It's a 3 page document that makes a number of false assumptions. They extrapolate bitcoin to having 300 billion transactions and means no lightening or L2 solution and the block size would be 3gigs. Nor do they even factor in the halving. Wasn't even researched properly, just utter nonsense for the papers.

And Joe Rogan was showing how easy it is to publish fake papers in journals. https://www.youtube.com/watch?v=AZZNvT1vaJg

here is the paper by the way behind a shitty paywall. https://www.nature.com/articles/s41558-018-0321-8

Full text here: http://sci-hub.tw/10.1038/s41558-018-0321-8

Data & analysis is here: https://github.com/moracamilo/Bitcoin/

The argument is:

1. Assume each mining pool's users are in the pool's home country, and use their country's average mix of fuel sources, and use the hardware in supplementary table 1 (which I haven't found), then Bitcoin emitted 69 MtCO2e in 2017.

2. Assume Bitcoin's usage increases at the same rate as a sample of other technologies (television, smart phone, washer+dryer...)

3. Assume that Bitcoin's carbon footprint increases with usage.

Then you can multiply the projected usage by the yearly carbon footprint, and estimate the effect on temperature, at 2°C.

A good counterargument is that Bitcoin's mining footprint is not proportional to usage. It's proportional to price * block reward = profitability.

However, this might balance out in the end. The block reward is decreasing exponentially every 4 years, but the price increases faster than usage. And when the block reward drops to zero, transaction fees will become the sole incentive for mining, which are proportional to usage.