[snarf21]

I really enjoyed it to but agree with GP that it falls down. Initially, he sets up a system around our pattern matching, biasing, etc. but as the book progresses I felt it got more and more hand wavy. Lots of the examples didn't really fit the plot of his system but he implies that the system still holds. Also, lots of the studies like the marshmallow test have failed to stand up. We take a single study of proof when it is usually no more than an indication. The conclusions are really just a data point. Perform the same test 100 times with different scientists and participants and look at the data again. Instead we take one test and cut the data to fit and then publish.