01 / OBSERVED IN FICTION
Recorded Tuesday
Mira boards; she arrives before the workshop begins, but loses her notes on the carriage.
AN OPEN RECORD OFFICE FOR IMPOSSIBLE EVIDENCECASEBOOK NO. 001–012 / V1
THE OTHER VERSION OF TUESDAY
What if the other ending wasn't the answer?
The thesis. Imagining an alternative can help expose assumptions and sharpen questions, but a plausible story about what could have happened is not evidence that it would have happened. Hindsight changes how compelling those stories feel.
Open a case file ↗Twelve fictional decisions · one controlled difference · zero forecasts
ILLUSTRATION 01 · One documented-in-fiction page; another page that nobody witnessed.
01Separate record from invention.
02Change one condition, not everything.
03Write down what stays unknowable.
ROOM 01 / THE BRANCH RECORD OFFICE
Alternative Timeline Builder
Select a fictional file. Compare the recorded branch with precisely one alternate assumption. If you change two, the clerk refuses to stamp the file.
APPLICATION / FORM CF–01
THREE VARIABLES · CHANGE AT MOST ONE
The counterfactual is a scripted illustration, not an estimate or measured probability.
RESULT / INSPECTABLE PROVENANCEBASELINE · FILED
Last Train, Tuesday
Mira reaches a station as the last train is announced.
02 / ILLUSTRATIVE ALTERNATIVE
The unchosen branch
Mira takes the bus; she arrives after introductions but keeps her notes and meets a helpful stranger.
Provenance ledger not an empirical dataset
OBSERVED IN FICTION
Within the authored record, Mira boards and arrives before the workshop.
ASSUMED
The platform gate remains open for two minutes.
UNKNOWN
The unwitnessed bus route and any unmentioned conditions.
How the file was compiled
- Baseline conditions are fixed.
- The scripted decision is compared with its alternate.
- No outcome is claimed as historical fact.
The alternate branch is imaginable. It is not observed.
EXHIBIT / 02
The unknowable drawer
Even the neatest branch cannot establish an unobserved outcome. We can specify a model, ask what changes, and check whether assumptions are supported; we cannot retrieve a second Tuesday from a drawer.
Open the adversarial file: when changing a choice changes nothing
In The Library Workshop, an extra reminder is sent, but attendance remains three. A different choice need not alter a result when an independent bottleneck holds. This guards against the seductive assumption that every intervention must change an outcome.
EXHIBIT / 03
The hindsight stamp
Read the ending before deciding how “inevitable” it feels. Fischhoff's research concerns how outcome knowledge can shift judgments of what could have been foreseen. An instructive question is not “Was it obvious?” but “What was genuinely knowable before the outcome?”
Inspect the research ledger ↗ROOM 02 / A METHOD WITH KNOWN GAPS
The one-variable rule is useful. It is not magic.
Counterfactual reasoning connects mental simulation, causal inquiry and narrative interpretation. These are related disciplines, not interchangeable kinds of proof.
We cannot observe both versions of one event.
In a simple causal question, we want to know what difference one action made while other relevant circumstances are comparable. That framing is close to potential-outcomes reasoning: the same unit under two incompatible conditions cannot be observed simultaneously in the ordinary way. Experimental design and defensible assumptions help researchers estimate differences across observations; a written story does not provide that evidence.
Here the original decision is part of an invented record. Pressing “Change only B” selects another prewritten narrative. The result demonstrates the logic of conditional comparison, not the actual likelihood of anything. It is especially useful when someone mistakes vivid detail for knowledge.
What is the strongest reasonable objection?
Altering only one visible variable can be too neat. In real systems the decision can change relationships, timings, incentives and information. Holding every other fact fixed may create an incoherent world. Critics of simple counterfactual accounts discuss preemption, overdetermination and dependencies that undermine naive but-for explanations. A controlled contrast is a question format, not a guarantee that the world obeys it.
Our response is to keep the ledger visible, refuse multiple simultaneous edits, and name the limitations. The rule forces discipline, but if the assumption of separability is false, the resulting story must be revised or discarded rather than dressed up as certain.
