MRmrb·Apr 9, 2019
No, Bitcoin won't raise global average temperatures by 2°C. This Nature article was debunked by other researchers. I too was asked to review it (recently, post-publication) and the gist of what they did wrong in their analysis is:
* They overestimate (probably by 3x-5x) the present power consumption of miners because they assume the distribution of types of mining machine is homogenous in this table: https://github.com/moracamilo/Bitcoin/blob/master/Randi_Tabl... (when in fact 60-70% of the mining power comes from a single row in this table: Antminer S9). This error causes their calculated average efficiency (J/GH) to be much overestimated as their dataset contains mostly inefficient obsolete ASICs
* They overestimate future power consumption by ~50x by assuming it grows linearly with the transaction rate (in reality it doesn't). Power co
Read the rest
nsumption grows with miners revenues, proportional to {block rewards + transaction fees}. Fees account for currently ~2% of revenues (average of last 60 days). Rewards decline over time, so that fees will eventually account for close to 100% of revenues, which will happen on the authors timeline of 100 years. So if fees per transaction remaimed constant, we could see a 50x tx rate growth with no increase of power consumption.
* They assume the proportion of CO2 emissions per kWh never improves over the next 100 years (great progress of renewables coming to a sudden stop?), and that Bitcoin consumes more fossil fuels than what the entire world currently consumes (infinite fossil fuel reserves?). It is not unrealistic to imagine the proportion of CO2 per kWh could be in a century 20% of what it is today.
Overall, these errors combined mean their scenario overestimates CO2 emissions by a factor of about a thousand: (3 to 5) x 50 ÷ 20% = 750x to 1250x
Edit: simplified bullet point #2