Let me lay out the argument from our post in a more technical way: the biggest problem with the Eyal/Sirer paper is that they don't think about the problem as an equilibrium problem, but rather argue about what's best from the perspective of a particular player. This leads them to propose a strategy which is not even optimal for any player (we prefer to think of Bitcoin as a kind of consensus game, in the game theoretic sense. See our earlier paper on the topic [1]).
They argue that Bitcoin is not incentive-compatible by virtue of the strategy they demonstrate. I think this question needs to be the crux of any Bitcoin research paper. I'll define "incentive compatible" to mean one of two things
(1) (weakly incentive compatible) If people follow their incentives, rather than the rules of Bitcoin as written down and understood by the community, then there exists an equilibrium in which all players follow the rules.
(2) (strongly incentive compatible) The above equilibrium is the only equilibrium in Bitcoin.
The question of whether the current Bitcoin ruleset is incentive compatible strikes me as the most important Bitcoin research question: it answers whether Bitcoin, as a system, will continue to be stable over the long term. A secondary question is to ask "what rule sets could exist which would be incentive compatible?" Obviously, if the answer to the first question is "no" then the second question is more important.