Mathematics books normally have an excellent signal to noise ratio. The difficulty is finding ones that are both important and not too difficult or specialized

Mathematics books normally have an excellent signal to noise ratio. The difficulty is finding ones that are both important and not too difficult or specialized; textbooks are usually a good starting point. I can recommend Feynman's lectures on physics and Bruno de Finetti's "Theory of Probability. A Critical Introductory Treatment".