Through Time

Reasoning when nothing is clear

How to build and take apart arguments, weigh odds, keep score of your forecasts, not give in to a machine out of habit.

Machines will offer ready-made decisions more and more often, and our aim is that people stay able to judge and dispute them: that is an aim, not a forecast. You will learn to take arguments apart, work out odds in plain words, keep score of your forecasts and decisions, and see which famous biases held up and which did not. This course is about your own decisions; other people’s claims and data belong to the courses next door, and why a model agrees with you to the course on dialogue.

So far there is only this table of contents. Lessons appear one at a time, and we promise no dates. Feed of new lessons (Atom)

16 lessons, 5 parts

Part 1 An argument as work

Before you agree or object, you need to see what an argument is built from and where it is weak.

  1. What reasoning is for

    The disputed hypothesis of Mercier and Sperber: reason evolved for arguing, so alone we look for support and together we find each other’s errors.

    coming soon

  2. What an argument is made of

    Premise, conclusion and hidden assumption: how to write out an argument, yours or someone else’s, so that it shows what needs checking.

    coming soon

  3. Fallacies in a real argument

    Why naming a fallacy does not refute it, and why in life the same move can be an error or a reasonable point.

    coming soon

Part 2 Odds in plain words

Most decisions need three ideas about probability and no formulas.

  1. Where to start: the base rate

    How to start your own estimate from the base rate when a vivid story tempts you to forget it; the idea itself is in the course on numbers.

    coming soon

  2. Bayes without the formula

    What a new fact changes in what you already believe, and how to work it out with imagined people in place of percentages.

    coming soon

  3. The outside view

    Kahneman’s point: your own plan feels special, so ask how similar ones ended, and imagine it has already failed.

    coming soon

Part 3 Forecasts and keeping score

The accuracy of your estimates can be measured and improved, but only if you write them down before the outcome is known.

  1. Experts, the chimp and superforecasters

    How Tetlock tested famous experts, and what the strongest forecasters do: they split up a question and often revise.

    coming soon

  2. Calibrating yourself

    What “70 percent sure” means, and how to keep score of your estimates so you can see whether to trust them.

    coming soon

  3. A decision journal

    What to write before the outcome so you can later judge the decision, not the luck: sensible practice, not a proven method.

    coming soon

Part 4 Biases: what held up

The list of biases in popular books turned out to be uneven, and knowing which are real is more useful than knowing them all.

  1. What survived replication

    Which famous effects stood up to large replications, anchoring among them, and why you can lean on them.

    coming soon

  2. What fell apart

    Social priming, where Kahneman himself admitted too much faith in weak studies, and what that says about popular books.

    coming soon

  3. What is still argued over

    Nudging and loss aversion: why some find a clear effect and others almost none, and how to decide while the argument stays open.

    coming soon

Part 5 The last word

Our aim is that the last word stays with a person who can justify it,, not one who merely clicks “agree.”

  1. Automation bias

    People follow a system’s prompt and miss its errors, novices and specialists alike: how to catch yourself doing it.

    coming soon

  2. The opposite mistake

    A formula often beats a specialist’s judgment, yet people abandon an algorithm once they see it err: how not to mistake stubbornness for vigilance.

    coming soon

  3. How to test a model’s argument

    Write your decision down before the talk, ask the model to break it, note what changed and why: you borrow a mind and still decide for yourself.

    coming soon

  4. A look ahead: machines that forecast

    Models are already being tried as forecasters beside people: what is left for a person if this goes on is, to us, this course’s central question.

    coming soon