> Measuring energy use per transactions makes no sense
I am well aware of the difference between average and marginal energy use. This would be particularly relevant if the blocks were not full, but mined anyway, so that any additional transaction could've been included "for free". However, more often than not, that's not the case, but the mempool is non-empty, and the constraint (of max transactions per block) is binding. [1]
Thus, one can not trivially argue that the marginal cost of a transaction is zero.
Next, yes, my figures (165 TWh/a = 19 GW) from Digiconomist [2]) are at the upper end of the estimates, but other figures (eg Cambridge Bitcoin Electricity Consumption Index) estimate 100 TWh/a, with reasonable bounds of 36 to 376 TWh/a, so 12 to 28 GW, so are largely in alignment, modulo a factor of 2. (Note that your source is also within a factor of 3 of those estimates, pegging BTC at 0.2% of global electricity consumption.)
At any rate, the average cost is just preposterous, even if off by a factor of 3.
I am well aware of the difference between average and marginal energy use. This would be particularly relevant if the blocks were not full, but mined anyway, so that any additional transaction could've been included "for free". However, more often than not, that's not the case, but the mempool is non-empty, and the constraint (of max transactions per block) is binding. [1]
Thus, one can not trivially argue that the marginal cost of a transaction is zero.
Next, yes, my figures (165 TWh/a = 19 GW) from Digiconomist [2]) are at the upper end of the estimates, but other figures (eg Cambridge Bitcoin Electricity Consumption Index) estimate 100 TWh/a, with reasonable bounds of 36 to 376 TWh/a, so 12 to 28 GW, so are largely in alignment, modulo a factor of 2. (Note that your source is also within a factor of 3 of those estimates, pegging BTC at 0.2% of global electricity consumption.)
At any rate, the average cost is just preposterous, even if off by a factor of 3.
[1] see for example https://jochen-hoenicke.de/queue/#BTC,1w,count
[2] https://digiconomist.net/bitcoin-energy-consumption/
[3] https://cbeci.org