Mining's energy mix, and why the estimate is contested

There is no instrument that reads the network's electricity consumption. What can be observed is the hash rate, which is itself estimated from block times, and the difficulty, which is written into the headers. To turn those into an energy figure, an analyst has to assume an efficiency for the installed hardware — how many joules each terahash costs — and that assumption is where the uncertainty enters. A more efficient installed base means less energy for the same hash rate.

The energy mix is a further step removed. Even if total consumption were known exactly, the share from each source would depend on where the machines are, and that is not observed. Analysts infer location from pool data, from public disclosures by operators, and from regional electricity prices, and each of those is an input to a model rather than a measurement. Two estimates can differ substantially and both be defensible, because they rest on different assumptions.

This is why an energy-mix figure must be quoted with its source, its methodology and its date. A percentage without those is not a fact about the network; it is a number from a model whose assumptions the reader cannot see. The honest presentation names the model and lets the reader judge it.