If correct answers take more work to enter, then I expect there to be fewer correct answers.
There was a deepfake experiment on the site recently asking people to use a slider to estimate their confidence that a particular video, audio or text clip was faked. That used a slider between -100 confidence and 100 confidence, and I really sincerely doubt that anyone spend any amount of time thinking about what number they entered. It was just a way to introduce noise.
uncertainty-based rescaling — leads people to systematically overestimate the size of small values and underestimate the size of large values.
Basically, when you're not sure what the actual number is, you tend to anchor your guess on 50%, not 0% or 100%.
It’s just people going “probably a small number”, being asked to quantify that, and throwing out 20-30% because that feels like one.