Tipping points in our climate predictions are both wildly dramatic and wildly uncertain. Can mathematicians make them useful?