But that’s not what happens. Data centers have been getting bigger and bigger and bigger and now we literally talk about planning gigawatt scale data centers. No “data center” took the efficiencies of Moore’s law and said “we good, we can stop making more complicated uses of computers now and just run what we already have on more efficient hardware”.

So then it comes down to, we’ll is that extra processing these data centers doing useful or wasteful? And that’s all subjective. Some of its more clear cut probably: work that leads to new medicine probably useful. Work that hijacks dopamine processes to addict people. Arguably not useful. But how do we determine that fairly? And where would we draw a line on this kind of moral policing of energy use?

Very good point – arguably many companies also just have a constant budget for their computing expenses and will buy the biggest/most boxes they can get for it.

I still do think that it's an important difference whether becoming more resource efficient is something that could be incentivized/enforced without endangering the underlying business model or whether resource waste is the business model.

No, the amount of energy "wasted" is not measured, the incentive is hashrate, and each generation of hardware is more efficient in terms of hashrate per joule.

> the incentive is hashrate, and each generation of hardware is more efficient in terms of hashrate per joule

That only affects who finds the next block. It will always take about 10 minutes, regardless of hashrate, energy usage, hardware efficiency, number of miners, etc. since the difficulty is adjusted to factor all of those out.

For anyone who isn't a miner (and doesn't care to become one), the only variables which matter are energy-usage and difficulty-of-51%-attack (i.e. double spending).

AFAIK the only reasons to advocate for bitcoin mining are:

- A vested interest. Since bitcoin is a zero/negative-sum game, nobody else should care about someone else's vested interest; if anything they should be opposed (in the negative case; e.g. as a waste of shared resources).

- belief that double-spending is currently unacceptably easy, and would become acceptably difficult given the increased hashrate.

Whilst I don't know of an objective 'acceptable optimum' for the difficulty of double-spending, it would take a lot to convince me that (a) there are credible threats given the current hashrate and (b) those threats would be solved by an increased hashrate. The only threats that even come close would be on the scale of nation-states, or perhaps the largest tech multinationals; yet even if we're paranoid enough to believe they could pull off an attack, I fail to see how burning even 10 times as much energy on hashing would disuade such a level of paranoia.

We might be talking past each other, the point I was disputing is that miners are incentivized to expend energy, which is not true. They are incentivized to produce hashes.

If you hold the hardware constant then hashrate correlates to energy usage obviously, but it's not quite the same thing.

Arguably. Hashrate was chosen as a proxy for compute power (which was chosen as a proxy for time). Compute power (or hashing power, now we have ASICs) is effectively a measure of energy, since the hardware cost gets amortised.

It does lead to the same thing though if you think it through: There's always a constant number of transaction space available in a 10 minute window, which is auctioned off to Bitcoin users and yields a certain sum of unclaimed mining fees. This sum is importantly unrelated to either the current hashrate or the energy efficienc in standing up that hashrate.

As long as the total hashrate is "cheaper" than that, in terms of hardware and electricity costs, there is an economic incentive for somebody to start mining (more) – so somebody will.