@joeposaurus this article is fucking dumb. They replicate the DK experiment and say the imp part from that experiment is the difference between real test and percceived score so they plot y-x against x. Then do random plot graph and get the same y-x plot. Not shit, Sherlock. In both cases that's just the graph of y = -x + C. The shit they specifically were pointing out is compared to percentile. There's a reason it was in the x-axis in DK. Plotting y = -x is shit we did in maths lessons at 16

@joeposaurus TL:DR essentially in this replicated experiment, they added an autocorrelation themselves and then were surprised when the results they obtained were due to autocorrelation