And it is, Bitcoin uses carefully aligned incentives to work around the problem.

Your paper is fundamentally flawed because it addresses nothing resembling the current dynamics at stake in Bitcoin. More precisely it ignores the incentives for miners to centralize (as they have shown to have) to mitigate propagation times. In effect your paper clearly demonstrates it is more profitable to do so under free-floating blocks and you essentially rely on their altruism to maintain the validity of your model to make decisions going forward. In short, your work might be sound from a technical standpoint, in a vacuum, but can not be used to construct security models that depend on worst-behaviours assumptions.

I cannot accept that you could comment on Greg's game theory intelligence when you continue to wave away the clear negative externality present in your models. You chose to ignore the obvious tragedy of commons at stake and that is why your opinion can never be considered as long as you don't recognize the cost externalized to node and the overall centralization pressure suggested by YOUR alignment of the incentives.