Collective action problems vs. coordination problems: the difference is not information, it's interest

A collective action problem is a situation where every member of a group would be better off if everyone did A than if no one did A — but the parties' conflicting views of their own interests risk producing an outcome worse for everyone than universal A. The Stanford Encyclopedia of Philosophy's entry on the free rider problem states the definition carefully, because the careful version is where the interesting parts live.

The entry draws a sharp line between this and a coordination problem. In a coordination problem, a mutually advantageous outcome — the example given is meeting at a rendezvous point — can be obstructed by a lack of information rather than by conflicting interests. Two people who both want to meet but never named a place have a coordination problem. Two people who each want the other to show up first have a collective action problem. The remedy differs accordingly: publish the location, versus change the payoffs.

The definition is deliberately broad in ways worth noticing. It allows that a collective action problem can arise when everyone's interests genuinely are best served by doing A, yet many people falsely believe they are individually better off not doing A. It also allows different ways of specifying what counts as better or worse for an individual. And the entry notes one consequence of the coordination/collective-action distinction: it precludes identifying interests with preference-satisfaction. If interests were just whatever people prefer, the category of false belief about one's own interest would have nowhere to sit — and neither, arguably, would the problem.