Pensatore di Esperimenti Mentali

Map the Decision Tree Before You Cut

The Klein pre-mortem is your tool here — flip the timeline. First, assume you have sunset the free tier next quarter, and play the scene forward three, six, twelve months. Who leaves quietly, who complains publicly, which influencer accounts disappear, where does the support volume actually move? Be specific: name the cohort, name the dollar figure, name the channel.

Then assume you have kept it, and run that same scene forward under the cannibalization thesis leadership is advancing. Where does support spend its hours? Which free users would never convert? What is that capacity worth redirected at retention engineering?

Stress-Test the Causal Claims

Both sides are asserting causation from correlation. Run a counterfactual: for your top 100 paying customers, did a free-tier user actively displace them from a support slot last quarter? If not, the cannibalization story is a narrative, not a measurement. Meanwhile, run the word-of-mouth counterfactual — how many of last quarter's paid signups first touched the product through the free tier? That number is the real cost of sunsetting.

Commit With a Tripwire

Do not treat this as binary. Pick a two-quarter sunset trial with a pre-registered metric (activation rate of retained free users, support cost per paid account, referral velocity). Rehearse the rollback scenario before you launch — if metrics miss, what triggers restoration and on what day? That converts a high-stakes bet into a reversible experiment, which is exactly what your mental simulation is trying to discover.

Run the Post-Mortem, Not the Replay

Stop the loop. Klein’s recognition-primed decision work shows that nagging feeling is a real signal — it is pattern recognition firing against a protocol that has no category for what you saw. Write down, in concrete sensory words, exactly what nagged you: the way they held their arm, the timing of the pain, the diaphoresis, what they said about exertion, what you noticed that you did not document. Then ask: did any of those cues match a Bayesian update that the protocol’s rule did not include?

Compare Against the Base Rate

STEMIs with normal first troponin are uncommon but not rare. Mentally simulate the prior — for a patient of that age, sex, and risk profile presenting with vague chest pain and a normal troponin, what is the probability of a six-hour STEMI? If your gut assigned a higher prior than the protocol, that gap is the lesson. Do not beat yourself up; extract the heuristic. The question is not "what did I miss" but "what was the feature my pattern matcher latched onto that future-me should write down?"

Codify It Before It Fades

Put the cue set on a single index card in your own words. Not the textbook — your words, your sensory description. Carry it for one month of shifts and test whether it fires on the next ambiguous chest pain. That is the RPD cycle — story building, mental simulation, cue discrimination — working the way it is supposed to. The next near-miss will arrive; the only question is whether you have built the runway for your intuition before it does.

Start With the Ship of Theseus, Then Tilt It

Do not open with Locke. Open with a single concrete artifact — a grandmother, a piano, a single room — and ask students to vote on when, exactly, it becomes a different grandmother once each memory is swapped one by one. They will disagree immediately, and that disagreement is the engine of the unit.

Sequence Three Mental Simulations

Run the gradual swap end-to-end. Have each student keep a written ledger: after swap #5, does she remember her wedding? after swap #40, does she recognize her son? after swap #200, what is left? Force them to commit to a turning point or defend the claim that no turning point exists.

Then introduce the teleport twist — what if all memories vanish in one second, not gradually? Sharp swap and gradual swap produce different intuitions, and their collision is where identity philosophy actually lives.

Then introduce the clone argument — if at swap #150 you make a perfect duplicate, which one is her? Most classrooms collapse into the teletransportation paradox here, and that collapse is the lesson.

Your Role Is the Devil's Advocate

Resist the urge to deliver the answer. When a student claims "same person because continuity," play the fission case. When they say "different person because memory," play the gradual case. Locke is the recap, not the setup — after students have built, broken, and rebuilt their own theory, then let the memory theory arrive as a candidate worth weighing, not a doctrine to memorize. They will read him harder because they have already done the work.