[Paul-E]

The paper that relies on turns out to have some inaccurate assumptions. The paper assumes that miners can flip their hardware on/off at essentially instantaneous intervals. The assume this to argue that miners will strategically turn their hardware off when the expected value of mining a block drops below the cost of power.

It turns out that many of the big miners have long term contracts with power companies to consumer power; they wouldn't save money by turning their hardware off for short periods of time. Power companies like this arrangement because it lets them predict demand better, and miners like it because they get "bulk rates" on electricity for being predictable in their consumption. The arrangement falls apart when miners turn their hardware on and off.