Decide & Adapt · Week 3 · Judgment · pre-mortem

Kill the Plan

Students take a polished plan, imagine it has already failed, list why, rank the causes and add fixes, then spot a new problem one of their fixes could create.

Level 1 AI after thinkingGrades 8–1245 minutesNone for students (AI step on the teacher’s screen, optional)

The big idea

It is easier to see how a plan fails once you pretend it already has.

Students will be able to:

  • Run a pre-mortem in a team
  • Rank risks by how likely and how bad they are
  • Write a fix for the top risks
  • Name a second-order effect of a fix

Lesson flow · the 5 P’s in decision mode

Think first. Then the tool.

Most of the lesson happens on paper. AI appears once, as a critic, and the teacher can run it from one screen. How the 5 P’s work in decision mode →

1

Ponder

6 min○ Screens off · paper only
  1. Hand out the plan. It is labelled “written by an AI assistant” but was written by ARQIV for this lesson, with weaknesses on purpose.
  2. Students read it alone and give it a score from 1 to 10 for “how likely is this to go well?”
Say to students“This plan sounds great. That is exactly why we are going to kill it.”
2

Plan

4 min○ Screens off · paper only
  1. Explain the pre-mortem: “It is a month later. The Fun Run was a disaster. Write down every reason why.”
  2. Each student writes at least four reasons alone, one per sticky note, before talking.
Say to students“Don’t ask ‘what might go wrong?’ Say ‘it went wrong, and here is why.’”
3

Partner

6 min● AI on · teacher screen, optional
  1. Optional, on the teacher’s screen: ask a chatbot to run a pre-mortem on the original plan.
  2. Compare: which causes did the class find that the AI missed, and the other way round? Which AI causes are generic?
Imagine it is one month after this event and it failed badly. List the five most likely reasons, specific to this plan, ranked from most to least likely: [paste plan].
Say to students“The AI is fast at listing risks. You know the school. Who found the better ones?”
4

Probe

17 min○ Screens off · paper only
  1. In teams, cluster the sticky notes and remove duplicates.
  2. Rank the causes: how likely, how bad. Circle the top three.
  3. Write one fix (a mitigation) for each of the top three.
  4. For one fix, ask: “What new problem could this fix create?” That is a second-order effect. Write it down.
Say to students“Every fix has a cost. Moving the date might fix the rain risk and clash with another event.”
5

Prove

12 min○ Screens off · paper only
  1. Each team writes a revised plan section: three mitigations and one second-order effect, with how they would handle it.
  2. Two teams present in 90 seconds. Others ask: “What would make your fix fail?”
Say to students“We want a plan that has already survived an attack. It doesn’t have to be perfect.”

Materials

What you need

Everything here is hand-authored and fictional unless a source is named.

Checklist

  • The “AI-written” Fun Run plan (hand-authored, below)
  • Sticky notes or slips of paper
  • Worksheet, one per student

Hand-authored example

The plan (hand-authored for this lesson; labelled as AI-written to mimic a polished AI draft)

Spring Fun Run · Plan v1
Goal: raise $3,000 for new library books. Date: Saturday, April 25, 9 am to noon, on the school field. Route: three laps around the field and the staff parking lot. Sign-ups: students register online through a class social media page; $15 entry includes a T-shirt. T-shirts: order 400 shirts in advance in one size (adult medium) to keep costs down. Volunteers: Ms. Rivera will coordinate everything. Water: one table at the finish. Rain plan: the forecast is usually good in late April. Promotion: one announcement two weeks before.

Hand-authored example

Weaknesses we planted (teacher key)

  • No rain plan; the date is fixed.
  • One person coordinates everything; no backup if they are ill.
  • Registration through social media excludes students without accounts and students under 13.
  • 400 one-size shirts paid up front: money is lost if fewer people come, and many won’t fit.
  • The route crosses the staff parking lot.
  • Only one water table, at the finish.
  • Promotion is late and one-off.
  • No check on the date clashing with other events.

Printable

Student worksheet

Print one per student. Everything students hand in is handwritten, so their thinking is visible.

Kill the Plan

Decide & Adapt · Week 3. Write your own thinking; it is what gets graded.

Name:Date:Class:
1 · My first score for the plan (1–10) and why
2 · Pre-mortem: “It failed because…” (at least four reasons)
3 · Our top three causes, ranked (likely × bad)
4 · A fix for each of the top three
5 · One fix’s second-order effect: what new problem could it create, and what would we do?
6 · My score for the revised plan, and what changed my mind

Assessment

Grade the process, not the answer

Every row scores something students did themselves. A lucky answer with no reasoning scores low; a careful decision that turned out wrong can score high.

Criteria4 · Excellent3 · Solid2 · Developing1 · Beginning
Range of causesFinds causes across people, money, timing and safety, including at least one not in the teacher keyFinds four or more distinct causesFinds two or threeOne or none
RankingRanks by likelihood and impact with reasonsRanks with a reason for the top causeRanks without reasonsNo ranking
MitigationsFixes are specific, affordable and address the causeFixes address the causesFixes are vague (“be careful”)No fixes
Second-order thinkingNames a realistic side effect of a fix and how to handle itNames a realistic side effectSide effect is unrelated to the fixMissing

Why this lesson

Suggestive

Imagining that an outcome has already happened produced about 30% more reasons for it than imagining it might happen. The study counted reasons; it did not measure their quality or decision accuracy. The research base for the premortem is thin.

The 1989 study found that imagining a failure had already happened produced about 30% more reasons. It did not measure whether decisions got better. We use the pre-mortem because it is cheap and makes risks visible, not because it is proven to improve accuracy.

Adjust it

Younger students: Run the pre-mortem as a whole class with four causes, and pick one fix together.

Older students: Ask teams to estimate a probability for each top cause and an expected cost, then rank by both.

No devices at all? Skip the Partner step or read this sample AI pre-mortem aloud: “Possible issues: bad weather, low turnout, budget overruns, volunteer shortages, safety concerns.” Ask: which of these is specific to this plan? Which ones did your team make more useful?

The AI step

Level 1 AI after thinking

Optional and short. Use it on the teacher’s screen as a red team: it critiques, it does not decide. Prompt:

Imagine it is one month after this event and it failed badly. List the five most likely reasons, specific to this plan, ranked from most to least likely: [paste plan].

Exit ticket

Name a plan in your own life (a project, a trip, a game) and one way it could fail that you hadn’t thought of until today.

Why this is hard to cheat

The plan is fictional and its weaknesses are planted, so there is nothing to look up. Students’ own causes and fixes, and the second-order effect, are written in class.

Go further

The Black Box classroom game →

Honest limits and sources

Where this comes from

Open the source before quoting it.

Honest limits

What this page leaves out

This is an initial edition, written by ARQIV in October 2026. It has not been classroom-piloted or independently reviewed, so its scenarios and timings may change after classroom use. The evidence named above supports the core move of the lesson, not this lesson as a whole. We do not claim the lesson measures or certifies a student’s judgment or “learning agility.”

Review status

Written by ARQIV. Initial edition, October 2026. This lesson has not been classroom-piloted or independently reviewed yet, and every question and scenario is hand-authored and checked against its sources by ARQIV. Tell us about anything wrong or unclear.

  1. Mitchell, Russo and Pennington (1989), as reviewed by Jason Collins, "Premortem"; Gary Klein, "Performing a Project Premortem," Harvard Business Review (2007).

All links opened and checked October 21, 2026 (accessed October 2026).