I used to be confused about the "nonzero sum game" term from game theory. I thought it's a confusing way to say "positive sum." But I've since read more (particularly reviewing Schelling) and thought about it more and realized that actually "zero sum" is truly the special case, and that negative sum games have a lot in common with positive sum games. And zero sum is really the exception.

In a negative sum game you often have strong structural incentives and reasons to cooperate (even if the best form of "cooperation" in practice is ignoring the other person completely^[1]). Canonical examples include war, gambling at a casino, and insulting ppl on the internet.

So there's an important sense in which good deals and "gains from trade" are possible, even if technically both sides might be better off if the other side initially didn't exist. But still the threat of conflict can allow you to eke out more concessions.

Structurally quite similar to positive-sum games, actually.

This is structurally very different from a zero-sum game, where my benefit comes entirely at your expense. No trades are possible, assuming both sides are playing optimally.

1. ^{^}

   cf War Games "the only winning move is not to play", though the tic-tac-toe example they gave is a zero-sum game.