TAROT — SIX FICTIONAL EXERCISES AND MODEL ANSWERS

Internal companion 0.1 — 6 November 2026

Six fully worked fictional cases, based on the instructions of corpus tarot-reflexion-0.12.0. The cards are chosen for teaching purposes: no draw or personal session. The examples reveal nothing about a person and validate no prediction or tradition. The model answers show one possible answer, not a correct psychological answer.


M8 — Justice — One identical rule, two effects
EXISTING INSTRUCTION: Take ten minutes to apply the same rule to two fictional examples, then note one difference in treatment.
Corpus criterion: You can explain the difference or change your rule.
Corpus limit: An identical rule does not always produce a fair effect; this exercise is not a legal opinion.
FICTIONAL DATA: In a fictional paper game, the rule is “a label has at most eight letters”. The two labels to examine are ROSE and GRANDBLEU. Count the letters, then discuss the limit of this rule without judging a person.
POSSIBLE MODEL ANSWER
ROSE has 4 letters: the rule is met.
GRANDBLEU has 9 letters: the rule is exceeded by 1.
Answer: The treatment is consistent with the written rule: one example is accepted and the other rejected. This does not demonstrate that the rule is suitable: if the aim is to make the names understandable, cutting GRANDBLEU could lose a piece of information. Possible revision in draft: allow for a second line, then check its legibility.
Check: Both examples are treated with the same rule; the justification and a limit are written down. Formal compliance is not enough to prove fairness.
Counter-test / limit: Replacing GRANDBLEU with BLEU produces two accepted labels. This variant shows that the rejection depends here on length, not on an intention attributed to someone.


M12 — The Hanged Man — Two points of view, some unknowns
EXISTING INSTRUCTION: In ten minutes, describe a fictional disagreement from two points of view without deciding which is true.
Corpus criterion: Each point of view contains one need and one piece of information to check.
Corpus limit: Changing perspective does not oblige you to accept a situation or to give up a limit.
FICTIONAL DATA: Fictional disagreement about a game poster: one person proposes the title “Atelier bleu”; another wants to add the day of the meeting. No real exchange takes place. Describe two possible needs, without claiming to know their intentions.
POSSIBLE MODEL ANSWER
Only two text proposals are given.
The place where the poster will be displayed and the information already available are unknown.
Answer: Point of view A, hypothesis: keeping a short title could make it easier to read; to be checked by a legibility comparison. Point of view B, hypothesis: adding the day could help to organise the meeting; to be checked by asking, within the fiction, where the date will appear. No need is established as a fact. One variant would be to keep the title and place the day on a separate line.
Check: Two possible needs and two checks are named. The disagreement can remain open; it is not necessary to choose who is right.
Counter-test / limit: If the date already appears just below the title, the requested addition may become redundant. If it appears nowhere, the short version may lack a useful piece of information.


M15 — The Devil — Examine a habit without judgement
EXISTING INSTRUCTION: In five minutes, describe an ordinary habit, what it brings you and a limit that you choose yourself.
Corpus criterion: The advantage and the limit are named without moral judgement.
Corpus limit: The card reveals neither malice nor coercive control; it does not allow you to label another person.
FICTIONAL DATA: Fictional habit: before each paper drawing session, sorting twelve pencils by colour. The session lasts only ten minutes. Describe one possible benefit and one freely chosen limit; no real pencil or behaviour is evaluated.
POSSIBLE MODEL ANSWER
The sorting by colour and the planned duration are declared in the fiction.
The time actually taken by the sorting and its usefulness are not measured.
Answer: Supposed benefit: finding a colour could be easier. Limit chosen in draft: spend at most two minutes on the sorting and stop even if it remains incomplete. Nothing yet proves that this habit helps. A future comparison could look at one fictional session with sorting and another without, without concluding that there is a cause from this single example.
Check: A hypothetical benefit and a limit are named. Noting “usefulness unknown” is enough; changing the habit is not required.
Counter-test / limit: If the colours are already easy to find without sorting, the supposed advantage may not appear. This says nothing about a person’s worth, health or intentions.


