About

The Evaluating Self is a reader's edition of the sources behind one question: what a person has to keep doing, and keep being able to do, when a machine can produce the work. It is built from eighty-seven sources, from the Upanishads and the Pali suttas to papers published in 2025, by way of the psychology of judgement, the study of expertise and automation, the design of everyday things, the philosophy of mind and the papers that describe how language models are trained and scored.

Each source is rewritten as a chapter that follows the author's own order, with hand-drawn figures for the mechanisms the argument turns on. The chapters are grouped by the question each source answers: how to get good output, how to know it is right, what to keep and what to hand off, and who is doing the work. The closing chapter compiles the argument the sources make together.

Around the chapters sits the apparatus that holds them to account. Every chapter yields a set of one-sentence claims, each marked as evidence, opinion or contested, with what it rests on, what limits it, and what it means once the delegate is a model. Where two sources from different traditions turn out to be making the same claim, a pairing records the shared claim, the evidence on each side, and the point where the analogy breaks. The claims are gathered into theses, and the theses into the rules on the Framework page, which are recomputed whenever a claim is withdrawn or a source is added. The timeline below places every source at the time it was written, in the lane of the question it answers.