counterfactual-reasoning

Counterfactual Reasoning

I. Overview Table

Item	Content
Capability Name	Counterfactual Reasoning
English Name	Counterfactual Reasoning
Capability Type	Atomic capability of legal reasoning
Core Question	"If X had not occurred/ had changed, would outcome Y still have appeared?"
Classic Formulations	But-for Test, Conditio Sine Qua Non
Applicable Fields	Tort law, contract law, criminal law, administrative law, insurance law, competition law, constitutional review
Input	The actually occurring chain of facts + the target variable that is to be hypothetically changed
Output	The inferred result in the counterfactual world + analysis of the differences from the actual world + legal conclusion
Difficulty Level	★★★★☆ (advanced reasoning capability, requiring rigorous logical control)
Risk Level	High—errors in counterfactual reasoning may directly lead to errors in causation determination

II. Legal Statement

Important Notice: Counterfactual reasoning is a legal analysis tool, and its conclusions depend on an accurate grasp of the facts and the correct application of legal rules. This skill document provides a methodological framework and does not constitute legal advice on any specific case. Counterfactual reasoning inherently involves hypothetical judgments about events that did not occur and has inherent uncertainty. In practical application, it should be combined with evidentiary rules, standards of proof, and precedents of the specific jurisdiction for comprehensive judgment.

III. Core Concepts

3.1 The Essence of Counterfactual Reasoning

Counterfactual reasoning is a kind ofhypothetical causal analysis method, whose core logical structure is:

现实世界：A发生 → B发生
反事实世界：假设A未发生 → B是否仍然发生？

若B不再发生 → A是B的原因（A对B具有因果贡献）
若B仍然发生 → A不是B的原因（A对B无因果贡献）

3.2 Definition of Key Terms

Term	Definition	Example
Target Variable	The fact/act/condition assumed to be changed or removed	"The defendant's speeding driving behavior"
Outcome Variable	The legal consequence that needs to be observed to see whether it changes	"The plaintiff's personal injury"
Counterfactual World	The hypothetical situation after the target variable is changed	"The world in which the defendant drove at a normal speed"
Actual World	The factual state that actually occurred	"The defendant drove at excessive speed and injured the plaintiff"
Background Conditions	Other facts that remain unchanged in counterfactual reasoning	"The weather, road conditions, and the plaintiff's position at the time"
But-for causation	The "but-for" test—but for A, B would not have occurred	"But for the defendant's speeding, the collision would not have occurred"
NESS test	Necessary Element of a Sufficient Set test	Used for complex scenarios with multiple causes and one result
Break in the causal chain (Novus Actus Interveniens)	An intervening factor cuts off the original causal chain	Intervention of a third party's intentional act

3.3 Philosophical Foundations of Counterfactual Reasoning

The application of counterfactual reasoning in law is built on the following assumptions:

Possible-worlds semantics: We can meaningfully discuss "what would have happened if things had been different"
Principle of minimal departure: The counterfactual world should be as similar as possible to the actual world, differing only in the target variable
Stability of causal laws: Natural laws and social laws still hold in the counterfactual world
Assumption of determinability: The outcome in the counterfactual world can be determined within a reasonable range

3.4 The Relationship Between the But-for Test and the NESS Test

                    因果关系检验
                        │
            ┌───────────┴───────────┐
            │                       │
      单因单果场景              多因一果场景
            │                       │
      But-for检验              NESS检验
     "若非A则非B"         "A是某个充分条件集
                           中的必要元素"
            │                       │
      简单反事实              复合反事实
   （移除单一变量）        （分析多变量交互）

IV. Complete Workflow

Stage One: Problem Identification and Framework Construction

Step 1: Identify the need for counterfactual reasoning

Trigger signal detection:

The appearance of expressions such as "if... then it would not..."
It is necessary to determine the causal relationship between a certain act and a certain result
It is necessary to determine the scope of damages ("what would have happened without the breach")
It is necessary to assess the degree of contribution of fault to the result
It is necessary to determine whether an intervening factor breaks the causal chain
The appearance of the defense that "even if... it would still..."

Operational requirements:

□ 明确识别当前法律问题是否需要反事实推理
□ 确定反事实推理在整体法律分析中的位置（是因果关系要件、损害范围要件还是其他）
□ 判断是简单反事实（单因单果）还是复杂反事实（多因多果）

Step 2: Precisely define the target variable

Core principle: The target variable must be specific, operational, and legally meaningful

错误示例："如果被告没有做错事" → 过于模糊
正确示例："如果被告在2023年3月15日未将含有缺陷的产品投入市场" → 具体可操作

错误示例："如果情况不同" → 无法操作
正确示例："如果被告在施工时按照GB50300标准设置了安全围挡" → 具体可操作

Checklist for defining the target variable:

□ 目标变量是否足够具体？（时间、地点、行为内容）
□ 目标变量是否对应法律上的义务违反或权利侵害？
□ 目标变量的改变是否在物理/逻辑上可能？
□ 目标变量是否是当事人可以控制的？（对于归责分析）
□ 是否需要定义多个目标变量？（多因一果场景）

Step 3: Precisely define the outcome variable

□ 结果变量是否对应法律上的损害/后果？
□ 结果变量是否可以观察/衡量？
□ 结果变量的范围是否明确？（全部损害还是特定损害项目）

Step 4: Lock down the background conditions

Application of the principle of minimal departure:

When constructing the counterfactual world, apart from the target variable, all other conditions should remain consistent with the actual world.

