Russell wrote this in a few months in 1912 for the Home University Library, a series of cheap introductions sold at a shilling. He called it his shilling shocker. It is a hundred and seventy pages, it is the best first book of philosophy in English, and it has not been displaced in over a century.

What reading it is like

Easy, which for philosophy is rare, and not simplified, which is rarer.

Russell's prose is the best in the subject — short sentences, no jargon, no throat-clearing — and the book moves. It opens with a table and asks what color it is, and by chapter three the existence of the external world is genuinely in question. Nothing is dumbed down; the arguments are the real ones, compressed.

Read it front to back. It is a sequence rather than a collection, each chapter built on the last, and it is short enough that there is no case for skipping.

The last chapter is the one people remember. "The Value of Philosophy" is four pages, it is not an argument so much as a statement of what the activity is for, and it is quoted more than anything else Russell wrote.

The argument, in order

The table. It opens by asking what we know about an ordinary table. The color depends on the light and on where you stand. The shape depends on the angle. The texture depends on whether you use a microscope. None of what the senses deliver is a property of the table simply; all of it is a relation between the table and an observer.

So Russell distinguishes sense-data — the colors, shapes and sounds immediately present to us — from the physical object we take to cause them. And the immediate consequence is that we never perceive the object. We perceive the data and infer the rest.

Why believe there is a table at all. His answer is not proof but the best explanation: the hypothesis of a persisting physical object accounts for the regularity and the agreement between observers far more simply than any alternative. It is an inference, it is defeasible, and he says so.

Acquaintance and description. The distinction he is proudest of. Some things we know directly — our own sense-data, our memories, ourselves in some degree. Everything else we know only by description, as the thing that stands in such-and-such a relation to things we are acquainted with. His rule: every proposition we can understand must be composed entirely of constituents we are acquainted with. It is the theory of descriptions doing epistemological work.

Induction. Chapter six is the clearest short statement of the problem in the literature, and it contains the chicken. The chicken is fed every morning and forms a reasonable expectation, and then one morning its neck is wrung. No amount of past regularity entails the next case. Russell concludes that the principle of induction cannot be proved from experience and must be accepted on its own account or not at all. See the problem of induction.

Universals. Chapters nine and ten argue that relations — being to the left of, being before — are real and are not reducible to particulars. It is the most technical stretch and the most dated.

The value of philosophy. Not that it delivers answers; he is explicit that it mostly does not. The value is in the questions, which enlarge what we conceive as possible and free us from the tyranny of custom. The phrase is that philosophy is to be studied for the sake of the questions themselves.

Key passages

Chapter I — the table; appearance and reality.

Chapter II — the argument for the existence of matter as the best explanation.

Chapter V — knowledge by acquaintance and knowledge by description; the principle of acquaintance.

Chapter VI — induction; the chicken.

Chapter IX–X — the world of universals; relations as real.

Chapter XIV — the limits of philosophical knowledge.

Chapter XV — the value of philosophy. The four pages.

What is contested

Sense-data. The central apparatus of the book is a theory almost nobody now holds. The objection that finished it is that positing an inner item we directly perceive explains nothing and adds a second problem — how the item relates to the object — to the one it was solving. Direct realism and adverbial theories both grew out of rejecting it.

Whether the inference to matter works. Russell argues that the physical object is the simplest explanation of the regularities in sense-data. Critics reply that simplicity is not evidence, and that the argument concedes the skeptical premise in its first move.

How much Russell still held. He revised continuously, and by The Analysis of Mind in 1921 he had moved to a neutral monism in which the sense-data picture of this book is gone. Reading it as his settled view is a mistake the book's popularity encourages.

Whether it dates. The universals chapters do. The first six chapters do not, and are still assigned.