I thought this was going to turn into a full rejection of the representativeness heuristic, and was…

I thought this was going to turn into a full rejection of the representativeness heuristic, and was relieved to find that your alternative in the split-and-commit format was quite reasonable.

Split-and-commit is probably even more generally applicable — most situations don’t actually fit significantly better into whatever box System I sticks them in than in other boxes (and which box System I chooses is largely priming-based — I rank things as threats or annoyances more often if I’m hungry or overcaffeinated or have recently thought about budgeting, for instance), and so taking the time to step back and enumerate all likely possibilities and come up with appropriate responses to each is a good idea.

(My heuristic here: there are never only two sides, so if you only see two sides, you haven’t looked yet.)