| name | pivot-card |
| description | Draws a random pivot card to break cognitive freeze when stuck on a problem. Each card is a concrete move that generates new information. Use when user is stuck, spinning, confused, or explicitly invokes /pivot-card. |
Pivot Card
Externalizes the choice of what to try next at the exact moment the brain defaults to ineffective loops (re-reading, wheel-spinning). Draw a card, apply it to the problem, execute — even badly. Wrong execution generates information. Hesitation generates nothing.
Cards
| # | Card | Move |
|---|
| 1 | Compute Something | Pick any numbers and calculate. Don't wait for the right values — wrong numbers still reveal structure. |
| 2 | Extreme Cases | Push inputs to limits: zero, infinity, all-identical, empty, maximum. Watch what breaks or simplifies. |
| 3 | Remove One Part | Strip out one term, constraint, or component. Solve the simpler problem. Whatever collapses tells you what that part was actually doing. |
| 4 | Reverse | Run time, direction, or causality backwards. What asymmetry appears? That asymmetry is a real fact about the problem, not a framing accident. |
| 5 | Rotate / Swap | Change the frame, coordinate system, or swap labels/inputs/roles. Ask: what stays the same? what breaks? The invariants are the structure. |
| 6 | Make Prediction | Commit to a checkable claim before looking anything up — "if X then Y." Then verify. Forces the vague model to pay rent. Works best after 4–5 other cards when you've gathered enough data. |
| 7 | Sit With Confusion | Don't move yet. Locate exactly where the confusion is. Narrow it to one sentence: "I don't know why ___." Precision about what you don't know is itself progress. |
Workflow
- User is stuck or invokes
/pivot-card
- Roll a die — pick a number 1–7 at random (do not optimize, do not pick what seems most relevant)
- State the card drawn
- Explain concretely what that move means for this specific problem
- Prompt the user to execute it now — explicitly grant permission to do it wrong
Principles
- Random pick breaks the freeze. The point is not to choose optimally — it's to stop the loop. Any card beats staring.
- Execute wrong on purpose. Don't evaluate "how exactly should I do this?" — that overhead kills the attempt. Use obviously wrong numbers, absurd cases, imperfect swaps. Wrong execution reveals why it's wrong, which exposes constraints and hidden structure.
- Motion precedes clarity. The understanding comes from doing the move, not from deciding which move to do.
- Over time the moves become internal. Weight toward cards that feel unfamiliar — those are still expanding how you think.
Source
Developed from a comment by Paul on David Bessis's Substack: https://davidbessis.substack.com/p/attention-is-all-we-have/comment/209394137
