Direct answer

Core relationship: Use square decision points for actions and branching chance points for uncertain events. Multiply outcome value by branch probability, then roll values backward while preserving the worst case.

For decision trees for small uncertain choices, begin with draw choices first and finish with roll back and inspect. The sequence keeps the starting facts, intermediate result, and final limit visible instead of relying on memory.

Method you can audit

Draw choices first

Include only actions that are actually available. Capture the result before moving to add uncertain events. A written output shows whether the next deduction follows from the evidence or quietly adds a new assumption.

Add uncertain events

Branches should be mutually exclusive and collectively useful. Compare this output with the warning about false branch completeness. If that failure still fits, revise the representation before using the result in another deduction.

Assign evidence-based inputs

Use ranges when probabilities or values are weak. Use assign evidence-based inputs as a checkpoint. Name the input, the change made, and the conclusion that becomes available so another reader can reproduce the move.

Roll back and inspect

Compare expected value, downside, and future options together. Finish by stating what this step establishes and what it leaves undecided. That boundary prevents a valid local conclusion from being stretched beyond the question.

Worked interpretation

Draw choices first in the case

A full launch offers a larger gain but a costly failure branch. A pilot offers smaller immediate value, cheaper failure, and information that can change a later launch decision.

Why roll back and inspect matters

The case starts with draw choices first, then uses add uncertain events to control the next inference. The conclusion is acceptable only after roll back and inspect. Changing any fact that supports those moves can change the answer, while an unrelated detail should not.

Limits and traps

False branch completeness

A tidy pair of outcomes may omit important middle cases. Return to draw choices first and rebuild only from information the problem actually provides.

Point estimates only

Sensitivity checks reveal whether uncertainty changes the choice. Compare the disputed move with assign evidence-based inputs; the two should use the same definitions and direction.

Ignoring learning value

A branch can create information as well as immediate payoff. Run the practice task while deliberately watching for ignoring learning value. Record the first point where the reasoning begins to drift.

Check against false branch completeness

Questions for add uncertain events

  • Can another reader reproduce draw choices first from the stated information?
  • Did you complete add uncertain events without reversing or enlarging the claim?
  • Does the result survive the failure mode false branch completeness?
  • Which part remains uncertain after roll back and inspect?

This check is specific to the method above. A clear answer should expose the evidence chain and the point where a different fact would produce a different conclusion.

Use it on a real problem

Model two options in the decision tree calculator, then vary each favorable probability by ten points and note whether the preferred option changes.

Continue after roll back and inspect

Use the problem prioritization guide for the broader method, then apply the idea in the decision matrix calculator. Keep the written reasoning so the next attempt can improve a specific step rather than simply repeat the exercise.

Compare this method with Reversible vs Irreversible Decisions, then continue to Clue Economy in Puzzle Design for a different reasoning task.