Nassim Nicholas Taleb February 07, 2014 Friends, this subject is for discussion, with back-up if you can: It seems to me that IQ tests favor turkeys. Standard tests of “fluid intelligence” that require the subject to complete a sequence favor a certain class of people who can rapidly detect naive patterns, and penalize those who are natural skeptics with richer imagination. In real life patterns are more complicated and having an ingrained skepticism that slows down inference is an invaluable asset. So my speculation is that it is OK to do well, but not to do very well. Consider the seemingly elementary sequence: a-b-a-b-a -?- [complete ]. Naive pattern matching would give [b] as solution. But in real life ecologies the sequence could have a more complex pattern, a-b-a-b-a-b-b (there is a repetition of the 6th letter) or meta-patterns to consider. These take time to examine and someone smart would need to fight to repress his imagination. So those who do well, but not great, should be much smarter than those who do better. Let us debate. Does it make sense?
