The Dunning-Kruger effect states that incompetent people overestimate their competence. You've heard of that before, probably. Turns out, there is a major error in the statistics: the effect is actually nothing more than an autocorrelation. If we measure the stuff correctly, the effect disappears.

A month after reading this article, I learned the same thing at the university.

Press keys D and F. After some time, the site will predict your next presses. I get accuracy 73 %. There is no such thing as human-generated randomness.