Cognitive Updating Test

Two of the six jobs in the Cognitive Updating Test were free marks

This test runs six jobs at once and tells you afterwards which one you let slip. Going back through every run anybody has saved, two of the six were being cleared by almost everyone: sensor sweeps at 99% and data entry at 98%. The camera job, by comparison, sat at 49%.

They were not easy because you are good at them. A data entry order could be answered for 165 seconds — nearly three minutes — while the camera gives you fifty. A sensor sweep, which is a prompt appearing and you pressing the matching button, could be answered for the best part of a minute. A job nobody fails is not measuring anything; it is handing out marks and making the test look harder than it is.

Data entry now allows 90 seconds, and the sweeps 38 and 45. All three are still comfortably doable if you are actually watching the panel, which is the only thing those jobs ask.

Nothing else about the test changed. Same six jobs, same controls, same scoring.

This moved scores. This makes the test genuinely harder, so expect to score lower than you did before, and your earlier scores are kept and shown separately rather than compared. We have left the target where it is rather than guessing how much harder it has become: our simulated operator has perfect attention and clears the new windows without noticing them, so it cannot tell us what they cost a person. The target gets corrected once enough of you have sat it.

CUT and the Dynamic Projection Test were far too easy to max out

Both were marked against what a simulated player scores, set below it on the assumption that a computer steering everything perfectly is out of reach. It is not. Of the nine people who have sat each, the best candidate on CUT outscores that simulation and the typical one on the Dynamic Projection Test beats it outright.

So the bar was set below almost everybody, and six of nine on CUT and eight of nine on the Dynamic Projection Test finished on the maximum of 150%. A score everyone gets tells a strong candidate and a struggling one exactly the same thing, and makes re-sitting pointless.

Both bars now come from what candidates actually score. CUT moves from 305 to 520 and the Dynamic Projection Test from 50 to 80. A typical run on either now reads around 100%.

This moved scores. Your CUT and Dynamic Projection Test scores will read lower than before, around 100% instead of 150%. Nobody got worse. Runs sat against the old bar are kept and shown separately rather than compared with new ones, because they were measured against a bar that was wrong.