Does the title apply to the book? Probably Approximately Correct by Harvard computer scientist Leslie Valiant suggests some interesting, important ideas for deciding and learning in real-world situations:

• being mostly right is good enough, if it's in time to take action
• learning a pattern requires a relatively stable environment and discernible differences

Prof Valiant explores these and related themes of computational complexity and machine learning in this 2013 popularization. But the prose is uninspired, the technical detail is skimpy, and the hubris is distracting ("I believe ...", "I call ...", "I expect ...", etc.). Applications to Darwinian evolution by natural selection appear important but are confusingly presented. And the clunky coined term, ecorithm, for the core thesis? Perhaps Valiant and his topic are far deeper than comes across. Maybe an assertive editor or coauthor could have helped.

^z - 2016-03-21