The theory of money is all about energy. However, proof-of-work is not quite the same as proof-of-hash. Until the computing device reaches the theoretical physics limit, hashing power per energy power is always improving. For example, the increasing historical difficulty in a time series can not be totally accounted by increase of energy power. This leads to the necessity of some solution to assess mining rig efficiency in a time series so that a real difficulty time series can be shown. Should efficiency x10 and the raw difficulty x200, the real difficulty only x20.
The data of mining rigs (the hashing rate, the power consumption, the sale price) is public. Rationally, people shall choose the rig with best profit in their business plan horizon. The best rig of a 2 year horizon is not necessarily the best rig of a 1 year horizon. Very often due to the sale price of a rig, the rational choice may not be the one with highest hashing rate per energy power. Recalling the arbitrage-free mining power distribution for different business horizon, one can compile the rig efficiency time series based on the available public data by the following steps:
As mentioned, until the speculation period ends, any deduction from arbitrage-free argument needs some grain of salt. Some other ad-hoc method to assess the network overall rig efficiency may be necessary, for examples:
Whatever approach always concludes the market share of the highest h tends to 100% as the business plan horizon becomes infinity.