http://reasonableapproximation.net/2015/05/05/farmers-dilemma.html

Suppose you and I are farmers, owning adjacent fields. One day you have a brilliant idea. If we dig a ditch from the nearby river, between our fields, then irrigating our fields becomes a lot less work. It would cost two utils to dig the ditch - one utilon each - and we'd get five utils each from its existence.

You come to me with this suggestion. "That sounds great," I say, "but I don't think I'm going to bother."

You object that I'm being dumb not to take those four utils.

"No," I say, "I just think that if I don't help you, you'll do all the work yourself. You still get three utils if you dig it alone, so you'll do it even if I don't help you. And by not helping, I get five utils instead of four. Why would I pay a utilon to help you?"

(Unfortunately for you, I am a member of H. economicus and have a natural immunity to arguments about "fairness" and "not being a douchebag".)

The farmer's dilemma is game-theoretically equivalent to chicken. Both of us choose to either cooperate by digging the ditch ("swerve" in chicken), or defect by sitting at home ("straight" in chicken). If both of us cooperate ("C/C"), we both get an okay result. If both of us defect ("D/D"), we both get a terrible result. If one of us cooperates while the other defects ("C/D"), then the defector gets the best possible result for themselves, and the cooperator gets a result between C/C and D/D.

If you're cooperating and I'm defecting (or vice versa), then neither of us have any incentive to change our strategies. I could start to cooperate, but then I'd just be giving you utility. You'd like that, but I wouldn't. And you could start to defect, but then you'd be throwing away utility. Neither of us would like that.

On the other hand, if we're both cooperating, then we both have an incentive to defect, as long as the other doesn't do the same; and if we're both defecting, we both have an incentive to cooperate, as long as the other doesn't do the same.

(Formally, there are two Nash equilibria, at C/D and at D/C. This distinguishes it from the prisoner's dilemma, which has an equilibrium at D/D.)