现实世界事实清单：
  F1: 2023年3月15日，晴天
  F2: 被告以80km/h速度行驶（限速60km/h）← 目标变量
  F3: 原告正在人行横道上行走
  F4: 路面干燥
  F5: 被告车辆制动系统正常
  ...

反事实世界：
  F1: 2023年3月15日，晴天 ← 保持不变
  F2': 被告以60km/h速度行驶 ← 目标变量改变
  F3: 原告正在人行横道上行走 ← 保持不变
  F4: 路面干燥 ← 保持不变
  F5: 被告车辆制动系统正常 ← 保持不变
  ...

Precautions for locking down background conditions:

Facts that have a causal dependency relationship with the target variable may need to be adjusted accordingly
Facts independent of the target variable must remain unchanged
The judgment of whether something is "independent" may itself be disputed

Stage Two: Construction and Simulation of the Counterfactual World

Step 5: Construct the counterfactual world

Construction methods:

方法A：简单替换法（适用于单一、明确的目标变量）
  将目标变量替换为其"合法/正常/应然"状态

方法B：消除法（适用于需要判断某因素是否为原因的场景）
  将目标变量从事实链中完全移除

方法C：渐进调整法（适用于需要判断程度/比例的场景）
  逐步调整目标变量的值，观察结果变量的变化

Step 6: Derive the outcome in the counterfactual world

Rules of derivation:

Apply known causal laws: Physical laws, medical knowledge, economic laws, social common sense
Apply rules of thumb: What usually happens under similar conditions
Apply expert opinions: For causal judgments in specialized fields
Apply statistical data: For probabilistic causation

Template for recording the derivation process:

反事实推演链：
  前提：[目标变量被改变为X']
  第一步推演：因为X'，所以[中间事实M1]会/不会发生
    推演依据：[因果规律/经验法则/专家意见]
    确定性程度：[高/中/低]
  第二步推演：因为M1，所以[中间事实M2]会/不会发生
    推演依据：[因果规律/经验法则/专家意见]
    确定性程度：[高/中/低]
  ...
  最终结论：结果变量Y在反事实世界中[会/不会/部分]发生
    推演依据：[综合以上推演链]
    确定性程度：[高/中/低]

Step 7: Compare the two worlds

比较矩阵：
┌──────────────┬──────────────┬──────────────┬──────────────┐
│   比较维度   │   现实世界   │  反事实世界  │     差异     │
├──────────────┼──────────────┼──────────────┼──────────────┤
│ 目标变量     │     A        │     A'       │   被改变     │
│ 中间事实1    │     M1       │     M1'      │   有/无差异  │
│ 中间事实2    │     M2       │     M2'      │   有/无差异  │
│ 结果变量     │     Y        │     Y'       │   有/无差异  │
│ 损害程度     │     D        │     D'       │   差额=D-D'  │
└──────────────┴──────────────┴──────────────┴──────────────┘

Stage Three: Extraction of Legal Conclusions

Step 8: Extract legal conclusions from the comparison results

Conclusion type comparison table:

Comparison result	Legal conclusion	Applicable scenario
Y does not occur at all in the counterfactual world	A is a sufficient cause of Y (but-for causation is established)	Simple causation determination
Y still occurs in the counterfactual world	A is not a but-for cause of Y	Causation denied / need to switch to the NESS test
Y partially occurs in the counterfactual world	A is a partial cause of Y, contribution degree = D-D'	Proportional liability / determination of the scope of damage
Y occurs in a different way in the counterfactual world	A changed the manner in which Y occurred but did not change the occurrence of Y	Complex causation scenarios
Y occurs later in the counterfactual world	A accelerated the occurrence of Y	Accelerated causation
It cannot be determined whether Y occurs in the counterfactual world	Causation is uncertain and must be judged in light of the standard of proof	Insufficient evidence scenarios

Step 9: Confidence assessment and qualification

Annotate the conclusions of counterfactual reasoning with confidence levels (see Section 9 for details).

Step 10: Integrate into the overall legal analysis

Integrate the conclusions of counterfactual reasoning as the analysis result of the causation element (or other elements) into the complete legal argument.

V. Comparison of Common Fields and Legal Sources

5.1 Tort Law Field

Specific scenario	Counterfactual question	Main legal source	Test method
General tort causation	Would the damage have occurred without the tortious act?	Article 1165 of the Civil Code	But-for test
Multiple causes, one result	What is the contribution proportion of each cause to the damage?	Articles 1171-1172 of the Civil Code	NESS test + proportional analysis
Product liability	Would the damage have occurred if the product had no defect?	Articles 1202-1207 of the Civil Code	But-for test
Medical damage	Would the patient's outcome have been different if the medical conduct had complied with standards?	Articles 1218-1222 of the Civil Code	But-for test + loss of chance theory
Environmental tort	Would the environmental damage have occurred without the pollutant discharge act?	Articles 1229-1235 of the Civil Code	But-for test (reversal of burden of proof)
Third-party intervention	Does the intervening act interrupt the original causal chain?	Case law doctrine	Double counterfactual test

5.2 Contract Law Field

Specific scenario	Counterfactual question	Main legal source	Test method
Damages for breach of contract	What would the economic position of the performing party have been if the contract had been performed normally?	Article 584 of the Civil Code	Expectation interest counterfactual
Culpa in contrahendo	Absent the culpable conduct in contracting, what would the relying party's economic position be?	Article 500 of the Civil Code	Reliance interest counterfactual
Foreseeability rule	Could the breaching party have foreseen this loss at the time of contracting?	Article 584 of the Civil Code	Reasonable person counterfactual
Duty to mitigate losses	If the non-breaching party had taken reasonable measures, could the loss have been reduced?	Article 591 of the Civil Code	Mitigation counterfactual

