Here is a paper modelling a contentious soft fork:
The fellow used warnet (a signet-like model bitcoin network for simulating the interaction of full nodes and miners)
Results:
To be honest, I had a difficult time making sense of these results. Perhaps it is presented for someone who has a better understanding of statistics than I (or perhaps I didn't wrestle with it enough). I'd be curious to hear if others draw any clear conclusions from this.
He also made an interactive fork explorer which you can find here: https://pfoytik.github.io/fork-explorer/index.html
Conclusions:
I believe these conclusions line up with what @optimism was expressing here: #1524712
Assumptions:
Perhaps I misunderstood this, but I think that a all or nothing outcome is more likely than not. A fork coin does not begin de facto with the same utility as the original coin. It only has as much utility as the rate at which merchants are willing to accept it. And I'm not sure his economic modelling accurately captures this (or it is possible that he models it at a rate much higher than I expect of any fork).
I feel the need to have a look at Voskuil's Fragmentation Principle again...
Confirm that that seems to be close to what I meant, but note that I only did that to entertain the notion that there is a situation in which to be mindful of on which chain your tx lives (and then honestly, really only for LN channels). I didn't simulate it but simply took the range (ultimately from https://arxiv.org/abs/1311.0243) that if the tips stay close to each other, 33-66% is committed to one fork side at all costs and 33-66% is committed to the other fork side, also at all costs, then there is a 0-33% danger zone where an impartial middle can keep both chaintips alive.
I'm not sure how the pools-of-pools that are the ultimate guarantor of FPPS payouts relate nowadays. Are there still 2 camps and how are they sized? I don't have recent facts and I don't have recent insights into pool operator relations either. My facts are dated, so I don't know. Example unknown: what's the Ant/F2/Spider/Via relation like these days and how many miners are going to leave these pools come flag day?
Note that I don't see any evidence that there is such a scenario playing out, but since I cannot with a straight face say that I know what's going to happen, I don't want to dismiss it either.
Yeah, read that again. I think that Voskuil argues the opposite - as long as there is conviction.
The only thing that can kill either chaintip is when no one mines it anymore (= lack of conviction.) Taking into account #1524412, it sounds like fork conviction is waning, but:
So time will have to tell.
I find Voskuil's writing difficult to think through, and do believe I managed to get the Fragmentation Principle backwards.
I'm trying to get some of the miners to do AMAs in the next few weeks. I believe we will have Luxor's CEO at the end of July, and I've reached out to Antpool's CEO (no idea if I'll get a response). I'd be very curious how they talk about any of this (if they even want to talk about it).