What this model cannot prove
It cannot reconstruct a real past, infer motives, rank actual decisions, establish causality, estimate probabilities, or predict your future. It cannot tell whether a missing variable would reverse the story. It can help visitors practise writing the difference between a fact, an assumption and a wish.
The same sentence can be an observation within a fictional world and an invention relative to reality. This is why the ledger's first column is deliberately labelled Observed in fiction, never simply “Evidence.”
ROOM 03 / TWELVE COMPLETE CASE FILES
Every Tuesday has a missing page.
All characters, outcomes and institutions in these twelve scenarios are fictional. Each file includes its premise, recorded outcome, changed decision, hidden context, a fair counterreading and a lesson.
FILE 01 · FICTIONAL RECORD
Last Train, Tuesday
Mira reaches a station as the last train is announced.
Open complete case record
Constraint: The platform gate remains open for two minutes.
Documented inside the fiction: Mira boards; she arrives before the workshop begins, but loses her notes on the carriage.
One changed choice: Take the night bus. Mira takes the bus; she arrives after introductions but keeps her notes and meets a helpful stranger.
Assumed context: The workshop may start late; punctuality might not matter.
Unknown: A service disruption could affect either route.
Counterexample: A late start makes both routes functionally equivalent; neither branch proves the smarter choice.
Lesson: Earlier preparation matters more than the transport choice alone.
Provenance: The arrival time is recorded inside this story; the alternative journey is not witnessed.
FILE 02 · FICTIONAL RECORD
The Borrowed Umbrella
Eshan sees storm clouds before leaving the neighbourhood library.
Open complete case record
Constraint: The library closes at six and rain is possible.
Documented inside the fiction: Eshan borrows it; he arrives dry but must return it the following day.
One changed choice: Leave without it. Eshan leaves without it; the rain stops, and he takes a longer stroll home.
Assumed context: A sheltered route may change the stakes.
Unknown: The forecast could have been wrong.
Counterexample: Weather remains dry in both stories, so the decision does not alter the immediate result.
Lesson: An observed dry arrival cannot establish that borrowing was unnecessary.
Provenance: The forecast is a prompt inside fiction, not a measured probability.
FILE 03 · FICTIONAL RECORD
The Quiet Meeting
Leena reviews an ambiguous slide just before a team meeting.
Open complete case record
Constraint: The chair allotted five minutes for questions.
Documented inside the fiction: Leena asks; the misunderstanding is surfaced, but the meeting runs over.
One changed choice: Stay silent. Leena stays silent; the meeting ends on time, then two colleagues interpret the slide differently.
Assumed context: The chair might already know the ambiguity.
Unknown: Another attendee may independently ask.
Counterexample: A colleague asks the same question, producing the same clarification in both branches.
Lesson: Uncertainty is not always resolved merely by speaking.
Provenance: Only the fictitious meeting record is fixed.
FILE 04 · FICTIONAL RECORD
The Community Garden
Noor is offered one free afternoon to help plant seedlings.
Open complete case record
Constraint: The soil is dry and volunteers have limited water.
Documented inside the fiction: Noor plants; a volunteer carries extra water and most seedlings survive in the story.
One changed choice: Wait for water delivery. Noor waits; water arrives later, but shade moves and planting takes longer.
Assumed context: Extra volunteers could change the bottleneck.
Unknown: The water shipment may be delayed.
Counterexample: A sprinkler starts regardless, removing the assumed water constraint.
Lesson: A helpful intervention can have knock-on effects outside the chosen action.
Provenance: No survival rates are real observations.
FILE 05 · FICTIONAL RECORD
The Unsent Draft
Asha has written feedback on a neighbourhood newsletter.
Open complete case record
Constraint: The editor will publish the next edition tomorrow.
Documented inside the fiction: Asha sends it; the editor uses one suggestion but misses a nuance.
One changed choice: Review it in the morning. Asha revises it in the morning; the message is clearer, but the edition is already laid out.
Assumed context: The editor might accept late copy.
Unknown: Delivery time can vary.
Counterexample: The edition is postponed, so both choices lead to inclusion.
Lesson: The best-crafted message is not necessarily the most influential.
Provenance: This is not a recommendation about personal communication.
FILE 06 · FICTIONAL RECORD
The Repair Queue
Irfan can repair a bicycle tyre now or inspect the whole wheel.