5.3 Criminal Law Field

Specific scenario	Counterfactual question	Primary legal source	Test method
Causation in result crimes	Absent the criminal act, would the harmful result have occurred?	Causation theory in the General Part of Criminal Law	But-for test + adequate causation
Crime of omission	If the actor had performed the duty to act, could the result have been avoided?	Theory of crimes of omission	Hypothetical act counterfactual
Intervening factors	Are the intervening factors abnormal enough to break the causal chain?	Objective imputation theory	Risk realization counterfactual
Alternative causation	Two independent acts each sufficient to cause the result	Revised condition formula theory	NESS test

5.4 Administrative Law Field

Specific scenario	Counterfactual question	Primary legal source	Test method
Illegality of administrative acts	If the administrative organ had acted in accordance with the law, would the result have been different?	Administrative Litigation Law	Lawful act counterfactual
Administrative compensation	Absent the illegal administrative act, would the party's loss have occurred?	State Compensation Law	But-for test
Protection of reliance	If the administrative organ had not made the prior act, would the party have made the same decision?	Principle of protection of reliance	Decision counterfactual

VI. Verification and Screening Rules

6.1 Validity verification of counterfactual reasoning

Rule 1: Substitutability test of the target variable

检验问题：目标变量的改变在物理上和逻辑上是否可能？
  ✓ 有效："如果被告没有闯红灯"（行为可以不同）
  ✗ 无效："如果被告不是人类"（物理上不可能的替代）
  ✗ 无效："如果地球没有引力"（违反基本物理定律）

Rule 2: Minimal departure principle test

检验问题：反事实世界是否仅在目标变量上与现实世界不同？
  ✓ 有效：仅改变被告的驾驶速度，其他条件不变
  ✗ 无效：改变被告的驾驶速度的同时，假设天气也变好了
  ⚠ 注意：如果目标变量的改变必然导致其他事实改变，则需要追踪这种连锁效应

Rule 3: Reliability test of the inference chain

对推演链中的每一步进行检验：
  □ 该步推演是否有可靠的因果规律支撑？
  □ 该步推演是否存在替代可能性？
  □ 该步推演的确定性程度如何？
  □ 推演链的总体确定性是否满足法律证明标准？

Rule 4: Overdetermination test

检验问题：是否存在其他独立的充分原因？
  场景：A和B各自独立地足以导致Y
  But-for检验结果：移除A，Y仍然发生（因为B）；移除B，Y仍然发生（因为A）
  → But-for检验失败，需转用NESS检验
  NESS检验：A是{A, C1, C2...}这个充分条件集中的必要元素
            B是{B, C3, C4...}这个充分条件集中的必要元素
  → A和B都是Y的原因

Rule 5: Preemption test

检验问题：是否存在"备用原因"——如果实际原因不存在，另一个原因会导致相同结果？
  场景：A先于B导致了Y，但如果A不存在，B也会导致Y
  分析：A是Y的实际原因（因为A实际上导致了Y）
        B是Y的潜在原因（被A先占）
  → 不能因为B的存在而否定A的因果关系

6.2 Applicable boundaries of counterfactual reasoning

适用反事实推理的条件：
  ✓ 存在明确的事实基础
  ✓ 目标变量可以被有意义地改变
  ✓ 存在可靠的因果规律支撑推演
  ✓ 反事实世界的结果可以在合理范围内判断

不适用反事实推理的情形：
  ✗ 纯粹的法律解释问题（不涉及事实因果）
  ✗ 目标变量无法被有意义地改变
  ✗ 缺乏基本的因果规律知识
  ✗ 反事实世界的结果完全不可预测
  ✗ 法律明确规定了替代性因果关系认定方法

VII. Output Format Templates

Template A: Standard counterfactual reasoning report

## 反事实推理分析报告

### 1. 分析目的
[说明为什么需要进行反事实推理，对应哪个法律要件]

### 2. 变量定义
- **目标变量**：[具体描述被假设改变的事实/行为]
- **结果变量**：[具体描述需要观察的法律后果]
- **反事实假设**：[目标变量被改变为什么状态]

### 3. 背景条件锁定
[列出在反事实推理中保持不变的关键事实]
| 编号 | 背景事实 | 状态 |
|------|----------|------|
| F1   | ...      | 保持不变 |
| F2   | ...      | 保持不变 |
| ...  | ...      | ... |

### 4. 现实世界事实链
[描述实际发生的因果链]
A → M1 → M2 → ... → Y

### 5. 反事实世界推演
[描述反事实世界中的因果链推演]
A' → M1' → M2' → ... → Y'

推演依据：
- 第一步：[依据说明]
- 第二步：[依据说明]
- ...

