Most arrivals came off a link in someone else's post, landed on the sample run, and left from it. The index we spent longest on was reached by a minority. The work was met sideways, at the one screen we had argued about most.
The wrong turn read as honesty. It was the passage people quoted when they linked the page, which settles what Lab 01 left open.
The failure state from Lab 04 never appeared in the wild — the condition that fires it is rarer than we modelled, so it shipped tested by nobody but us. A whole branch of the interface has no evidence behind it.
The sample run became the entry page and the index behind it was cut. One decision from Lab 03 reversed: the second navigation came back, scoped to the run, because people arriving mid-artifact had no way up.