Open complete case record
Constraint: A repair shop closes in forty minutes.
Documented inside the fiction: Irfan replaces the tube; he rides away, then notices the rim still rubs.
One changed choice: Inspect the rim first. Irfan inspects the rim; he finds a burr but runs out of time to finish the repair.
Assumed context: A nearby shop might be open later.
Unknown: The rim problem may be cosmetic.
Counterexample: The tyre remains flat for a separate valve reason in both branches.
Lesson: The fast fix and the thorough fix optimize different goals.
Provenance: The defect is invented, not a safety assessment.
FILE 07 · FICTIONAL RECORD
The Exhibition Label
Tara curates a small photo exhibition with an ambiguous image.
Open complete case record
Constraint: The gallery permits one additional line on each caption.
Documented inside the fiction: Tara adds a line; viewers understand the setting but debate the intended meaning.
One changed choice: Leave the image untitled. Tara leaves it untitled; viewers generate diverse readings, some unsupported by context.
Assumed context: Visitors may read the catalogue.
Unknown: The photographer might revise the title.
Counterexample: A curator talk supplies the missing context in either version.
Lesson: More context can reduce one confusion while narrowing an artwork.
Provenance: There is no measurement of visitor understanding.
FILE 08 · FICTIONAL RECORD
The Cooking Class
Dev has enough ingredients for one group meal and a late guest arrives.
Open complete case record
Constraint: The kitchen has one small oven.
Documented inside the fiction: Dev changes the recipe; everyone receives a portion, but cooking finishes later.
One changed choice: Keep the original recipe. Dev keeps the recipe; it finishes promptly, and guests share smaller portions.
Assumed context: A second tray may be available.
Unknown: The guest may bring a dish.
Counterexample: The late guest brings extra food, eliminating the shortage in both branches.
Lesson: What counts as success depends on a value judgment about timing and fairness.
Provenance: These are fictional outcomes, not controlled experimental observations.
FILE 09 · FICTIONAL RECORD
The Museum Ticket
Sana arrives at a museum with time for only one gallery.
Open complete case record
Constraint: Both galleries will close at five.
Documented inside the fiction: Sana visits sculpture; she notices a technique she later sketches.
One changed choice: See the archive room. Sana visits archives; she discovers a letter that reshapes her understanding of the exhibition.
Assumed context: Gallery staff could provide a shortcut.
Unknown: One wing may close unexpectedly.
Counterexample: A shared tour moves through both rooms; the choice loses its exclusivity.
Lesson: There is no single objective best visit without specifying the goal.
Provenance: No artwork or letter represents a real museum object.
FILE 10 · FICTIONAL RECORD
The Library Workshop
Ravi organizes a public reading club with only three attendees signed up.
Open complete case record
Constraint: The room holds twenty people and posters are already printed.
Documented inside the fiction: Ravi sends a reminder; attendance remains three because everyone who can come has already registered.
One changed choice: Do not send a reminder. Ravi sends no reminder; attendance remains three for the same reason.
Assumed context: A local event may compete for attention.
Unknown: A walk-in visitor could appear.
Counterexample: This is the adversarial fixture: changed B, same attendance, distinct action trace.
Lesson: A single change may have no effect if the outcome depends on another bottleneck.
Provenance: The shared attendance count is an authored fictional stipulation.
FILE 11 · FICTIONAL RECORD
The Shared Desk
Pia and Arun both need the same desk for an afternoon.
Open complete case record
Constraint: There is one power outlet and a quiet corner.
Documented inside the fiction: They schedule shifts; each gets quiet time but loses the overlap for discussion.
One changed choice: Work together side by side. They share the desk; a useful exchange happens, but concentration becomes difficult.
Assumed context: Another desk could become free.
Unknown: Neither person may need mains power.
Counterexample: The power supply fails; both plans need a different workspace.
Lesson: A seemingly fair allocation can impose hidden coordination costs.
Provenance: The workload and preferences are invented.
FILE 12 · FICTIONAL RECORD
The Festival Map
Omar volunteers to draw a map for a local festival.
Open complete case record
Constraint: Visitors have only one printed page to navigate.
