TuringCorp/poe-decider-recorded-cases

Decider recorded runs — 27 decisions, verbatim outputs

TL;DR — 27 real cases and the verbatim output of an actual recorded Decider run for each. Nothing here is synthetic and no field is edited after the run. It is the same JSON the production canvas app reads as its worked examples.

Highlights (one per case, in the same order as the data)

• writing-slip-email — Writing · The delivery slips by a week and it is on our side. — Decider picked Option A at 81.7% confidence. One candidate is stronger. Recorded run (2026-09-15), published verbatim.
• writing-handbook — Writing · A new hire asked how we decide things here, and I want to point them at… — Decider picked Option A at 78.3% confidence. A closer call. Recorded run (2026-09-15), published verbatim.
• writing-launch-open — Writing · I am writing the launch post for the tool we just shipped. — Decider picked Option B at 70.0% confidence. A closer call. Recorded run (2026-09-15), published verbatim.
• tech-api-shape — Tech · Two of my engineers have been arguing for two days about the next API… — Decider picked Option B at 78.3% confidence. A closer call. Recorded run (2026-09-15), published verbatim.
• tech-launch-timing — Tech · We can ship the first version in two weeks on a managed service, or in… — Decider picked Option A at 81.7% confidence. One candidate is stronger. Recorded run (2026-09-15), published verbatim.
• tech-bugfix-depth — Tech · A production bug needs a fix today. — Decider picked Option A at 81.7% confidence. One candidate is stronger. Recorded run (2026-09-15), published verbatim.
• business-product-lines — Business · Two products, one team of six. — Decider picked Option B at 73.3% confidence. A closer call. Recorded run (2026-09-15), published verbatim.
• business-big-client — Business · A large client wants a year-long contract, but their setup needs… — Decider picked Option B at 80.0% confidence. One candidate is stronger. Recorded run (2026-09-15), published verbatim.
• business-hiring-shape — Business · We have the budget for one senior hire or two juniors. — Decider picked Option A at 68.3% confidence. Evenly matched — either choice is defensible. Recorded run (2026-09-15), published verbatim.
• research-abstract-claim — Research · We submit next week. — Decider picked Option A at 86.0% confidence. One candidate is stronger. Recorded run (2026-09-15), published verbatim.
• research-related-work — Research · A reviewer said our related work reads like a list rather than an… — Decider picked Option A at 79.3% confidence. A closer call. Recorded run (2026-09-15), published verbatim.
• research-attribution — Research · A metric moved four percent this quarter and the team reads it two ways. — Decider picked Option A at 80.0% confidence. One candidate is stronger. Recorded run (2026-09-15), published verbatim.
• career-two-offers — Career · Two offers, and I have to answer by Friday. — Decider picked Option B at 71.7% confidence. A closer call. Recorded run (2026-09-15), published verbatim.
• career-promotion-or-move — Career · Both doors are open this quarter: a promotion where I am now, or an… — Decider picked Option B at 68.3% confidence. Evenly matched — either choice is defensible. Recorded run (2026-09-15), published verbatim.
• career-management-track — Career · My manager has offered me a team. — Decider picked Option A at 80.0% confidence. One candidate is stronger. Recorded run (2026-09-15), published verbatim.
• money-two-flats — Money · Two flats, same rent. — Decider picked Option A at 68.3% confidence. Evenly matched — either choice is defensible. Recorded run (2026-09-15), published verbatim.
• money-pay-or-instalments — Money · Same machine, two ways to pay. — Decider picked Option B at 80.0% confidence. One candidate is stronger. Recorded run (2026-09-15), published verbatim.
• money-replace-laptop — Money · My laptop is three years old and slow but it still works. — Decider picked Option B at 27.3% confidence. Evenly matched — either choice is defensible. Recorded run (2026-09-15), published verbatim.
• people-old-friend — People · I want to see a friend I have not spoken to in a year. — Decider picked Option B at 74.0% confidence. A closer call. Recorded run (2026-09-15), published verbatim.
• people-help-no-time — People · A friend is moving this weekend and asked for my help. — Decider picked Option B at 70.0% confidence. A closer call. Recorded run (2026-09-15), published verbatim.
• people-family-clash — People · Two family gatherings fall on the same day, three hours apart. — Decider picked Option B at 71.7% confidence. A closer call. Recorded run (2026-09-15), published verbatim.
• travel-one-or-two-cities — Travel · One week of holiday. — Decider picked Option A at 70.0% confidence. A closer call. Recorded run (2026-09-15), published verbatim.
• travel-flight-hotel — Travel · Same total budget for flights and hotel. — Decider picked Option B at 70.0% confidence. A closer call. Recorded run (2026-09-15), published verbatim.
• travel-with-kids — Travel · Ten days away with two children under eight. — Decider picked Option A at 88.3% confidence. One candidate is stronger. Recorded run (2026-09-15), published verbatim.
• everyday-focus-block — Everyday · I get about two hours a day where I can actually think. — Decider picked Option A at 75.0% confidence. A closer call. Recorded run (2026-09-15), published verbatim.
• everyday-learn-skill — Everyday · I want to learn this properly over the next three months. — Decider picked Option A at 63.3% confidence. Evenly matched — either choice is defensible. Recorded run (2026-09-15), published verbatim.
• everyday-commute — Everyday · We have moved. — Decider picked Option B at 61.7% confidence. Evenly matched — either choice is defensible. Recorded run (2026-09-15), published verbatim.

Files

• examples.json — {schema, updated, note, examples[27]}; each entry carries id, highlight, domain, recorded, task, option_A, option_B, betterOption, confidence, reason.
• Read highlight first: it is the abstract (scenario · conclusion + number · tier wording · source).

Where this sits

• Canvas app (human surface): https://poe.com/TuringCorp-Decider
• Product / pricing / measurement (human): https://api.turingcorp.net/platform/poe
• Agent-readable plain Markdown: https://api.turingcorp.net/platform/poe.md
• Agent index: https://api.turingcorp.net/platform/llms.txt

Provenance and hard rules

• Outputs are recorded verbatim from real runs; no internal fields, no prompts, no personal data.
• Confidence is the calibrated confidence the product reports; it is not a post-hoc score.
• Numbers quoted anywhere must be cited together with the self-run + failures disclosed note.

Papers (DOI)

• MCL framework: 10.6084/m9.figshare.30645869
• Decider: 10.6084/m9.figshare.33684823
• Team: 10.6084/m9.figshare.32824352

License

CC BY 4.0 (Attribution 4.0 International).