The Art of Thinking Clearly

Ninety-nine short chapters, each naming a bias or fallacy, illustrating it with a scenario, and offering a brief corrective. There is no through-line, no chapter order that matters, and no theory connecting them.

Dobelli is explicit that this is a compilation rather than original work, and the book is best understood as an index to a literature rather than a contribution to it.

Ninety-nine entries with no connecting argument. It is an index to a literature, and says so.

Devmark’s reading — not a quotation from the book

The well-established ones are handled cleanly: survivorship bias, confirmation bias, sunk cost fallacy, availability, anchoring, base-rate neglect, hindsight bias and the illusion of control.

For these, the three-page treatment is adequate — the concept is simple, the example is apt, and the corrective is actionable. This is the book working as intended.

His opening chapter is the book's best and does more work than the length suggests: the systematic invisibility of failures means any inference drawn from visible successes is drawn from a sample selected on the outcome.

He applies it to authors, musicians, entrepreneurs and investment strategies, and the accumulation makes the point more forcefully than any single case would.

His organising philosophy, borrowed openly from Taleb and Munger: avoiding stupidity is easier and more reliable than achieving brilliance, so a list of errors to avoid is more usable than a list of virtues to acquire.

This is the closest thing the book has to an argument, and it justifies the format — a catalogue of ways to be wrong is exactly what that philosophy calls for.

This must be stated. The book presents ninety-nine effects at uniform confidence, and they are not uniformly supported. Several rest on studies from areas hit hard by the replication crisis, and Dobelli provides no indication of which findings are robust and which are single studies.

A reader has no way to tell from the text whether an entry describes a well-replicated phenomenon or a striking result that did not survive.

Ninety-nine effects presented at identical confidence. They are not equally supported, and the book gives no way to tell.

Devmark’s reading — not a quotation from the book

Worth knowing. Nassim Taleb publicly accused Dobelli of drawing on his work without adequate attribution, and there was a subsequent dispute about sourcing more broadly.

The material genuinely is derivative by design — Dobelli says so — but the attribution in the original German edition was thinner than the borrowing warranted, and readers should know the book is a compilation of other people's research.

Several entries describe substantially the same underlying phenomenon under different names, which inflates the headline count without adding any content a reader can use.

Anyone who knows the underlying literature will notice that ninety-nine distinct errors is a generous accounting, and that perhaps forty genuinely distinct mechanisms are being described across the ninety-nine chapters. The number is doing marketing work rather than analytical work.

Against Kahneman it is far shallower and far easier — a reference card versus a research programme. Against Cialdini it covers more ground with less depth and no applied method.

Its genuine advantage is format: three pages is short enough to read one entry while waiting for something, which is how most people actually use it.

Useful for someone who wants the vocabulary quickly and will follow up elsewhere on anything that matters to them. Useful as a reference to check a decision against.

Poorly suited to anyone who wants to understand why these effects occur, or who will treat the entries as established science without further checking.

Read it as a list of hypotheses about your own reasoning rather than a list of facts. The entries are prompts to look more closely, not conclusions.

The most valuable habit it can install is the survivorship question — who is not in this sample — which appears first for good reason and is worth more than the ninety-eight that follow.