The counterpoint from my colleague Ed Felten is based on a flawed understanding of how mining pools work and how pool participants are rewarded.

Explained here: http://hackingdistributed.com/2013/11/08/fairweather-mining/

The argument in that counterpoint seems to be as follows:

1. Assume that selfish mining doesn't work.

2. Because selfish mining doesn't work there will be fair weather miners who will only mine on whichever chain is furthest ahead, defaulting to the public chain in the case of a tie.

3. Since the selfish mining pool won't be ahead all the time nobody will mine for it.

4. Therefore selfish mining doesn't work.

It's not what I'd term a strong rebuttal.

I think that the "selfish mining" in your first point isn't truly selfish mining. The idea is that the particular method of selfish mining being discussed would be defeated by a truly selfish mining method. No matter what you want to act selfishly and the debate is whether or not their is an optimal selfish strategy that is bad for the overall bitcoin infrastructure. Since little to no incentive exists to stick with a mining pool that is selfishly mining it is my understanding that that the attack using a mining pool mentioned in the previous post will not work well.

That's a valid point, but not at all what I got out of the argument. I picked up primarily on the specific counter-strategy.

It's certainly true that selfish mining strategy A which gave a 5% increase in results would lose out to another selfish mining strategy B which gave 10% increases in results. However that's not an argument that strategy A doesn't break the system, merely that the optimal selfish strategy breaks the system at least badly as strategy A.

Now it is possible that there isn't a stable strategy to use. If we supposed for the purposes of argument that the fair weather strategy was more profitable than the selfish strategy described in the first article you posted, then it would seem that the optimal strategy would oscillate. As more people participate in a selfish pool consistently the more profitable it is to be a fair weather miner. However, supposing that leads to a dissolution of the selfish pool all those fair weather miners turn honest. Against honest miners, however, the selfish strategy is proven to be more profitable. And so you'd see an oscillation where people constantly shift between being honest, selfish and fair weather.

Of course, if the long run gains keeping up on this strategy treadmill is less than simply sticking to the selfish strategy through thick and thin, then perhaps the fair weather strategy isn't better in reality.