Here’s the full paper. And some of my notes / excerpts below:
- Use of multi-sig in a public blockchain reduces privacy not only of user, but of other users as well
- Use of mixers is a poor privacy measure and potential attack vector, as shown with Bitcoin and Dash
- A good chunk of Bitcoin blockchain transactions still carry no fee, and often reflect a direct relationship between miner and transaction generator
- Bitcoin transaction fees still amount to only 20% of miner reward (but growing)
- Bitcoin velocity is much lower than expected (this, IMO, was the most interesting part):
- the velocity of Bitcoin works out to 1.4 per month averaged over the period January 2016–July 2017, compared to 5.4 with the naive metric. Our revised estimate is not only much lower but also much more stable over time.
- only a small percentage of bitcoins are used for activities other than investment and speculation, although that fraction has been gradually increasing over the past year