is mostly demand driven), the same operational costs would be justified, so the same (order of) energy will be spent;
- I assume that memory bound hardware to be (much) more expensive than current ASICs (for the same value (as in mined coins) per input energy); this seems to work only for the benefit of HW vendors and increase the threshold for entry, thus working against the decentralization.
In the end (which means after some initial period - exemplified by the raise of ASICs for CPU POW) it seems that we end up with a network that costs roughly the same to the environment, but more centralized because of higher HW costs.
Does this reasoning make sense. Are some assumptions too speculative or outright false?