I skimmed over the original article (Burson et al '06) to decode "noise and bias model": The authors found out that on easy tasks less skilled people overestim

I skimmed over the original article (Burson et al '06) to decode "noise and bias model": The authors found out that on easy tasks less skilled people overestimated their performance while more skilled people estimated correctly that they did ok. On hard tasks more skilled people underestimated their performance, while bad performers correctly estimated that they performed poorly. In other words, on easy tasks everybody thought they were doing ok, on hard tasks everybody thought they were doing badly. Thus it seems nobody did any metacognition; people of all skill levels just assessed task difficulty. On easy tasks "smart" people happened to be correct, on difficult tasks it was "dumb" people turn.