Code After — Frequently Asked Questions
Code After — Frequently Asked Questions
Code After is a research and publishing project on how artificial intelligence is reshaping the institutions that govern modern life — law, accounting, language, evidence, work, and public authority. It treats AI not as a technology in isolation, but as an institutional event with consequences for how rules are made, value is measured, responsibility is assigned, and authority is exercised.
AI is a general-purpose decoupling force. It separates governance from execution, rules from systems, and meaning from measurement. Institutions built for an earlier substrate — human-paced, territorial, deterministic — strain against systems that are machine-paced, transnational, and probabilistic. Professional services firms have moved into the gap. They are now the de facto constitutional layer of AI governance.
Because what AI changes is institutional, not only technical. The rules of evidence, the categories of property, the methods of measurement, the conditions of credentialing, the boundaries of jurisdiction — these are the surfaces AI reshapes. A technical event changes what machines do. An institutional event changes what humans can do with the institutions they inherited.
Code After: Law, Accounting, and the Governance of Artificial Intelligence* (v0.9, April 2026) is the founding text of the project. Open-access on Zenodo with persistent DOI [10.5281/zenodo.19537473](https://doi.org/10.5281/zenodo.19537473). Licensed under Creative Commons Attribution–NonCommercial–NoDerivatives 4.0. It diagnoses four structural gaps through which AI governance fails and proposes a Third Architecture designed to close them.
The gap between AI systems and the institutions meant to govern them. Existing laws, accounting standards, regulatory tools, and professional practices were built for human organisations operating on human timescales within territorial borders. AI systems are none of these. The mechanism behind the failure is decoupling — the systematic separation of governance instruments from the operational reality they were designed to reach.
Visibility, Rule-Execution, Categorical, and Measurement. Each names a point at which an instrument built for an earlier technological substrate fails to engage the systems it was designed to govern. Each has been independently recognised by readers in jurisdictions the manuscript did not address. The diagnostics travel.
Sovereigns cannot see what they govern. AI systems are technically opaque, jurisdictionally fragmented across borders, and temporally accelerated beyond legislative cycles. The gap has three dimensions and no single fix. Better hiring does not close it. Faster processes do not close it. International harmonisation produces principles, not execution. The Visibility Gap is structural, not transitional.
Written rules do not reach deployed systems. The distance between legislative intent and technical implementation is filled — necessarily — by professional services firms translating abstract principles into operational templates. Statutes constitute. Translation operationalises. Without the Translation Layer, rules do not arrive at the systems they name.
Legal concepts cannot attach to computational objects. Personhood, property, agency, intent, control — these were drafted for humans and the artefacts humans made. AI systems decide, classify, recommend, and generate without fitting the categories. The compiler throws errors. Until the categories are rebuilt, the law speaks past the systems it claims to govern.
Industrial-era accounting cannot price probabilistic, self-updating intelligence. AI models, data assets, compute infrastructure, and algorithmic capability resist the recognition rules of GAAP, IFRS, and CAS — and the three regimes are diverging in how they handle the resistance. The same model becomes a capital asset in one jurisdiction, an expense in another, and state infrastructure in a third. Three balance sheets. Identical physics.
The professional services ecosystem — global law firms, the Big Four, technical consultancies, and the standard-setting bodies adjacent to them — that converts state policy into model-level specifications, audit-grade documentation, and compliance templates. Sovereigns set rules. The Translation Layer determines what those rules mean inside the systems. Code After argues this layer has become the de facto executive branch of the AI economy.
The structural response Code After proposes. After the Westphalian sovereign and the global multilateral, a third construction is required to govern systems that respect neither. The Third Architecture is institutional infrastructure designed for governance at machine speed — deliberately built rather than retrofitted from existing instruments. It is the constructive answer to the four gaps.
The framework's diagnostic tool for predicting which governance treatments will travel and which will not. Expressed as **I = V × W × N**: a treatment's incorporation weight is the product of its **Visibility**, its **Workability**, and its **Necessity**. The structure is multiplicative — failure on any one variable is dispositive. The heuristic is applied throughout v0.9 and reappears at the linguistic level in *Code After Language*.
A six-paper research series extending the v0.9 framework into the major institutional domains AI is reshaping: language, law, accounting, evidence, work, and public authority. Open-access publication on Zenodo, bilingual English–Chinese from launch, with additional language editions to follow where genuine collaborators are present. Paper 1 — *Code After Language* — targets release in September 2026.
The structural condition under which a jurisdiction's language is not the language in which AI systems are trained, documented, benchmarked, or governed. It sits upstream of the four gaps of v0.9. A regulator who cannot read a model card in the language of governance cannot audit. A court that cannot examine training documentation in the language of law cannot adjudicate. The Gap cannot be closed by exercising existing authority. It can only be closed by constructing leverage.
The proposed framework for closing the Linguistic Gap. Three layers — **Visibility** (corpora and benchmarks), **Workability** (model performance in the language at governance-grade tasks), and **Necessity** (procurement and market-access conditions that make the language a requirement). The three layers must be built together. Hong Kong, with bilingual law and trilingual professional capacity built over decades, is the closest existing approximation of that architecture. Evidence that the construction is possible. Not a model to be copied wholesale.
Citizens, professionals, policymakers, scholars, regulators, courts, educators, and the institutions that serve them — in the G3 economies (United States, China, European Union) and in the one hundred and seventy jurisdictions outside them. The framework is general. The application is local. The series is structured so each paper can be read on its own and read in its own language.
Citation: Yan, N., Yan, J., & Yan, R. (2026). Code After: Law, Accounting, and the Governance of Artificial Intelligence(v0.9).
Zenodo. https://doi.org/10.5281/zenodo.19537473
Licence: Creative Commons Attribution–NonCommercial–NoDerivatives 4.0 International (CC BY-NC-ND 4.0). The work is freely shareable with attribution. Commercial use and derivative works require permission.
Project home: codeafter.ai