Documented inside the fiction: Omar prioritizes landmarks; first-time visitors orient themselves visually but miss minor streets.
One changed choice: Prioritize street names. Omar prioritizes street names; regulars navigate precisely, while first-time visitors hesitate.
Assumed context: Signage onsite could supplement the map.
Unknown: A temporary road closure could invalidate either layout.
Counterexample: A volunteer at the gate gives directions, reducing the difference between versions.
Lesson: Different users need different evidence about what a map makes clear.
Provenance: Visitor responses are illustrative and not survey data.
THREE SHORT ESSAYS
Interpretation, filed in triplicate.
The branch we never saw
There is an asymmetry in every imagined alternative. The actual outcome is a memory, a log or a record; the alternative is constructed. The mind may make the construction feel similar to a recollection by filling it with vivid weather, dialogue and detail. But vividness measures the craftsmanship of the story, not the truth of its ending.
A disciplined case asks which facts are fixed, which causal links have evidence, and how many incompatible stories fit the same starting scene. If more than one reasonable sequel survives, a categorical “would have” is too strong.
Lessons without time travel
Counterfactuals can be useful without establishing what happened in the unrealized branch. They can identify a condition worth studying or an intervention worth testing. In a well-designed experiment, a comparison across units can support a causal claim under explicit assumptions. The fantasy of revisiting an individual past is not required.
The practical lesson is to change the question. Instead of “Would everything have gone better?” ask “Which constraint was under my control, what else might have changed, and what evidence could distinguish the competing explanations?”
Regret is not evidence
After disappointment, one attractive alternative can come to dominate attention: “If only I had chosen B.” The feeling may identify something deeply valued, but it cannot certify the future of a different choice. Sometimes alternative B would have brought a different cost; sometimes a common bottleneck would have made the same ending.
The bureau avoids personal confession and refuses to measure regret. It invites a modest epistemic habit: make room for uncertainty even when the imagined branch feels emotionally convincing.
ARCHIVE NOTE / TWO RESEARCH CONTRASTS
Documented research is not a fictional branch.
Hindsight in judgment
DOCUMENTED EVIDENCE. Fischhoff's 1975 research studied how knowing an outcome can affect judgments of what could have been anticipated, and subsequent work has explored how that effect complicates retrospective inference. The precise study conditions matter. We do not reproduce those experiments here.
FICTIONAL COMPARISON. A fictional clerk sees that the night bus arrived late and decides the train choice was obviously correct. The clerk does not know about the lost notes or whether the bus was always late. That imagined clerk is a teaching device, not a study participant.
Controlled comparisons in causal inference
DOCUMENTED METHODOLOGY. Rubin (1974) and Holland (1986) frame causal effects in terms of potential outcomes and the methodological difficulty of comparison. Randomization and careful design can help make causal questions answerable, although each design has limits.
FICTIONAL COMPARISON. Changing a setting in our case compiler simply retrieves a different authored scene. No random assignment, measurements or research participants are involved. That is why the bureau uses no effect sizes or probability dials.
Counterposition: Philosophers debate how counterfactual dependence maps onto causation, especially in cases of preemption and overdetermination. The workshop is an editorial aid, not a resolved causal theory.
PERSONAL DESK / PRINT OR THINK
Which part of your imagined version was knowledge—and which was desire?
Try one fictional case. Record what the file says, one change you introduced, a competing explanation, and evidence that would change your mind. Don't use a vivid ending as your only reason.
CASE REFLECTION
Observed in fiction: __________________________
One variable changed: ________________________
Assumption I need: ____________________________
Alternative reading: __________________________
Evidence that could discriminate: ______________
Download the Counterfactual Casebook ↓Substantive, print-friendly PDF with 12 cases and a reusable worksheet.READING ROOM / SOURCE LEDGER
What we can verify. And what we cannot.
These bibliographic details and broad propositions were checked against publisher, institutional and abstract pages on 28 October 2026. Full articles were not universally inspected. Cases on this website are original fiction and have no empirical provenance.
- 01Roese, Neal J. (1997). Counterfactual Thinking. Psychological Bulletin 121(1), 133–148. Source recordSupports: Counterfactual thought has multiple cognitive functions and is an established research subject. Counterreading: A review of thought patterns does not validate scripted forecasts.
