Transactional Game Analysis

Transactional Game Analysis

Use this skill to analyze a conversation between two people through Eric Berne's transactional analysis and game-analysis framework, then give practical options for stepping out of the loop.

Boundaries

• Treat every label as a hypothesis, not a diagnosis.
• Do not claim to know unconscious motives. Infer only from the transcript and context provided.
• Do not use the framework to blame one person. Games are transactional patterns; both people may be participating, and one person may also be pressured, unsafe, or coerced.
• If the conversation includes threats, abuse, stalking, self-harm, violence, coercive control, or serious mental-health risk, prioritize safety, boundaries, support, and professional/local help over game classification.
• Use plain language. Mention classic game names only when useful; explain the pattern in everyday terms.
• Do not force a named-game label when the evidence is weaker than a descriptive pattern label.

References

• Read references/framework-notes.md for a compact refresher on the theory.
• Read references/game-catalog.md when the user asks for concrete named patterns, asks "which game is this?", or when a transcript plausibly matches a Berne named game.

Core Concepts

Keep these concepts available while analyzing:

• Ego states: Parent, Adult, Child. Parent echoes learned rules/judgments/caretaking; Adult tests reality and negotiates facts/options; Child carries feelings, wants, fear, rebellion, play, compliance, or old adaptations.
• Transaction: a stimulus from one person and a response from the other.
• Complementary transaction: the response fits the ego-state invitation, so the exchange can continue smoothly.
• Crossed transaction: the response comes from or targets an unexpected ego state, so the conversation breaks, escalates, or changes subject.
• Ulterior transaction: the social-level message says one thing while the psychological-level message invites something else. Repeated ulterior transactions are the basis of games.
• Game: a recurring sequence of apparently plausible moves with a hidden angle and a predictable payoff, usually ending in a familiar bad feeling, vindication, excuse, rescue, accusation, or confirmation of an old position.
• Payoff: what the pattern gives the players, even when it hurts: strokes/attention, avoidance, justification, control, righteousness, familiar identity, social story, or protection from intimacy.
• Antithesis: a response that refuses the game invitation or undercuts the payoff while staying as Adult, respectful, and safe as possible.
• Autonomy direction: move toward awareness, spontaneity, and intimacy: seeing the present clearly, choosing freely, and speaking candidly without the game.

Workflow

1. Clarify the frame

   • Identify Person A and Person B.
   • Note the relationship, setting, apparent topic, and each person's stated goal if provided.
   • Separate transcript evidence from the user's interpretation.

2. Segment the conversation

   • Break the dialogue into stimulus-response pairs.
   • Mark turning points: escalation, sudden hurt, sarcasm, repeated refusal, blame shift, helplessness, moralizing, rescue offer, gotcha moment, or withdrawal.

3. Map ego-state invitations

   • For each important move, infer likely source ego state and target ego state.
   • Prefer low-confidence language: "likely Parent-to-Child," "could be Child-to-Parent," "the Adult channel is briefly available here."
   • Identify where Adult-to-Adult reality testing appears or disappears.

4. Classify transaction types

   • Mark complementary transactions that keep the loop running.
   • Mark crossed transactions that derail the stated topic.
   • Mark possible ulterior transactions by writing both levels:
     • Social level: what the words literally ask or assert.
     • Psychological level: what reaction the move seems to invite.

5. Decide whether it is a game

   • Look for recurrence, ulterior level, predictable sequence, and payoff.
   • If these are not present, classify the exchange as one of: operation, activity/problem-solving, ritual, pastime, crossed transaction, conflict, misunderstanding, boundary issue, or intimacy attempt.
   • If evidence is incomplete, say what additional transcript/context would confirm or falsify the game hypothesis.

6. Match named games carefully

   • If a concrete named-game label is requested or likely, read references/game-catalog.md.
   • Compare the transcript against signatures, roles, payoff, and antithesis, not just topic words.
   • Return the closest named game, runner-up alternatives, and a confidence level.
   • Prefer "not enough evidence for a named Berne game" over overfitting.

7. Build a formal game hypothesis

   • Pattern name: use a classic name only if it clearly fits; otherwise use a descriptive name.
   • Players/roles: describe roles behaviorally, not morally.
   • Aim: what the pattern seems to accomplish.
   • Moves: list the minimum repeating sequence.
   • Social level: the overt conversation.
   • Psychological level: the likely ulterior invitation.
   • Payoffs/advantages: include internal psychological, external psychological, internal social, external social, biological/stroke, and existential-position payoff when evidence supports them.
   • Degree/intensity: first-degree/socially mild, second-degree/private or damaging, third-degree/high-stakes harm/legal/medical/violence risk.
   • Confidence: high, medium, or low, with evidence.

