A new paper has a crisp economic analysis that captures many observations about the security and stability of Bitcoin and proof-of-work blockchains. The most interesting part is the three “collapse scenarios”. I’ll summarize here, with comments. [Thread] faculty.chicagobooth.edu/eric.budish/re…
First, ASIC-resistant blockchains are vulnerable to sabotage, because they allow the attacker to repurpose mining chips after using them for an attack. So the cost to the attacker is temporary and small. With ASICs, in contrast, the attacker’s chips are useless after sabotage.
This has been observed at least as far back as 2014 freedom-to-tinker.com/2014/12/12/why… Incidentally, the fact that so many blockchain projects are developing/deploying ASIC resistance without acknowledging this basic flaw doesn’t inspire confidence.
The paper also speculates about a future where repurposable chips are almost as efficient as ASICs, but this seems like a stretch to me. There’s also a section about a 51% double-spend attacker, which I didn’t find as compelling as the analysis of sabotage.
Collapse scenario #2 arises from the fact that even the ASIC investment for sabotage isn’t that costly — “only” $1.5Bn – $2Bn for Bitcoin today. This is well known, but there's the practical challenge of one entity acquiring 51% of Bitcoin’s mining power.
It also ignores the possibility that in response to an attempted sabotage, cryptocurrency holders might attempt to protect their investment by mining at a loss, presumably leading to a war of attrition with the attacker. This is fun to think about!
Collapse scenario #3 is by far the most interesting. Suppose ASIC improvement plateaus in the future, and the latest chips are only 20% more efficient than the previous generation. That’s still enough to make the latter uncompetitive, and the attacker can pick them up cheaply.
In this scenario, acquiring 51% hashpower for sabotage is far cheaper than acquiring 51% hashpower for mining profitably/at equilibrium. This is a nice observation and I’m not sure it’s been discussed widely before.
(To be clear, the three collapse scenarios in the order I discussed them above roughly correspond to #2, #3, and #1 respectively in Section 3.3 of the paper. This is my own interpretation of the paper's arguments.)
The paper says that if Bitcoin ever becomes digital gold — a significant store of value — then there will be incentives for sabotage, and suggests that therefore Bitcoin can’t/won’t become economically important akin to gold. Overall, a thought-provoking read.
