A Treatise on Probability

Keynes wrote this in 1921, but it feels urgently modern. He argues that probability isn't merely numerical, it is a logical relationship between what we know and what we believe, a measure of reasonable faith. Before this book, probability was mostly algebra; Keynes made it epistemology. He builds a framework where probability degrees are not always comparable, where evidence has "weight," where certainty is a spectrum rather than a binary. This was not just theory, it was a fundamental rethinking of how rational beings should navigate uncertainty. Its influence ripples through Bayesian statistics, decision theory, and now, remarkably, into artificial intelligence research. For anyone who has ever wondered what it really means to be "probably right", and why numbers alone cannot capture the complexity of reasonable belief, this remains the deepest book on the subject ever written.