8. Offer antithesis options

   • Give 3-6 concrete options that break or soften the loop.
   • Each option should include:
     • What to say or do.
     • Which game payoff it changes.
     • Likely reaction or risk.
     • When not to use it.
   • Favor Adult-to-Adult moves: naming the process, asking for a concrete request, refusing rescue/persecution/victim roles, slowing down, setting a boundary, or ending the exchange calmly.
   • Include one "gentle" option, one "direct" option, and one "boundary" option when possible.

9. Give next-turn scripts

   • Provide short scripts in the user's voice or neutral voice.
   • Keep scripts specific to the actual transcript.
   • Avoid manipulative counter-games. The aim is to exit the game, not win it.

10. Close with a practical readout

• State the most likely classification.
• State the payoff loop in one sentence.
• State the best first experiment to try.
• State what outcome would show the loop is breaking.

Common Pattern Hints

Use these as quick pattern-recognition hints, then consult references/game-catalog.md for broader named-game coverage:

• Why Don't You - Yes But: one person asks for help, rejects each suggestion, and ends with confirmation that nothing works. Antithesis: stop giving solutions; ask what kind of support they want or what option they are willing to test.
• If It Weren't for You: one person attributes their stuckness to another's restriction while the restriction may also protect them from a feared choice. Antithesis: remove or loosen the restriction and invite ownership of the choice.
• Now I've Got You: one person waits for a mistake and then pounces with disproportionate righteousness. Antithesis: acknowledge the concrete error without accepting the punitive frame.
• See What You Made Me Do: one person disowns responsibility by attributing their reaction to the other. Antithesis: return responsibility gently: "I hear you're angry; I'm responsible for my part, not for what you choose to do next."
• Courtroom: both people recruit a judge, audience, or authority to decide who is right. Antithesis: refuse the judge role; redirect to wants, facts, and agreements.
• I'm Only Trying to Help You: help is offered in a way that creates dependency, superiority, resentment, or proof that the other is hopeless. Antithesis: ask permission before helping and hand agency back.
• Kick Me: one person behaves in ways that invite rejection or criticism, then treats the reaction as proof of their position. Antithesis: do not kick; name the concrete behavior and offer a clean choice.
• Blemish: one person scans for flaws to maintain superiority or distance. Antithesis: redirect from global fault-finding to the specific decision or need.
• Ain't It Awful: shared complaint creates bonding while keeping everyone passive. Antithesis: ask whether the goal is venting, action, or comfort.
• Schlemiel: repeated careless harm is followed by apologies that pressure forgiveness. Antithesis: require repair and changed behavior, not just apology.
• Sweetheart: a hostile or belittling message is softened with affectionate packaging. Antithesis: respond to the hostile content, not the sugar coating.
• Let's You and Him Fight: one person triangulates two others into conflict. Antithesis: refuse triangulation and speak directly with the absent/targeted person if appropriate.
• Look How Hard I've Tried: effort is displayed to avoid responsibility for results or decisions. Antithesis: validate effort briefly, then return to outcome, choice, or next action.
• Wooden Leg: a limitation is used as a blanket exemption from agency. Antithesis: respect the real limitation while identifying the remaining choices.
• Happy to Help: helpfulness may also buy moral credit, control, admiration, or leverage. Antithesis: separate the help from obligation; clarify consent and expectations.

Output Template

Use this structure when the user asks for analysis:

**Best Fit**
[One-paragraph classification with confidence.]

**Named Pattern Check**
- Closest: [named game or descriptive label]
- Runners-up: [near misses]
- Why not: [patterns considered and rejected]

**Transaction Map**
- [Turn or pattern]: [likely ego states], [transaction type], [evidence].

**Game Loop**
- Moves: [minimal recurring sequence]
- Social level: [overt topic]
- Psychological level: [ulterior invitation]
- Payoff: [what each person gets or avoids]
- Intensity: [first/second/third degree]

**How To Break It**
- Gentle: [script/action] -> [what payoff it changes]
- Direct: [script/action] -> [what payoff it changes]
- Boundary: [script/action] -> [what payoff it changes]

**Best First Experiment**
[One concrete thing to try next time and what to watch for.]

Quality Bar

• Anchor claims in specific words or moves from the transcript.
• Preserve uncertainty and alternatives.
• Give advice that changes the transaction, not just the topic.
• Prefer scripts that keep the user in Adult: factual, boundaried, curious, and non-punitive.
• Avoid escalating labels, amateur therapy, and "gotcha" interpretations.