Spending material wealth on mining is what makes it secure. We work on improving the per-hash cost because it makes it more difficult for someone with a brilliant technological innovation to compromise the security of the system.

And there is incentive to improve the total energy use. Every participant in the system pays for the electricity that secures bitcoin in the form of inflation and transaction fees. If some alternative to Bitcoin can provide all the same benefit but with less inflation and lower fees, then there is a good reason to switch to that alternative. And people are trying to innovate here continuously, the blockchain sector at this point is getting more investment than the AI sector.

Basically your argument is, "Bitcoin may not incentivize improvements in energy efficiency, so we should scrap Bitcoin and use something more efficient." Great, let's do it. The technology is already here, in fact, the technology was always there. Have a consortium of banks operate a distributed public ledger. The banks will charge fees to include transactions on the ledger, just like miners charge fees now, and the economic incentives to raise or lower fees will be similar -- but the fees will be lower because the consortium members will not need to cover the electricity cost of mining, only the cost of operating their servers.

The only problem Bitcoin mining solves is the lack of identification among the miners (i.e. the "Sybil attack" problem). Seems pretty obvious to solve the problem by introducing identity, particularly given that the large miners / mining pools are not really anonymous. Sacrifice the small miners who contribute little to the system, set up a consortium, and enjoy whatever benefits the distributed ledger provide without the waste.

(Yes I know, banks are evil, how dare anyone suggest that we acknowledge any authorities in a system, why should we trust the banks, etc.)

This is a pretty classic HN take. "It's really easy... just do X".

This is a multi-trillion dollar opportunity, if it's that easy, then just do it!

aka https://en.m.wikipedia.org/wiki/Jevons_paradox

Nope, not the same. Bitcoin mining does not "produce" hashes, it "produces" Bitcoin blocks. All that changed with the improvement in hashing speed is the number of hashes that must be computed before a block is added to the block chain.

i dont follow. moving from cpu to gpu to fpga to asics (each a gain in efficiency) did not reduce electricity consumption, because it in effect accelerated the rate at which the hashing complexity was raised by the network, in effect negating up any efficiency gains. that's exactly the phenomenon.