### 6. 比较分析
| 维度 | 现实世界 | 反事实世界 | 差异 |
|------|----------|------------|------|
| 目标变量 | A | A' | 被改变 |
| 结果变量 | Y | Y' | [描述差异] |
| 损害程度 | D | D' | ΔD = D - D' |

### 7. 法律结论
[基于比较分析得出的法律结论]

### 8. 置信度评估
- 推演链可靠性：[高/中/低]
- 结论确定性：[高/中/低]
- 主要不确定因素：[列出]

### 9. 限定与说明
[对结论的限定条件和需要注意的事项]

Template B: Brief counterfactual reasoning (embedded)

**反事实分析**：若[目标变量的改变]，则根据[推演依据]，[结果变量]将[会/不会/部分]发生。
理由：[简要推演过程]。
因此，[目标变量]与[结果变量]之间[存在/不存在/部分存在]因果关系。
（置信度：[高/中/低]）

VIII. Confidence Annotation System

8.1 Confidence of the inference chain

Level	Mark	Meaning	Typical scenario
High certainty	`[CF-HIGH]`	The inference is based on established natural laws or uncontroversial rules of experience, and the conclusion is almost certain	Physical causation (e.g., no ignition means no combustion)
Relatively high certainty	`[CF-MED-HIGH]`	The inference is based on reliable professional knowledge or statistical regularities, and the conclusion is highly probable	Medical causation (e.g., timely surgery means a 90% survival rate)
Moderate certainty	`[CF-MEDIUM]`	The inference is based on reasonable experiential judgment; the conclusion is possible but alternative possibilities exist	Economic causation (e.g., the profit that could have been obtained if there had been no breach)
Relatively low certainty	`[CF-MED-LOW]`	The inference is based on general speculation, and the conclusion is uncertain	Behavioral prediction (e.g., whether the person would have made a different decision if informed of the risk)
Low certainty	`[CF-LOW]`	The inference is highly speculative, and the conclusion is extremely uncertain	Long-chain inference involving multiple uncertain variables

8.2 Overall conclusion confidence

整体置信度 = min(各步推演置信度)

即：推演链的整体置信度取决于最薄弱的环节。

示例：
  步骤1置信度：高
  步骤2置信度：中
  步骤3置信度：高
  → 整体置信度：中

8.3 Correspondence between confidence and standards of proof

Legal field	Standard of proof	Minimum required confidence
Civil cases (general)	High probability (>75%)	CF-MEDIUM or above
Criminal cases	Beyond reasonable doubt	CF-HIGH
Administrative cases	Clearly preponderant evidence	CF-MED-HIGH or above
Insurance claims	Preponderance of the evidence	CF-MEDIUM or above

IX. Common errors and prevention

9.1 Fatal error table

Number	Error name	Error description	Consequence	Preventive measure
E1	Ambiguous target variable	Failure to precisely define the fact changed by the hypothesis	The inference cannot proceed or the conclusion is unreliable	Strictly implement the checklist in step 2
E2	Excessive deviation	The counterfactual world differs too greatly from the real world, and not only the target variable was changed	The inference result cannot be attributed to the target variable	Strictly implement the principle of minimal deviation
E3	Ignoring overdetermination	Using only the But-for test in a multi-cause, single-effect scenario	Erroneously denying causation	Detect whether multiple independent sufficient causes exist, and switch to the NESS test when necessary
E4	Ignoring preemptive causation	Denying the actual cause because a backup cause exists	Erroneously denying causation	Distinguish actual causes from potential causes
E5	Break in the inference chain	Skipping key intermediate steps during the inference process	The conclusion lacks support	Infer step by step and label the basis for each step
E6	Confidence inflation	Assigning excessively high confidence to speculative conclusions	Misleading legal judgment	Strictly implement confidence labeling
E7	Counterfactual回溯过远	Tracing a counterfactual hypothesis back to too distant a past	The counterfactual world becomes uncontrollable	Limit the time range of the counterfactual hypothesis
E8	Confusing factual causation with legal causation	Equating de facto But-for causation with legal causation	Ignoring legal causation limitations (such as adequacy, foreseeability)	Clearly distinguishing the two levels of factual causation and legal causation

9.2 Common Pitfalls

Pitfall 1: Hindsight Bias

问题：知道实际结果后，倾向于认为该结果在反事实世界中也是"显而易见"的
示例：知道手术失败后，倾向于认为"如果换一种方案就一定能成功"
防范：
  - 从事前视角（ex ante）而非事后视角（ex post）进行推演
  - 考虑反事实世界中的多种可能结果
  - 引用事前的统计数据和专业标准

Pitfall 2: Slippery Counterfactual

问题：反事实假设引发连锁改变，导致反事实世界与现实世界差异过大
示例："如果被告没有出生" → 整个世界都不同了
防范：
  - 将反事实假设限定在最近的、最具体的行为层面
  - 不要追溯到行为人的性格、成长经历等深层因素
  - 使用"最近可能世界"原则

Pitfall 3: Cherry-picking Counterfactual

问题：选择对己方有利的反事实假设，忽视其他同样合理的假设
示例：原告主张"如果被告没有延迟交货，我就能获得100万利润"
      但忽视了"即使按时交货，市场价格下跌也会导致利润减少"
防范：
  - 系统性地考虑所有合理的反事实假设
  - 对每种假设分别进行推演
  - 明确标注哪些假设被采用及理由

Pitfall 4: Probability Neglect

问题：将"可能"等同于"确定"，忽视反事实结果的概率性
示例："如果及时治疗，患者就不会死亡"（实际上及时治疗的存活率可能只有60%）
防范：
  - 明确标注反事实结果的概率
  - 使用"机会丧失"理论处理概率性因果关系
  - 区分"确定会避免"和"有一定概率避免"

Pitfall 5: Time Frame Error

问题：反事实推演的时间框架选择不当
示例：评估违约损失时，将反事实世界推演到无限远的未来
防范：
  - 根据法律规则确定合理的时间框架
  - 考虑可预见性规则对时间范围的限制
  - 对远期推演降低置信度

Ten. Handling of Special Scenarios

10.1 Scenario of Multiple Causes Leading to One Result

Scenario Description: Multiple causes jointly lead to one result

Handling Method:

