To quote from the article:

“ If none of them knows what the other is doing, they will sell each other out. … say ‘“’Then I will be running against you.‘”’ Or if not you, say you’ll find someone who will. Tell your Congressperson that you are a one-issue voter, and that issue is chucking the madman from the White House. Now, who knows? Maybe you won’t pull your support. But they don’t know what you’ll do, so they have to act. It’s just basic game theory.’

He’s clearly equating the prisoner’s dilemma with uncertainty, and this is objectively wrong. You write:

“ However, the ambiguity in this article’s context has to do with a coalition being broken up, so that players are suddenly in the position where they can no longer rely on each other’s actions. That is the case that takes this coalition from the known, reliable, cooperative outcome and spirals them into the Prisoner’s Dilemma.”

This sounds like a coordination game with multiple equillibria, which is not a PD. You’re right that the PD illustrates why coercive institutions (obligations) can make players better off, but if you’re talking about coalitions being necessary to sustain those institutions then you’re now talking about some larger model.

This entire article repeatedly abuses game theoretic terminology, and it’s quite frankly infuriating to those of us who study politics using game theory for a living. This may come off as nitpicking, but it’s really not — game theory is a branch of mathematics, and we use it precisely because we want to make rigorous, deductively closed arguments. We do not throw around statements like “Hillary Clinton was our stable equilibrium” (that’s not what equilibrium means) or “new strategies are the only things that disrupt stable equilibriums.” Equillibria are defined with respect strategy sets, and there’s no concept of a “new strategy” in game theory. Either a game is one-period, or the evolution of available actions is part of the game, and a “strategy” is actually a function that gives an action for every period contingent on the history of the game up until that point. (And thus an equilibrium is defined for the entire game.)