- 02Fischhoff, Baruch (1975). Hindsight ≠ Foresight: The Effect of Outcome Knowledge on Judgment Under Uncertainty. Journal of Experimental Psychology: Human Perception and Performance 1(3), 288–299. Source recordSupports: Outcome knowledge can distort retrospective likelihood judgments. Counterreading: The study does not establish how any particular fictional choice would turn out.
- 03Rubin, Donald B. (1974). Estimating Causal Effects of Treatments in Randomized and Nonrandomized Studies. Journal of Educational Psychology 66(5), 688–701. Source recordSupports: Controlled design and randomization bear on causal inference. Counterreading: Our narrative branches are not estimates from randomized data.
- 04Holland, Paul W. (1986). Statistics and Causal Inference. Journal of the American Statistical Association 81(396), 945–960. Source recordSupports: The potential-outcomes framework distinguishes causal questions from simple associations. Counterreading: A fictional same-person alternative cannot be observed as ground truth.
- 05Byrne, Ruth M. J. (2005). The Rational Imagination: How People Create Alternatives to Reality. MIT Press. Source recordSupports: Mental alternatives are structured by reasoning processes. Counterreading: Human imaginability is not proof that the alternative occurred.
- 06Byrne, Ruth M. J. (2016). Counterfactual Thought. Annual Review of Psychology 67, 135–157. Source recordSupports: Counterfactual thought supports reasoning about the past and future. Counterreading: Its affective functions make it vulnerable to preference and regret.
- 07Kahneman, Daniel & Tversky, Amos (1982). The Simulation Heuristic. Judgment under Uncertainty: Heuristics and Biases, 201–208. Source recordSupports: Mental simulation is a distinct judgment operation. Counterreading: Ease of simulation can mislead about probability.
- 08Woodward, James (2004). Making Things Happen: A Theory of Causal Explanation. Oxford University Press. Source recordSupports: Interventions and explanatory dependencies matter to causal questions. Counterreading: Changing words in a story is not an empirical intervention.
- 09Stanford Encyclopedia of Philosophy (2024 revision). Counterfactual Theories of Causation. Stanford Encyclopedia of Philosophy. Source recordSupports: Simple one-factor accounts of causation face counterexamples and refinements. Counterreading: The philosophical disputes are not settled by this interface.
- 10Stanford Encyclopedia of Philosophy (2025 revision). Counterfactuals. Stanford Encyclopedia of Philosophy. Source recordSupports: Counterfactual conditionals have philosophical and logical subtleties. Counterreading: Different semantics need not yield one preferred narrative.
- 11Fischhoff, Baruch (2025). Fifty years of hindsight bias research—Reflection on Fischhoff (1975). Journal of Experimental Psychology: Human Perception and Performance 51(2), 143–150. Source recordSupports: Hindsight bias remains a live research subject. Counterreading: The review does not guarantee immunity from bias through one exercise.
- 12Byrne, Ruth M. J. (2017). Counterfactual Thinking: From Logic to Morality. Current Directions in Psychological Science 26(4). Source recordSupports: Counterfactual thinking can influence moral judgments and reasoning. Counterreading: This ethical link is context-sensitive and should not be used to assign blame.
HELP DESK / SIX QUESTIONS
Before you close the file.
Is this an alternate-life predictor?
No. The branches are written fiction. They teach a distinction between a possible story and evidence about what would have occurred.
Why change only one input?
Controlling a single difference makes assumptions auditable. It does not itself make a narrative causal estimate; that also requires defensible causal structure and evidence.
What if the changed decision has no effect?
That is a valuable counterexample. A causal story may omit a bottleneck, shared cause, or a pathway that neutralises the intervention.
Can I use this to settle regrets?
It is not a personal therapy, diagnostic or regret-ranking tool. If a scenario resonates, use it as a fictional thinking exercise, not a verdict on your past.
What does hindsight bias have to do with it?
Once an outcome is known, alternatives may feel less plausible than they did in advance. That retrospective feeling is not a record of genuine foresight.
Does the workshop collect my selections?
No. It runs in the browser and makes no analytics, backend calls, account, cookies or local storage. Print or download creates a local document only.