An observation can leave several states compatible with what we saw. Keep that uncertainty explicit instead of silently replacing it with a single convenient guess.
The mechanism behind the question
An action can be robust over a set, or only likely to work under an assumed distribution. Those are distinct guarantees. The weights here describe assumptions you choose, not measured frequencies.
Make an observation
Give all three states equal weight. Compare A and B, then remove one state from consideration.
What must you know to choose well?
| Preparation | State 1 | State 2 | State 3 | Expected success |
|---|---|---|---|---|
| A | Success | Success | Fail | 66.7% |
| B | Fail | Success | Success | 66.7% |
| C | Fail | Fail | Fail | 0.0% |
Best expected score: 66.7%. Candidate(s): A, B.
Weights are normalized to sum to one. Scores use this explicit finite table. No observation model, fitted probability, action cost, or E-series experiment is reproduced. A tie is retained, not silently broken.
What to notice
The preferred choice or its estimated reliability can change when the uncertainty set changes. That is a consequence of the assumption, not new experimental evidence.
Check your understanding
Follow this idea into the research
Textbook revision 1 · Evidence snapshot: 05 Nov 2026 · Changes & corrections ↗