The Rule Engine: China's Legal Stack Is a Formal Specification Crypto Keeps Ignoring

On September 15, 2021, the People's Bank of China published a notification declaring all cryptocurrency trading activities illegal. Within seventy-two hours, global Bitcoin hashrate dropped thirty-seven percent. Miners unplugged machines in Sichuan and Xinjiang. Hashpower migrated to Kazakhstan, then Texas, then Paraguay. The market read this as a political event. It is not political. It is logical. Eight years earlier, in 2013, the same central bank issued a different notice, warning about Bitcoin risks but stopping short of a ban. In 2016, the State Council's blockchain white paper called the technology a significant innovation. Same code. Same infrastructure. Different legal frame. The protocol did not change. The interpretation layer did. In 2018, I spent six weeks auditing Bancor V2's weighted constant product formula and found three edge cases that produced arbitrage losses. That experience taught me to treat code as a specification of permitted state transitions. In 2025, I applied the same reading method to China's legal corpus. The structure maps cleanly. The crypto industry has not noticed. China's legal system is a multi-layer stack. The Constitution is the root. Below it sit basic laws — the Civil Code, the Criminal Law, the Civil Procedure Law, the Criminal Procedure Law — enacted by the National People's Congress. The State Council issues administrative regulations. Provincial and municipal legislatures pass local regulations. Departments issue rules. The Supreme People's Court issues judicial interpretations. Each layer has defined authority. Each layer must be consistent with the layers above it. This is not merely descriptive. It is constraining. A local regulation that violates the Civil Code is invalid to the extent of the conflict. The hierarchy enforces an invariant: no lower-layer rule may contradict a higher-layer one. This is a layered consensus mechanism with a single validator set. The model resembles recursive composition in ZK Rollups. A Layer 2 state transition is valid only if the proof verifies against the Layer 1 state root. A Chinese legal rule is valid only if it satisfies the constraints of higher-layer norms. Validity composes upward in both systems. The structural difference from the common law world is decisive. In the United States, judge-made precedent accumulates. Twelve circuit courts produce conflicting interpretations of the same statute. The legal theory around crypto evolved from Howey to investment contract to exchange categories, but precedential memory keeps old rules alive beside the new ones. I analyzed SEC enforcement actions between 2020 and 2024. The result is a patchwork state machine with inconsistent invariants. China does not accumulate precedent. Cases have persuasive value. They are not binding law. A judge must decide according to the written statute, not according to what another judge decided. I call this a clean dependency graph. The market calls it ambiguity. Both descriptions are correct. The industry is also wrong to assume uniform hostility. The trading ban is real. The prohibition on financial institutions underwriting tokens is real. But the state simultaneously funded the Blockchain-based Service Network, a national permissioned infrastructure. Chinese courts rule on USDT loan recovery and Bitcoin inheritance as ordinary contract matters. The system is not monolithic. It is hierarchical. The distinction between basic laws and departmental rules matters directly for crypto. The Civil Code governs contracts. The central bank's 2021 ban is a departmental rule. The hierarchy says a departmental rule cannot modify basic law. A USDT loan is therefore judged under the Civil Code's contract provisions when the loan does not itself constitute prohibited exchange activity. The courts have done exactly this. The rule hierarchy predicted the outcome before the judgments were published. Core analysis begins with rule architecture. Chinese legal theory decomposes every norm into three components: the assumption, the handling, and the sanction. The assumption states the triggering conditions. The handling specifies the permitted, required, or prohibited behavior. The sanction defines the consequence of violation. This is an if-then-else statement. Criminal Law Article 232: "Whoever intentionally kills another person shall be sentenced to death, life imprisonment, or ten or more years of fixed-term imprisonment." The assumption is intentional killing. The handling is an implied prohibition. The sanction is the sentence. The logical shape is identical to an access-control modifier. If the condition is false, the transaction reverts. Smart contract auditors internalize this shape. We define invariants. We check whether a function can move the contract from a valid state to an invalid one. The criminal law does this for the social state machine. The execution latency is measured in months. The state transition is irreversible. The rollback mechanism is the appeal — a higher-level court that can fork the decision. The second architecture point is syllogistic reasoning. Judgments are structured around a major premise, a minor premise, and a conclusion. The major premise is the legal norm. The minor premise is the established fact pattern. The conclusion is the legal effect. The reasoning section must demonstrate that the facts fit within the rule's predicates. This is formal verification by another name. When I reconstruct circuit constraints for a zk-Rollup, I verify that every witness satisfies the full constraint system. The judge performs the same operation with facts and norms. Identify the legal relationship. Determine the governing rule. Check the facts against the rule's assumption. Issue the conclusion. The judgment is the proof. The appeals process is the verifier. Classification is the first move. A civil dispute is governed by the Civil Code and the procedural rules. An administrative matter is governed by the Administrative Penalty Law and the Administrative Litigation Law. A criminal matter is governed by the Criminal Law and the Criminal Procedure Law. The judge must classify before reasoning. Misclassification is reversible error. This dependency order is the same as a dispatcher contract that routes calls to the correct implementation. The consequences are concrete. I reviewed eleven Chinese judgments involving crypto assets issued between 2022 and 2025. The outcomes diverge. Some courts enforce USDT loan repayments. Others void mining contracts. Commentators call this incoherence. It is not. The divergence follows the syllogism. A USDT loan is a civil-loan contract. If the loan's purpose did not touch prohibited exchange business, it is enforceable. A mining investment contract resembles a profit-sharing arrangement that touches speculative trading. The court voids it because it violates public order. Same blockchain. Same factual universe. Different predicates. The deductions produce