步骤1：识别所有可能的原因 A1, A2, A3, ...
步骤2：对每个原因分别进行But-for检验
  - 移除A1，Y是否仍然发生？
  - 移除A2，Y是否仍然发生？
  - ...
步骤3：根据But-for检验结果分类

情形A：移除任一原因，Y都不发生
  → 所有原因都是But-for原因
  → 按照各原因的贡献程度分配责任

情形B：移除某些原因，Y仍然发生（过度决定）
  → 转用NESS检验
  → 判断每个原因是否是某个充分条件集中的必要元素

情形C：各原因单独不足以导致Y，但共同导致Y
  → 所有原因都是But-for原因（因为移除任一原因，剩余原因不足以导致Y）
  → 共同因果关系，按贡献比例分配责任

10.2 Counterfactual Reasoning for Omission

Scenario Description: The actor failed to perform a duty to act, and it needs to be determined whether the result would have been different if the duty had been performed

Special Difficulty: It is necessary to construct a counterfactual world in which "the actor actively acts," but the specific content of the active act may be uncertain

Handling Method:

步骤1：确定行为人的作为义务内容
  - 法律规定的具体义务
  - 行业标准/注意义务的具体要求
  - 合同约定的具体义务

步骤2：构建"履行义务"的反事实世界
  - 假设行为人按照义务要求采取了行动
  - 如果义务内容有多种履行方式，选择"合理人"会采取的方式

步骤3：推演反事实世界的结果
  - 注意：不作为的反事实推理通常置信度较低
  - 因为"如果做了X会怎样"比"如果没做X会怎样"更难判断

步骤4：特别注意概率性结论
  - "如果医生及时诊断，患者有70%的存活概率"
  - 需要结合"机会丧失"理论进行法律评价

10.3 Hypothetical Causation

Scenario Description: The defendant claims that "even if I had not committed the tortious act, the same damage would have occurred due to other reasons"

Handling Method:

步骤1：区分"合法替代行为"抗辩和"假设因果关系"抗辩

合法替代行为抗辩：
  "即使我合法行事，同样的结果也会发生"
  示例：医生虽然误诊，但即使正确诊断，患者也无法治愈
  → 通常可以否定因果关系

假设因果关系抗辩：
  "虽然我的行为导致了损害，但如果我不做，另一个人也会做同样的事"
  示例：被告虽然砍了树，但如果被告不砍，暴风雨也会把树吹倒
  → 通常不能否定因果关系（因为实际上是被告的行为导致了结果）

步骤2：判断假设原因是否已经"启动"
  - 如果假设原因在实际损害发生时已经在进行中 → 可能影响损害范围
  - 如果假设原因纯粹是假设性的 → 通常不影响因果关系认定

步骤3：区分对因果关系的影响和对损害范围的影响
  - 假设因果关系通常不否定因果关系本身
  - 但可能影响损害赔偿的范围和期限

10.4 Cumulative Causation

Scenario Description: Multiple causes are each insufficient to cause the result, but cumulatively lead to the result

示例：多家工厂各自排放少量污染物，单独不足以造成损害，
      但累积排放导致了环境损害

处理方法：
  - But-for检验：移除任一工厂的排放，损害可能仍然发生
    （因为其他工厂的排放仍然超过阈值）
  - 但这不意味着该工厂没有因果贡献
  - 使用"实质贡献"标准：该工厂的排放是否对损害有实质性贡献？
  - 或使用市场份额责任等替代理论

10.5 Expectation Interest Counterfactual in Contract Law

Scenario Description: When calculating damages for breach of contract, it is necessary to construct a counterfactual world of "normal performance of the contract"

步骤1：确定合同的具体内容和履行条件
步骤2：构建"合同正常履行"的反事实世界
  - 假设违约方完全按照合同约定履行
  - 保持其他市场条件、第三方行为等不变
步骤3：计算守约方在反事实世界中的经济状态
步骤4：计算差额 = 反事实世界经济状态 - 现实世界经济状态
步骤5：适用可预见性规则和减损义务进行限缩

特别注意：
  - 市场价格波动的处理：使用违约时的市场价格还是判决时的？
  - 守约方自身的履行能力：守约方是否有能力在反事实世界中获得预期利益？
  - 第三方因素：第三方的行为在反事实世界中是否会不同？

Eleven. Quality Checklist

11.1 Checklist After Completing Counterfactual Reasoning

□ 基础检查
  □ 目标变量是否精确定义？
  □ 结果变量是否精确定义？
  □ 反事实假设是否明确表述？
  □ 背景条件是否完整列出并锁定？

□ 逻辑检查
  □ 是否遵守最小偏离原则？
  □ 推演链的每一步是否有依据？
  □ 是否存在推演链断裂？
  □ 是否考虑了替代推演路径？
  □ 是否检查了过度决定问题？
  □ 是否检查了先占因果关系问题？

□ 偏差检查
  □ 是否存在后见之明偏差？
  □ 是否存在选择性反事实？
  □ 是否存在概率忽视？
  □ 是否存在时间框架错误？
  □ 是否存在滑坡反事实？

□ 法律检查
  □ 反事实推理的结论是否对应正确的法律要件？
  □ 是否区分了事实因果与法律因果？
  □ 是否考虑了法律上的因果关系限缩规则？
  □ 置信度是否满足适用的证明标准？

□ 表述检查
  □ 推演过程是否清晰可追溯？
  □ 置信度是否标注？
  □ 不确定因素是否说明？
  □ 结论是否有适当的限定？

Twelve. Complete Examples

Example One: Simple Scenario—Causation in a Traffic Accident

Facts

At 3:00 p.m. on January 10, 2024, defendant Zhang drove a small car westbound along Jiefang Road in a certain city and, upon reaching the intersection of Jiefang Road and Heping Road, ran the red light through the intersection. At this time, plaintiff Li was crossing the pedestrian crosswalk southbound along Heping Road while the light was green. Zhang's vehicle collided with Li, causing Li's left leg to fracture, 45 days of hospitalization, and 80,000 yuan in medical expenses and 30,000 yuan in lost wages.

