You’re welcome, thanks for reading!
[…] integrating Schnorr for example, might increase the probability of these attacks, due to the increased privacy.
I would say so, yes — anything which a) strengthens privacy, and b) increases block capacity is good for these attacks.
That said, I am dissapointed that appearantly nobody with technical credibility has picked this up with a reply. I sure hope that does eventually happen.
Agreed. I’m looking for more technical people to get their eyes on this too and see if it can be falsified. It’s right on the edge of what has been studied / published in economics papers as far as I’m aware.
I found plenty of discussion about the future emergence of a transaction fee market to take over miner compensation when the block reward runs out (all of it pretty pessimistic — that’s its own problem!), but no-one had gone further into thinking about what an attack would actually be like in practice.
I discovered this attack while writing about exactly that subject, and thinking about how Dr Evil would go about double-spending one hundred… billion dollars, so (if it doesn’t get proven wrong) any credit belongs to Mike Myers.
Have you had any discussion with Hasufl on this? What has been the response so far?
Yes, I first mentioned this point to Hasu on twitter when he said he was writing a research paper. Then we had a long discussion the same day he posted that paper where at first I couldn’t understand how he was getting these gigantic numbers for how much an attack costs.
Eventually figured out, after talking back and forth for a while, that it came down to the assumptions he was making about hardware costs: Hasu’s model was saying that miners are (at all times) very heavily bought-in / committed to the price of BTC, such that a 51% attack would cost them billions in useless post-attack hardware.
I had reasoned that people would find ‘jumping off’ points in their CapEx cycles instead, as written in the ‘Objections’ section of this article. And earlier this week went on to realize that old/decommissioned equipment could pose a major problem:
