PPracticeFROM THE QUIET COST

A DAILY EXPLANATION CHALLENGE DAILY GAME

Spell It Out.

You know how. Can someone else follow?

TODAY’S CHALLENGE
01 Look closely02 Write the steps03 Find the gaps

The starting point

THE FINISH LINE

THE COST WE ARE PRACTICING AGAINST

When AI supplies the explanation or instructions, we can miss the practice of working them out. Here, you put the steps into your own words and look for what you assumed.

YOUR FIRST TRY

Leave nothing to guess.

0 steps

Write one action per step. Aim for 10–15 steps, with a maximum of 20. Clear coverage counts, not extra words.

Before you start, how complete do you think your answer will be?
Optional. This is your prediction, not part of the score.

Your draft stays in this browser.

The first check looks for key actions and connections. You can review its matches if you used different wording.

THE COLLECTION

Ordinary things. Missing steps.

A different challenge each day, cycling through this set of 12. Try any challenge here as extra practice. Each draft is saved separately.

THE QUIET COST

Keep the explanation yours.

Working out an explanation asks you to retrieve what you know, put events in order, connect causes to effects, and notice where your understanding runs out. These are the actions this challenge asks you to practice.

Familiarity can hide a gap

Rozenblit and Keil’s 2002 experiments included explaining a zipper. People’s confidence in their understanding of mechanisms fell after trying to explain them. Their procedure study showed a different pattern, with no significant confidence drop. Our zipper and instruction challenges therefore practice related, distinct skills.

Read Rozenblit & Keil, 2002 ↗

Generating an explanation can help learning

A 2018 meta-analysis found that prompting self-explanation improved learning on average across the included studies. That supports asking you to explain before showing a reference. It does not validate this particular game or show that gains transfer to work.

Read Bisra and colleagues, 2018 ↗

AI can change how much thinking we do

A 2025 survey of 319 knowledge workers linked greater confidence in AI to less reported critical thinking. Because it was a survey, it does not establish that AI caused lasting skill loss.

Read Lee and colleagues, CHI 2025 ↗

A 2026 randomized coding study found weaker unaided understanding after AI-assisted learning of a new library. It is a preprint about a specific task and short-term skill formation. It does not show that this challenge prevents decline.

Read Shen & Tamkin, 2026 preprint ↗

What the score means

The percentage describes coverage of this challenge’s reference checklist. Each milestone has equal weight. In instruction challenges, a matched action earns half credit if an essential prerequisite is missing or comes later. Independent actions can change order. Mechanism challenges assess connections without imposing one writing order.

The automatic check uses phrase patterns. It can miss valid wording or accept a mistaken sentence. Review the evidence, revise your explanation, or link a missed milestone to the step where you explained it. Those corrections are labeled self-reviewed. This is a practice score, not a validated measure of intelligence, writing quality, or skill loss.

Bring in your steps.

One step per line, with or without numbers. This replaces the current draft for this challenge.