Digna Legi / Watch · 17 October 2026 · Selection 02

Five ideas you can put to work.

A slide that proves its claim, a tool that reveals a hidden path, a visual probability rule, a reason for readers to care and a way to score confidence. Choose the problem you are working on.

If you watch only one

Make the slide prove one sentence from 14:53 ↗ In under three minutes, Alley shows how a sentence headline determines which evidence belongs on a slide. You can test the move on your next presentation.

01Underseen

Communication · Science Communication Lab · Full runtime 25:56

Make the slide prove one sentence

Michael Alley · Science Communication Lab

The idea A sentence headline gives you a test for deciding which evidence belongs on a slide.

Why spend the time Alley turns an aquifer diagram into support for a specific claim. Try stating the conclusion of one slide in a sentence, then check whether every retained element helps establish it. The short passage demonstrates the first principle of his approach.

Reservation The supporting experiment changed several design features together; it does not isolate the effect of sentence headlines. This is guidance for spoken scientific presentations, not a rule for every document.

Suggested passage: 14:53–17:50

Watch from 14:53 ↗Full video ↗Source check ↗

Tools and design · Rui Oliveira · CUSEC recording · Full runtime 54:19

Inventing on Principle

Bret Victor · Rui Oliveira · CUSEC recording

The idea A live preview can still hide the behavior you need to understand. Replaying identical inputs and showing the trajectory reveals what a change actually does.

Why spend the time A small platform game makes the distinction visible: changing the bounce, replaying the same controls, then drawing the path. The four-minute demonstration offers a concrete criterion for evaluating a tool: can you inspect the behavior that matters?

Reservation The replay demonstrates one set of inputs, not correctness for every possible interaction. The old recording is soft; use a larger screen for the code.

Suggested passage: 10:26–14:40

Watch from 10:26 ↗Full video ↗Source check ↗

Probability · 3Blue1Brown · Full runtime 15:10

Bayes theorem, the geometry of changing beliefs

Grant Sanderson · 3Blue1Brown

The idea Evidence filters possibilities unevenly. The updated probability is your hypothesis’s share of the possibilities left standing.

Why spend the time Sanderson builds the rule from counts and areas. Sketch the evidence-compatible cases before trusting a conditional-probability claim; a low base rate can outweigh a suggestive description.

Reservation The example's personality likelihoods are assumptions, and the prior depends on how the person was sampled. The later Linda-problem aside is outside this passage and overstates how consistently a change of wording eliminates errors.

Suggested passage: 3:24–10:14

Watch from 3:24 ↗Full video ↗Source check ↗

Communication · UChicago Social Sciences · Full runtime 1:21:51

Make the problem matter before explaining it

Larry McEnerney · UChicago Social Sciences

The idea A gap in your knowledge is not automatically a problem your reader has a reason to solve.

Why spend the time McEnerney distinguishes an unresolved question from a consequential instability. In a proposal’s opening, identify whose understanding is disrupted and what changes if the problem remains unresolved. The passage develops a reader-value diagnosis rather than a sentence-polishing rule.

Reservation The examples assume academic expert communities, and the handouts are difficult to read on screen. Establishing relevance should not become manufacturing urgency or pretending every reader shares the same interests.

Suggested passage: 53:27–1:02:40

Watch from 53:27 ↗Full video ↗Source check ↗

Probability · The Royal Institution · Full runtime 53:52

The art and science of uncertainty

David Spiegelhalter · The Royal Institution

The idea Write down a probability before seeing the answer, then score the squared error. Confidence becomes a commitment you can check.

Why spend the time Spiegelhalter supplies the questions, confidence choices and scoring table. The exercise makes an overconfident wrong answer costly and shows why reporting your actual belief gives the best expected score.

Reservation Five trivia questions demonstrate the scoring method, not your forecasting ability or calibration. The wider lecture’s blanket claim that intervals are always too narrow is too strong.

Suggested passage: 15:23–23:05

Watch from 15:23 ↗Full video ↗Source check ↗