> but none of them make a case that the current subsidy level is necessary.I wasn't actually arguing that, yes there are a bunch of unknowns. I'm only making the argument that there is a clear formula connecting mining rewards, potential double-spend profits and the cost of attacking a PoW coin. I don't care enough to actually derive the formula and plug numbers in it to answer that question.
> Recipients can adapt to the behavior by requiring more confirmations, restricting who and how they attack
More confirmations don't really work as a solution. The cost to hold a double-spend open is known, so we could try requiring enough confirmations to make such a double-spend unprofitable, but the attacker can execute multiple double-spends in parallel to spread their costs.
For large transactions (relative to the hashrate of that coin), the confirmations can can get quite high, which destroys the usefulness of cryptocurrency for fast and easy transactions.
For one of the "random shitcoins" I was tracking, I found the centralised exchange I suspect the attacker was targeting (presumably trading them to another currency and then withdrawing). The exchange had set really high numbers of confirmations (2-5 days worth of blocks) but the because these coins were so cheap to attack, the attacker had no problem doing reorgs that large.
And then the exchange delisted the coin due to "51% attacks".
> The total value of activity on the chain isn't equivalent to the value available to attackers in an attack, especially to the extent that attacking undermines the value of the system as a whole.
Yes, another thing to go into that formula.
Centralised exchanges and other automated systems interacting with cryptocurrencies really mess things up. They provide a lot of liquidity open to double-spend attacks.
In a world were cryptocurrencies were only used as the peer-to-peer cash in the way a lot of it's early proponents advocated for (and still do), where everyone can manually verify their transactions and establish trust, then your suggestions do work somewhat better.
But in the highly centralised nightmare that cryptocurrencies have become... well it changes numbers going into the formula.
> That's also not clear to me, there are an awful lot of idle two generation old miners out there at any given time.
When hashrate is trending up, I suspect the two generation old miners are slow enough to not worry about.
But the economics might change as block reward subsidies drop, or if the potential profit of double-spends goes up.
> Perhaps it isn't but if you really strongly believe that there are about 1000 cryptocurrencies that you should be figuring out how to short hard instead of posting on HN. :P
I mean, I considered it. Hence the research.
But these were really, really, shit shitcoins, not listed on exchanges with futures markets, so the only real way to profit was to execute the 51% attacks myself, and I didn't want to go down the black hat path.
Also, I moved onto other hobbies.