The Target Recognition Test spent its first three minutes nearly empty
The four panels in this test are meant to arrive one after another over the first half minute, so you build up to juggling all four. They were arriving over the first three and a half minutes instead. The scene panel sat saying "Standing by" for two and a half minutes of a twenty-three minute test, and the system panel for over three.
It was not broken, it was mistimed. The arrival times were written as a share of the run back when this test was four minutes long. The test is now twenty-three minutes, the shares were never revisited, and everything stretched with it. All four panels now arrive inside the first thirty-five seconds, the way they were meant to.
The scene panel has changed too, and it is harder. The shapes are drawn a third smaller and there is see-through cloud drifting across the ground. That came from one of you asking for it — smaller targets, more scattered, harder to find, maybe some cloud — and from our own figures agreeing: you were finding 91% of what that panel asked for while the two panels next to it sat at 35% and 44%. It was the easy one pretending to be hard. The shapes are harder to spot; they are not harder to click, which is a different thing and not what this test measures.
Your earlier scores on this test are kept and shown separately rather than compared with anything you do now.
This moved scores. Two changes that pull in opposite directions: the panels arriving sooner gives you more to score from, and the smaller shapes under cloud make one panel harder. We can measure the first and not the second, so rather than guess we have left the target where it was. Expect your first run on this version to read differently in either direction, and expect the target to be corrected once enough of you have sat it.