At the time of the accident, the weather was clear, the road surface was dry, and visibility was good. Zhang's speed was about 40 km/h (the speed limit on that road section was 60 km/h). The traffic police determined that Zhang bore full responsibility for the accident.

Legal Issue: Was there a causal relationship between Zhang's act of running the red light and Li's damage?

Counterfactual Reasoning Analysis

1. Purpose of Analysis

To determine whether there is factual causation (But-for causation) between Zhang's act of running the red light (tortious act) and Li's personal injury (damage result), as one of the constitutive elements of tort liability.

2. Variable Definition

Target Variable: Zhang's act of running the red light at 3:00 p.m. on January 10, 2024, at the intersection of Jiefang Road and Heping Road
Result Variable: Li's personal injury (left leg fracture and related losses)
Counterfactual Assumption: Zhang stopped and waited at the red light and passed through the intersection after the light turned green

3. Background Conditions Locked

Number	Background Fact	Status
F1	3:00 PM on January 10, 2024	Remains unchanged
F2	Clear weather, dry road surface, good visibility	Remains unchanged
F3	Li was crossing the pedestrian crosswalk along Heping Road from south to north under a green light	Remains unchanged
F4	Zhang's vehicle speed was about 40 km/h	Remains unchanged
F5	Zhang's vehicle braking system was normal	Remains unchanged
F6	The traffic signal at the intersection was operating normally	Remains unchanged

4. Real-World Fact Chain

张某闯红灯进入路口
  → 张某车辆与正在绿灯通行的李某处于同一空间
  → 张某车辆与李某发生碰撞
  → 李某左腿骨折
  → 李某住院治疗45天，产生医疗费8万元、误工费3万元

5. Counterfactual World Inference

前提：张某在红灯时停车等待

第一步推演：张某停在停车线后，不进入路口
  推演依据：遵守红灯信号的直接物理结果
  确定性程度：高 [CF-HIGH]

第二步推演：李某在绿灯状态下安全通过人行横道
  推演依据：路口无其他异常情况（背景条件锁定），
           李某在绿灯状态下正常通行不会遇到来自张某方向的车辆
  确定性程度：高 [CF-HIGH]

第三步推演：张某车辆与李某不会发生碰撞
  推演依据：两者不在同一空间，物理上不可能碰撞
  确定性程度：高 [CF-HIGH]

最终结论：李某的左腿骨折及相关损害在反事实世界中不会发生
  确定性程度：高 [CF-HIGH]

6. Comparative Analysis

Dimension	Real World	Counterfactual World	Difference
Target Variable	Zhang ran the red light	Zhang waited for the red light	Changed
Intermediate Fact	Zhang's vehicle entered the intersection	Zhang's vehicle stopped behind the stop line	There is a difference
Collision Event	Collision occurred	Collision did not occur	There is a difference
Outcome Variable	Li suffered a left leg fracture	Li passed through safely	There is a difference
Degree of Damage	Medical expenses 80,000 + lost wages 30,000	No damage	ΔD = 110,000 yuan

7. Legal Conclusion

If not for Zhang's red-light running, Li's personal injury would not have occurred. Zhang's red-light running is the but-for cause of Li's injury, and factual causation is established.[CF-HIGH]

8. Confidence Assessment

Reliability of the inference chain: High—each step of the inference is based on established physical causation
Certainty of the conclusion: High—the outcome in the counterfactual world is almost certain
Main uncertainty factors: Almost none (this is a typical simple causation scenario)

Example Two: Complex Scenario—Multiple Causes and Loss of Chance in Medical Injury

Case Facts

Patient Wang, a 55-year-old male, visited Hospital A on February 1, 2024, due to persistent chest pain. The attending physician, Zhao, performed an electrocardiogram, which showed mild ST-segment changes, but Zhao did not take it seriously, diagnosed it as "intercostal neuralgia," prescribed painkillers, and sent Wang home.

In the early morning of February 3, 2024, Wang suddenly suffered an acute myocardial infarction at home and was rushed to Hospital B for emergency treatment. Hospital B diagnosed "acute anterior wall myocardial infarction" and immediately performed PCI surgery (percutaneous coronary intervention). Although the surgery successfully opened the blocked blood vessel, because the myocardium had already undergone extensive necrosis, Wang's cardiac function was severely impaired and was assessed as a level-three disability.

Medical appraisal opinion:

Zhao was negligent in the diagnosis on February 1—he failed to recognize the early manifestations of acute coronary syndrome
If the diagnosis had been confirmed and PCI surgery performed on February 1, Wang would have had about a 70% probability of avoiding extensive myocardial necrosis, and his cardiac function might have remained at a normal or mildly impaired level
Even if diagnosed and operated on on February 1, there was still about a 30% probability of similar myocardial necrosis due to severe coronary artery disease
Wang himself had underlying diseases such as coronary heart disease, hypertension, and diabetes, and these diseases were the fundamental cause of the myocardial infarction

Legal issues: Was there a causal relationship between Zhao's misdiagnosis and Wang's harm? If so, what proportion of liability should Zhao (Hospital A) bear?

