Learn

Critical Thinking & First Principles

Taking a claim apart: what it actually says, what would have to be true, and what the numbers hide.

7 chapters · all free during the beta · ~98 min total

Syllabus in review — not final. These chapters are outlined, not written. Their checkpoint scenarios are live now, and they are the part you can actually train on.

  1. 1
    What Is Being Claimed

    Before asking whether a claim is true, find out what it says. Three questions do most of the work: what the number counts, what would show the claim wrong, and whether its "because" is a mechanism or only a name.

    Free
  2. 2
    Check What the Decision Rests On

    A plausible report now costs minutes to produce and as long as ever to check, so nobody can check all of it. Find the few claims the decision rests on, check what their sources say and who is saying it, and tell the room what you checked and what you didn't.

    Free
  3. 3
    What Is This a Case Of?

    Every piece of evidence reached you through a process that chose what you would see. Ask how often things of this kind turn out this way, ask who is missing from the sample, and ask how many out of how many before one vivid story becomes a trend.

    Free
  4. 4
    Did This Cause That?

    Why two things moving together have four explanations, how to name the confounder instead of reciting "correlation isn't causation", and what comparison would actually show that a change caused a result.

    Free in beta
  5. 5
    How Sure, and Says Who

    How to state confidence as a number someone can score, carry it honestly from the weakest input to the summary, and read a risk figure as a claim with an author: when the official number and the working estimate are far apart, the gap is the finding.

    Free in beta
  6. 6
    Arguing Against Yourself

    The checks you run on other people's reasoning, turned on your own: tests that catch you defending instead of looking, the strongest version of the case you're rejecting, and the one result that would prove you wrong.

    Free in beta
  7. 7
    Fences and Anomalies

    Two mirror-image mistakes in a system that works: removing a rule because nobody can say why it is there, and accepting a fault because it hasn't hurt anyone yet. Both act on something nobody understands; the fix for both is to find out, in proportion to what is at stake.

    Free in beta

Skip ahead and practise this domain →