M16 — The House of God — Plan a counter-test before the test
EXISTING INSTRUCTION: For ten minutes, choose a low-risk project hypothesis and imagine a small test that could disprove it.
Corpus criterion: A result contrary to your hypothesis is defined before the test.
Corpus limit: No accident, ruin or break-up is foretold; do not look for a danger in order to make the card fit.
FICTIONAL DATA: Fictional mock-up hypothesis: “all the proposed titles have at most twelve letters, with no space”. The small test set comprises ATELIER, BLEU and CONSTRUCTIONS. Before counting, define what would contradict the hypothesis; do not modify any real project.
POSSIBLE MODEL ANSWER
ATELIER has 7 letters, BLEU 4 and CONSTRUCTIONS 13.
CONSTRUCTIONS exceeds the limit of 12; one counter-example is enough here to refute “all”.
Answer: Counter-observation defined before counting: finding at least one title of thirteen letters or more. The third title meets this condition. Bounded assessment: the hypothesis is false for this list; no catastrophe or fragility of a person is discovered. Possible revision: describe the two short titles and the long title separately, then try a layout on a copy.
Check: The contrary result was defined before the test and then actually observed. The conclusion remains limited to this list.
Counter-test / limit: Another list containing only ATELIER and BLEU would be compatible with the hypothesis for those two titles only; it would not prove that every possible title respects the limit.


M18 — The Moon — One impression, two explanations
EXISTING INSTRUCTION: Take ten minutes to write down one impression then two possible explanations, distinguishing the facts already observed.
Corpus criterion: At least one uncertainty remains explicitly open.
Corpus limit: An impression is not proof of a lie, a secret or a health problem.
FICTIONAL DATA: On a fictional mock-up, a thumbnail shows only “Atelier bl…”. The declared source title is “Atelier bleu”. No real software has been tested. Separate the impression, the facts of the fiction and two possible explanations.
POSSIBLE MODEL ANSWER
The declared thumbnail is abbreviated and the declared source title is complete.
The thumbnail settings, the available space and the rendering at normal size are unknown.
Answer: Possible impression: “the title seems cut off”. Hypothesis A: the thumbnail only abbreviates its preview; possible check on the source and at normal size. Hypothesis B: an area that is too narrow hides the end; possible check on a copy by widening this area. Neither of these causes has been observed here. Assessment: cause unknown, no real error demonstrated.
Check: Two explanations remain distinct from the declared observations and at least one unknown remains open. Finding a definite cause is not the result requested.
Counter-test / limit: A complete title at normal size would weaken the hypothesis of a loss in the document; an ending that is still missing would call for further checks. No lie, secret or health problem is inferred.


M20 — Judgement — An assessment that keeps what fails
EXISTING INSTRUCTION: Spend ten minutes on the assessment of a trial already carried out: observed result, surprise and question to take up again.
Corpus criterion: The assessment also contains what did not work or remains inconclusive.
Corpus limit: An assessment is not a verdict on your worth; no outside call or event is predicted.
FICTIONAL DATA: Fictional mock-up trial already carried out: shortening four titles to a maximum of ten letters, with no space. The declared outputs are ATELIER, BLEU, CONSTRUCTION and DESSIN. The protocol measures neither speed nor satisfaction. Write an assessment, a surprise and a question to revisit.
POSSIBLE MODEL ANSWER
The lengths are 7, 4, 12 and 6 letters. Three titles respect the limit; CONSTRUCTION exceeds it by 2.
The result concerns only the four declared outputs, not a user trial.
Answer: Assessment: objective partly achieved, three titles out of four, and not four out of four. Surprise to record as a fictional reaction: one output remains long despite the stated objective. Question to revisit: what information would be lost by shortening CONSTRUCTION? Keeping the long output and trying a second line may be an option to examine, without announcing it as validated.
Check: The observed result, the failure kept, the fictional surprise and the question are present. No improvement in efficiency, mood or personal worth is demonstrated.
Counter-test / limit: Four short titles would only make it possible to observe that the limit is respected for these four outputs; this would prove neither better understanding nor future success.


Internal editorial review only; no expert or independent reader consulted. The notebook and its notes remain in their tab; this document reuses none of their data. A downloaded file remains on your device.