Counterfactual reasoning analysis

1. Purpose of analysis

To determine the causal relationship between Zhao's misdiagnosis and Wang's harm (large-area myocardial necrosis resulting in a Grade III disability), and to determine Hospital A's proportion of liability. This case involves multiple causes for one result (misdiagnosis + underlying diseases) and probabilistic causation (loss of chance), requiring composite counterfactual reasoning.

2. Variable definitions

Target variable: Zhao's misdiagnosis on February 1, 2024 (failure to identify the early manifestations of acute coronary syndrome and failure to perform further examination and treatment)
Outcome variable: Wang's large-area myocardial necrosis and Grade III disability
Counterfactual assumption: Zhao correctly identified acute coronary syndrome on February 1, immediately arranged hospitalization and performed PCI surgery

Note: This case has two independent damage-causing factors:

Factor A: Zhao's misdiagnosis (delayed treatment)
Factor B: Wang's own underlying diseases such as coronary heart disease (fundamental cause)

3. Locking of background conditions

No.	Background fact	Status
F1	Wang, a 55-year-old male, had coronary heart disease, hypertension, and diabetes	Remains unchanged
F2	Wang had severe stenosis/blockage in the coronary arteries	Remains unchanged
F3	On February 1, 2024, Wang developed chest pain symptoms	Remains unchanged
F4	The medical technology level and PCI surgical conditions at the time	Remains unchanged
F5	Hospital A had the conditions to perform PCI surgery	Needs confirmation (assumed to have them)

4. Real-world chain of facts

王某冠脉严重狭窄（基础疾病）
  + 赵某误诊为肋间神经痛（误诊行为）
  → 王某未获得及时治疗
  → 2天后（2月3日）冠脉完全堵塞，急性心肌梗死发作
  → 大面积心肌坏死
  → 虽经乙医院PCI手术抢救，但心功能严重受损
  → 三级伤残

5. Counterfactual world projection

前提：赵某在2月1日正确诊断，立即安排PCI手术

第一步推演：王某在2月1日接受冠脉造影检查
  推演依据：正确诊断急性冠脉综合征后的标准诊疗流程
  确定性程度：高 [CF-HIGH]

第二步推演：冠脉造影发现严重狭窄，立即进行PCI手术
  推演依据：急性冠脉综合征的标准治疗方案
  确定性程度：较高 [CF-MED-HIGH]

第三步推演（关键分歧点）：PCI手术的效果
  情形A（概率70%）：手术成功，血管开通，心肌未发生大面积坏死
    → 王某心功能正常或轻度受损
    → 不构成三级伤残（可能为十级或无伤残）
    推演依据：医学鉴定意见——70%概率避免大面积坏死
    确定性程度：中等 [CF-MEDIUM]（概率性结论）

  情形B（概率30%）：因冠脉病变过于严重，即使及时手术仍出现大面积坏死
    → 王某心功能严重受损
    → 仍可能构成三级伤残或类似等级
    推演依据：医学鉴定意见——30%概率仍出现类似坏死
    确定性程度：中等 [CF-MEDIUM]（概率性结论）

6. Comparative analysis

Dimension	Real world	Counterfactual world (Scenario A, 70%)	Counterfactual world (Scenario B, 30%)
Target variable	Misdiagnosis, 2-day delay	Correct diagnosis, immediate surgery	Correct diagnosis, immediate surgery
Myocardial necrosis	Large-area necrosis	None/mild necrosis	Large-area necrosis
Disability grade	Grade III	None/Grade X	Grade III or similar
Difference from reality	—	Significant difference	No significant difference

7. Legal analysis

First layer: factual causation (but-for test)

Removing Zhao's misdiagnosis (assuming correct diagnosis and treatment), Wang had a 70% probability of avoiding large-area myocardial necrosis
But there was still a 30% probability of a similar result
Conclusion：The but-for test yields a probabilistic result—the misdiagnosis deprived Wang of a 70% chance to avoid serious harm

Second layer: Analysis of multiple causes and one result

因素A（误诊延误）：使王某丧失了及时治疗的机会
因素B（基础疾病）：是心肌梗死的根本病因

两个因素的关系：
  - 因素B是必要条件（无冠心病则不会心梗）
  - 因素A是加重条件（延误治疗使损害从"可能避免"变为"确定发生"）
  - 两者共同作用导致了最终的严重损害

Third layer: Determination of the proportion of liability

方法一：机会丧失理论
  赵某的误诊使王某丧失了70%的避免严重损害的机会
  → 甲医院应对王某损害的70%承担赔偿责任

方法二：原因力比较
  基础疾病是根本原因（原因力较大）
  误诊是加重原因（原因力较小但显著）
  → 综合考虑，甲医院承担40%-60%的责任

方法三：综合判断（实务中常用）
  考虑因素：
  - 误诊的过错程度（未识别ST段改变，过错明显）
  - 基础疾病的严重程度（冠心病+高血压+糖尿病，病情本身严重）
  - 机会丧失的概率（70%）
  - 患者自身健康管理责任
  → 建议甲医院承担50%-70%的责任

8. Confidence assessment

Reliability of the reasoning chain: Moderate—the key step (surgical outcome) depends on probabilistic medical judgment
Certainty of the conclusion:
- Existence of causation: Relatively high [CF-MED-HIGH](the misdiagnosis did indeed delay treatment)
- Specific proportion of liability: Moderate [CF-MEDIUM](depends on the application of the loss of chance theory and the assessment of causal force)
Main uncertain factors:
- The actual success rate of the February 1 surgery (the expert opinion gives 70%, but this itself is an estimated value)
- Even with timely surgery, Wang's specific degree of recovery is uncertain
- The impact of Wang's underlying disease on long-term prognosis