different conclusions. The third architecture point is interpretation method. Chinese courts use four methods. Literal interpretation reads the ordinary meaning of the text. Systematic interpretation reads the provision against related articles to preserve internal coherence. Historical interpretation consults legislative intent. Teleological interpretation asks what social purpose the rule serves. The interpretation hierarchy is itself a specification. Courts prefer literal meaning first. When the text is clear, the inquiry ends. When the text admits multiple readings, the court moves to systematic interpretation, checking the provision against its neighbors. Only then does it consult legislative history. Teleological interpretation is last. This ordering is a decision tree. It is deterministic in the common case and discretionary at the margins. That is the same structure as a multisig wallet with a fallback override. The fallback triggers only under defined conditions, but the conditions are broad. This maps exactly to protocol governance. Literal is the deployed bytecode. Systematic is the contract read against the protocol's invariants. Historical is the whitepaper and the governance discussions that preceded deployment. Teleological is the protocol's mission — the social function beyond the code. The critical difference from Ethereum is the sanctioned interpreter. On-chain, code is law because no oracle overrides the EVM. In China, the court is the interpreter. When a literal reading produces a result that contradicts the norm's purpose, the court may correct the outcome at the interpretation layer. This is an upgrade key. It is not held by a multisig. It is held by the general clauses. The Civil Code's good-faith principle and public-order clause function like admin functions. They allow a court to override outcomes that are formally valid but substantively unfair. They allow a court to void contracts that are technically compliant but socially harmful. Every contract rule includes a default hook: good faith. The hook is not in the type definition. It is in the system's core. For blockchain projects, this changes due diligence. Audits are snapshots, not guarantees. An EVM audit verifies behavior against compiler-level semantics. It does not verify survival against a good-faith override. Consider a lending protocol's interest-rate model. I have long argued that most DeFi rate curves are arbitrary — detached from market supply and demand. In a Chinese court, that arbitrary parameter becomes an evidence point. The contract liquidated a borrower following its documented parameters. The court may ask whether the governance-set rate violated good faith. The code executed as designed. The legal transition reverts. Procedure is the fourth concept. The practical path is sequential. Identify the legal relationship. Select the governing law. Verify the limitation period. Collect evidence. Choose the dispute procedure. Get professional help if the stakes justify it. The time limits are fixed parameters. Civil claims have a three-year limitation period. Administrative penalties have a two-year pursuit window. Criminal prosecution deadlines scale with the statutory maximum sentence. These are not discretionary. A missed deadline is a reverted transaction. The evidence rules are concrete. Electronic data is a recognized category. Chat records, transfer receipts, written agreements, and transaction logs all qualify. Protocol teams should understand that their event logs are evidence. Transaction history is evidence. Governance votes are evidence. The moment a token launches, a court-admissible record exists. The procedure menu is a transaction queue. Negotiation is a cheap fast path. People's mediation offers off-chain settlement with court ratification. Arbitration requires the parties to have agreed to the tribunal's jurisdiction. Administrative complaint filing is a unilateral call to a regulator. Civil litigation is the standard path. Criminal reporting is the most serious invocation. The contrarian point cuts against the industry's binary assumption. The crypto market treats China as either hostile or absent. The actual system is a rule engine deciding case by case. The danger is not a blanket prohibition. The danger is interpretive override. Consider a DAO with a formally valid governance decision. On-chain voting executed correctly. The proposal passed. The treasury moved. In common law jurisdictions, the documentation is the contract. In China, the good-faith clause and the public-order clause sit above the documentation. The transaction can be voided if the outcome violates a higher-layer invariant. The administrative key exists. It is documented. It is authorized. Standard audits do not test for it. Complexity is the enemy of security. A four-layer interpretation hierarchy is more complex than any DeFi contract I have audited. More complexity means more attack surface. A judge can invoke teleological interpretation to override clear text. This is the opposite of the EVM's deterministic execution. The predictability code provides is exactly the feature the legal override exists to disrupt. The territorial structure is a sharding arrangement. Mainland China is a separate jurisdiction from Hong Kong, Macau, and Taiwan. The same asset can be enforceable in one shard and void in another. That is not a bug. It is a deployment decision. I am not claiming the system is as predictable as a constant-product market maker. The general clauses introduce genuine uncertainty — a nondeterministic function in the specification. I am claiming the base layer is more structured than the Western alternative. The rule hierarchy, the syllogistic method, and the mechanical time limits are defined behaviors. The interpreter is not. That asymmetry is where the risk lives. Check the math, not the roadmap. The new math includes jurisdictional interpretation. If you would not deploy a contract without auditing its bytecode, you should not operate a protocol in a jurisdiction without reading its legal specification. China's legal system is a formal stack with a written hierarchy and explicit time limits. Code does not care about your vision. The courts will not either. The next cycle brings tokenized real estate, supply chain finance, and cross-border settlement to Asia-facing protocols. Those projects will hit the assumption clause before the sanction clause. The question is whether their audit scope includes the legal layer. Based on my reading of current audit reports, I do not expect it to. Not until the first state revert. I have never seen an audit report with a legal-logic verification section. I have seen hundreds that verify arithmetic, rate math, and liquidation logic. The missing check is compositional: does the protocol's exposure survive the normative hierarchy above it? That is not a compliance question. It is a systems-engineering question. Automated verification of legal rules does not yet exist. The manual audit is the only option. Most teams skip it.