It’s possible that the camel has two humps (pdf). We don’t have much evidence regarding the shapes of the subskill distributions. The connection between “rationality” and debiasing is questionable (correlation versus causality selection effects tag alongs bla bla bla). Beware motivated cognition on this subject.
No, I think he’s simply saying some subjects have more aptitude for the topic than others, and this distribution may be bimodal—explaining why some results are good and some are bad. None of these studies address whether the epistemic debiasing leads to improved or worsened instrumental results. (That particular idea, that improved epistemic rationality can lead to decreased instrumental rationality, would be the ‘Valley of Bad Rationality’.)
It’s possible that the camel has two humps (pdf). We don’t have much evidence regarding the shapes of the subskill distributions. The connection between “rationality” and debiasing is questionable (correlation versus causality selection effects tag alongs bla bla bla). Beware motivated cognition on this subject.
Speaking of the camel has two humps, replication has turned out to be difficult and the idea doesn’t look very good anymore: see first quote in http://www.gwern.net/Notes#the-camel-has-two-humps
Sweet, thank you.
Your notion being that being able to avoid bias might not give many practical benefits?
No, I think he’s simply saying some subjects have more aptitude for the topic than others, and this distribution may be bimodal—explaining why some results are good and some are bad. None of these studies address whether the epistemic debiasing leads to improved or worsened instrumental results. (That particular idea, that improved epistemic rationality can lead to decreased instrumental rationality, would be the ‘Valley of Bad Rationality’.)