Testing confirmation screens for clarity was the wrong test

Building StareBrain — say a command, see exactly what it's about to do, confirm before it fires.

A conversation today about an unrelated product — an AI course-building tool with an approval gate — surfaced a mistake I've been making in StareBrain's own confirmation screens. I've been testing them for legibility: can someone read the confirmation in two seconds. That tells you nothing about whether the screen actually works. A screen can be perfectly clear and still let a mistake straight through, because clarity and error-detection are different things that happen to look similar.

The better test, borrowed directly from that thread: deliberately seed known-wrong states — a stale calendar slot, a wrong contact, a mismatched time — and measure whether a real user actually catches it before confirming. Not "does this feel clear," but "does this catch the thing that's wrong."

Rebuilding my own testing process around that before shipping any more confirmation screens.

Building in public as I go — waitlist link in profile.