9. Limitations and explanations

This analysis relies on the conclusion in the medical expert opinion that there was a "70% probability of avoiding extensive necrosis"; if that conclusion is overturned, the causation analysis must be redone
Determination of the proportion of liability involves legal policy judgment, and different courts may have different discretionary scope
The application of the "loss of chance" theory in this case remains controversial in Chinese judicial practice, and some courts may adopt a different analytical framework
It is recommended to further collect the following evidence to improve the certainty of the analysis:
- Whether Hospital A had the conditions to perform PCI surgery on February 1
- The specific severity of Wang's coronary artery lesions (angiography results)
- Statistical data on similar cases

Appendix: Counterfactual Reasoning Quick-Reference Decision Tree

开始：需要判断A与Y之间的因果关系
  │
  ├─ Q1: 是否只有一个可能的原因A？
  │   ├─ 是 → 使用简单But-for检验
  │   │        移除A，Y是否仍然发生？
  │   │        ├─ Y不发生 → A是Y的原因 ✓
  │   │        ├─ Y仍发生 → A不是Y的原因 ✗
  │   │        └─ 不确定 → 标注置信度，结合证明标准判断
  │   │
  │   └─ 否 → 存在多个可能原因A1, A2, ...
  │            │
  │            ├─ Q2: 各原因是否各自独立地足以导致Y？（过度决定）
  │            │   ├─ 是 → 使用NESS检验
  │            │   │        每个原因是否是某个充分条件集中的必要元素？
  │            │   │        ├─ 是 → 该原因是Y的原因 ✓
  │            │   │        └─ 否 → 该原因不是Y的原因 ✗
  │            │   │
  │            │   └─ 否 → 各原因共同导致Y
  │            │            使用But-for检验（移除任一原因，Y不发生）
  │            │            + 原因力比较确定各原因的贡献比例
  │            │
  │            └─ Q3: 是否存在概率性因果关系？
  │                ├─ 是 → 使用机会丧失理论
  │                │        计算目标变量对结果概率的影响
  │                └─ 否 → 使用标准多因分析
  │
  ├─ Q4: 是否存在介入因素？
  │   ├─ 是 → 双重反事实检验
  │   │        (1) 移除原始行为A，Y是否仍然发生？
  │   │        (2) 移除介入因素I，Y是否仍然发生？
  │   │        (3) 介入因素是否"异常"到足以中断因果链？
  │   └─ 否 → 继续标准分析
  │
  └─ Q5: 是否为不作为场景？
      ├─ 是 → 构建"假设作为"的反事实世界
      │        注意：置信度通常较低
      └─ 否 → 使用标准反事实推理流程

About this skill

I. Overview Table

II. Legal Statement

III. Core Concepts

3.1 The Essence of Counterfactual Reasoning

3.2 Definition of Key Terms

3.3 Philosophical Foundations of Counterfactual Reasoning

3.4 The Relationship Between the But-for Test and the NESS Test

IV. Complete Workflow

Stage One: Problem Identification and Framework Construction

Step 1: Identify the need for counterfactual reasoning

Step 2: Precisely define the target variable

Step 3: Precisely define the outcome variable

Step 4: Lock down the background conditions

Stage Two: Construction and Simulation of the Counterfactual World

Step 5: Construct the counterfactual world

Step 6: Derive the outcome in the counterfactual world

Step 7: Compare the two worlds

Stage Three: Extraction of Legal Conclusions

Step 8: Extract legal conclusions from the comparison results

Step 9: Confidence assessment and qualification

Step 10: Integrate into the overall legal analysis

V. Comparison of Common Fields and Legal Sources

5.1 Tort Law Field

5.2 Contract Law Field

5.3 Criminal Law Field

5.4 Administrative Law Field

VI. Verification and Screening Rules

6.1 Validity verification of counterfactual reasoning

Rule 1: Substitutability test of the target variable

Rule 2: Minimal departure principle test

Rule 3: Reliability test of the inference chain

Rule 4: Overdetermination test

Rule 5: Preemption test

6.2 Applicable boundaries of counterfactual reasoning

VII. Output Format Templates

Template A: Standard counterfactual reasoning report

Template B: Brief counterfactual reasoning (embedded)

VIII. Confidence Annotation System

8.1 Confidence of the inference chain

8.2 Overall conclusion confidence

8.3 Correspondence between confidence and standards of proof

IX. Common errors and prevention

9.1 Fatal error table

9.2 Common Pitfalls

Pitfall 1: Hindsight Bias

Pitfall 2: Slippery Counterfactual

Pitfall 3: Cherry-picking Counterfactual

Pitfall 4: Probability Neglect

Pitfall 5: Time Frame Error

Ten. Handling of Special Scenarios

10.1 Scenario of Multiple Causes Leading to One Result

10.2 Counterfactual Reasoning for Omission

10.3 Hypothetical Causation

10.4 Cumulative Causation

10.5 Expectation Interest Counterfactual in Contract Law

Eleven. Quality Checklist

11.1 Checklist After Completing Counterfactual Reasoning

Twelve. Complete Examples

Example One: Simple Scenario—Causation in a Traffic Accident

Facts

Counterfactual Reasoning Analysis

Example Two: Complex Scenario—Multiple Causes and Loss of Chance in Medical Injury

Case Facts

Counterfactual reasoning analysis

Appendix: Counterfactual Reasoning Quick-Reference Decision Tree