Looks to me like the author also fell for the Dunning-Kruger effect.
They proved, if a person has no idea about their performance and randomly judges their performance, they will overestimate their performance if it is really bad.
And that's not something anyone had to prove. If the performance is <10%, and the estimation is randomly between 0-100%, of course, on average, the estimate is higher than the performance.
But what Dunning-Kruger was about is not that people judge their performance randomly, but that people judge their performance wrong.
The naive assumption is "performance = estimation", and Dunning-Kruger evidences that "performance != estimation", with a bias towards the center. Underperformers overestimate, overperformers underestimate.
And it's not about "if people can't estimate their performance at all, this is the result", but it's about "people are bad at estimating their performance". That's the core learning of Dunning-Kruger.