travelin

Have started skipping pages and chapters, seeking the really intriguing insights, which seem, most often, to be research other than that of Dr. Kahneman. Dr. Kahneman can devote most of a chapter to how he saved standardized tests for the Israeli military in the 1960´s. I think modesty is a big problem for him.

Also, while it´s hard to imagine an academic community once utterly in thrall of perfect rationality, game theory, etc., the fact that mankind is not rational by rigid standards of logic now feels like a truism, and listing traps in statistical judgment that students fall into doesn´t make working rationality any more or less true. I sense that Dr. Kahneman also knows this.

Like a book about the human brain I was picking through at roughly the same time, this is really a series of short papers written over quite a long time. It's likely that they contradict each other and somehow convey the illusion of complexity in the process.

Is this the place to talk about problems with statistics as a science? Because Dr. Kahneman loves statistics, and wherever people fail at statitistics, Dr. Kahneman considers them irrational. This is self-serving, because he takes pains to point out that writers of statistics textbooks make basic statistical errors. He himself says he forgets to use large enough samples in studies sometimes.

I attended a university which dedicated entire courses to the soft science of statistical failure, and then required I take those courses. One absolutely soporific professor Dr. Kahneman praises a great deal stood in front of my class and "proved" that statistically there is no reason to vote. Another "proved" that attacking Iraq was George Bush's only option in the Gulf War.

Dr. Kahneman likes to say that a few simple measures are better than expert opinion. This is self-serving again, because results are measured by the same simple metrics. So, for example, experts are asked to predict the likelihood of a very limited number of distinctly defined options, multiple choice style. Or, in an example such as that of predicting the market for wine, the same growing conditions which are used to grade the quality of wine are used to predict its market value in the growing season. What a shock.

This review is in danger of becoming as diffuse and confusing as the book. As another review points out, the book is fast and loose with statistics itself and essentially plays all sides of the field at once. I did browse some of the academic papers in the appendix and they weren't remarkable more consistent. It would be interesting to see the peer review process for those papers.

Without having experienced the stifling milieu of absolute rationality in social sciences, I imagine we should be grateful for the comparative freedom Dr. Kahneman gave some students in the 50s.

Still, per my update, the concluding pages of the book nearly sound like fascism. If statistics is the magical order of the universe, only to be discovered and worshipped by the select elite, well, the Dr. tells us that governments and companies, which are magically immune from (statistical) failures should be doing all of our thinking for us.