Antecedent Edit Card¶
Reasoning artifact — instantiates Minimal-Change Counterfactual Construction
States the one antecedent change being imagined — and only that change — as a surgical edit to a single node of the causal model.
The Antecedent Edit Card is a small, explicit statement of the one thing being changed — and nothing else. It names the single variable the counterfactual sets to a new value, records that variable's actual value beside its counterfactual value, and pins the change to one identified node of the causal model so everyone is imagining the same intervention. Its defining move is isolation: it fixes the lever and refuses to bundle anything else with it. It does not follow the consequences of the edit downstream — that is the entailment audit's job — and it does not judge closeness or preservation. If the antecedent cannot be written as a single, well-located edit, the card forces that fact into the open before any "what would have happened" is allowed to proceed.
Example¶
An epidemiology team reviews a district's pandemic response and debates the counterfactual "if school closures had started a week earlier." Before arguing outcomes, they fill an Antecedent Edit Card. The variable is the closure start date; its actual value is the date closures actually began, its counterfactual value one week earlier. The card locates this as a single node — the school-contact rate at that time step — in the transmission model, and classifies the change as an intervention that sets the node, not an observation that merely correlates with it. Just as importantly, the card lists what it is deliberately not changing: testing capacity, public compliance, hospital staffing, and the response's competence in every other respect all stay at their actual values. That last line is what prevents the debate from sliding into "if the whole response had been better," a package deal masquerading as one lever. With the edit pinned to a single node, the team can hand a clean, well-posed change to the audit that traces its consequences.
How it works¶
- Name the exact variable. State precisely what changes, at what granularity, avoiding vague verbs like "improved" that hide a cluster of edits.
- Record actual value beside counterfactual value. The pair — what was, what is imagined — makes the size and direction of the edit explicit.
- Locate one node in the causal model. Point to the single node the edit sets, so the change enters the model at a defined place rather than as free-floating narrative.
- Declare intervention semantics. Mark the edit as setting the node (a surgical intervention), which severs its normal causes, rather than merely observing a different value.
- List the deliberate non-edits. Spell out what the antecedent is not bundled with, so nothing rides in under the same label.
Tuning parameters¶
- Edit atomicity — one node versus a small bundle. Truly atomic edits keep the counterfactual clean; bundling several under one banner ("a better rollout") is where cherry-picking hides.
- Value specificity — a point value versus a range for the counterfactual setting. A range acknowledges genuine uncertainty but complicates the downstream trace.
- Node placement — how far upstream the edited node sits. An upstream root-cause node makes many things contingent and hard to hold fixed; a proximate node keeps the edit tight but may miss the real lever.
- Intervention vs. observation — whether the new value is imposed on the node or conditioned on. The choice changes which upstream facts stay fixed.
- Explicitness of non-edits — how thoroughly the "not changed" list is enumerated. Fuller lists prevent smuggling but take effort.
When it helps, and when it misleads¶
Its strength is that it turns a slippery "what if things had gone differently" into a single, well-posed lever everyone can reason about. Written as an intervention on one node, it matches the surgical logic of the do-operator in causal inference: set this variable, sever its ordinary causes, and leave the rest of the model in place.[n1] That precision is what lets downstream mechanisms trace consequences and hold background facts fixed without ambiguity.
Its failure mode is the "package deal" edit — a card that says "if the launch had gone well" and quietly imports a dozen favorable changes under one heading. A related misuse is choosing a node so far upstream that nearly everything downstream becomes negotiable, letting the analyst reach any conclusion. The discipline that keeps it honest is one stated node, an explicit actual-and-counterfactual value pair, clear intervention semantics, and a written list of what is not being changed.
How it implements the components¶
counterfactual_antecedent_edit— the card is the declared edit: the exact variable, its actual and counterfactual values, and its intervention semantics.causal_model_alignment— it pins the edit to a single identified node in the causal model, so the change enters at a defined place and can be propagated coherently.
It sets the lever but does not follow it. The entailment_cascade_map — the downstream consequences the edit forces — is produced by Entailment Cascade Audit, its nearest relative here: the Card fixes the one intervention node, while the Audit traces everything that node moves.
Related¶
- Instantiates: Minimal-Change Counterfactual Construction — the Card supplies the single, well-located antecedent the rest of the construction builds on.
- Consumes: Actual Fact Ledger supplies the variable's actual value that the edit departs from.
- Sibling mechanisms: Actual Fact Ledger · Entailment Cascade Audit · Counterfactual Preservation Checklist · Gratuitous Change Challenge · Closest-World Ranking Table · Nearby-World Sensitivity Review
Editorial Notes¶
Form Classification¶
Form family: Representation, Specification & Plan
Rationale: States the one antecedent change being imagined — and only that change — as a surgical edit to a single node of the causal model, making its operative form a non-executable information artifact that externalizes static or prospective structure.
Independent corroboration: The frozen evidence defines Antecedent Edit Card as 'States the one antecedent change being imagined — and only that change — as a surgical edit to a single node of the causal model', so its operative form is Representation, Specification & Plan.
Review outcome: Independent reviewer agreement; high confidence.
Origin Attribution¶
Primary origin: Philosophy
Origin pattern: Cross-disciplinary synthesis
Present-day reach: Multi-domain
Rationale: Philosophy of causation supplies minimal-change counterfactuals and the distinction between intervening on and merely observing an antecedent.
Related originating lineages:
- Cognitive Science — Mental simulation uses minimal edits to compare possible worlds.
- Medicine & Healthcare — Epidemiologic causal models provide a major applied tradition.
- Statistics & Experimental Design — Causal inference formalizes single-variable interventions and counterfactual comparison.
Review resolution: Minimal-change counterfactual semantics in philosophy are primary. Mental-model editing, clinical counterfactual reasoning, and intervention-based causal inference materially shape the card; its explicit non-edit field is an Encyclopedia synthesis.
Encyclopedia synthesis: The exact catalogued form synthesizes established practice rather than reproducing a single standard historical label.
Review outcome: Reconciled after independent review; high confidence.
Notes¶
Atomicity is the whole discipline: a counterfactual is only informative if the one thing that differs is doing the explanatory work. The moment two things change together on a single card, the analysis loses the ability to say which one mattered — and that is exactly the failure the surrounding archetype exists to prevent.
[n1] Judea Pearl's do-operator models an intervention that sets a variable to a value while cutting the arrows from its usual causes, distinguishing "making X happen" from merely "seeing X happen." An antecedent edit is this operation applied to a single node. ↩