When Does Being Selfish Pay Off? Temporal Composability and Profitability in Selfish Mining eprint.iacr.org
Selfish mining, now with a stopwatch: it studies how fast attacks turn profitable, not just whether they ever do.
4 comments
Do they mean the attack can become profitable after just a few blocks (or is this more like some expected-time calculation), because the stopwatch part feels like the bit that would matter in practice...
> after just a few blocks
No, that sounds too optimistic. The stopwatch angle is about how long it takes for the revenue gap to close, which is a very different question from “can I print money in 3 blocks?”
If anything, that time-to-profitability is the part people usually hand-wave away when they talk about selfish mining.
No, not a few blocks. They mean the attacker can run negative EV for a long stretch, then cross break-even once the accumulated revenue gap and difficulty retarget effects catch up, so the metric is more like an expected time to recoup losses than a short-run jackpot.
What surprised me a bit is that fee dynamics can move that crossover a lot, the abstract says incentive transactions can make some strategies hit profitability up to 15x faster at the same hash rate, which is a pretty big gap if you're thinking in terms of weeks not blocks.
Does that stopwatch still matter if the attack never gets close to a retarget window, or is the whole point that the revenue gap can linger long enough for difficulty adjustment to turn a barely-bad strategy into a profitable one (which feels like the part people skip over)...