Why does written natural language remain a primary authoritative form of law?
Writing note: This article uses a controlled technical writing style inspired by ASD-STE100. It does not claim full ASD-STE100 compliance.
1. Law Uses Text
Most modern legal systems use written natural language as a primary authoritative form. In this article, authoritative form means a form that relevant legal institutions give legal authority.
Legislatures approve statutes. Governments issue regulations. Courts publish judgments. Public authorities apply these legal sources.
Law is not only text. Legal systems can also depend on precedent, custom, conventions, and institutional practice. The exact sources differ between legal systems.
Computers changed how people store, search, copy, compare, and distribute legal text. These changes improved access to legal sources.
They did not necessarily change the authoritative form of law. A digital statute is still natural-language text.
A database can help a person find an applicable legal source. A person must still determine what that source means for a specific case.
2. Written Language Is a Technology
Written law is familiar. This familiarity can make written language seem necessary.
A long history of written law does not prove that law must use written language. Written language can be a technology that law uses.
A legal system must contain information about rights, duties, permissions, prohibitions, powers, procedures, conditions, and consequences.
Natural-language text can express this information. This does not prove that natural-language text is the only possible form.
A sentence can express a right. A paragraph can express a duty.
This does not show that the text and the legal relation are the same thing.
This gives the main question:
Must law use written language, or is written language one technology that law uses?
3. Law Also Uses Interpretation
A legal source does not apply itself. A person or institution must connect the legal source to facts.
Some cases are simple. Other cases contain unclear facts, conflicting sources, broad terms, exceptions, or legal discretion.
Interpretation is not only an error caused by natural language. Some interpretation is part of legal operation.
A term can have more than one reasonable meaning. Two legal sources can point in different directions.
A court can give weight to purpose, precedent, or principle. A public authority can have legal discretion.
Interpretation can therefore change a legal result.
A computational legal system cannot only store legal text. It must also show where interpretation enters the process.
It must show when more than one interpretation is possible. It must show when a legal result depends on discretion.
It should also separate source text, interpretation, discretion, and selected facts.
If a machine hides these choices, the machine does not remove interpretation. It only moves interpretation into another layer.
A machine can therefore make interpretation less visible, even when it makes the final result look more precise.
This gives a second question:
Can a legal system make interpretation more explicit without pretending that interpretation can disappear?
4. The Computer Did Not Replace the Document
Computers changed legal work. They made legal text easier to store, search, and compare.
Artificial intelligence can now analyze and discuss legal text. However, authoritative legal sources still appear primarily as text.
The computer operates on the document. The document remains central.
Science fiction often keeps the same structure. Star Trek imagined faster-than-light travel, transporters, artificial persons, and holographic doctors.
Its legal stories still use codes, precedent, hearings, lawyers, and judges.
In Star Trek: Voyager, Captain Janeway reads the Federation Judicial Code and checks legal precedent.
In The Measure of a Man, a hearing examines the legal status of Data.
The computer provides information. Humans interpret the law.
The technology changed. The legal form did not change in the same way.
This can be a reasonable design. Law can need context, interpretation, and judgment.
However, familiarity is not sufficient evidence that the design must remain unchanged.
A future legal system can still use documents. The use of documents should be a design choice, not an automatic inheritance.
5. Law as Code Starts With Text
Current Law as Code work already asks how legal rules can become machine-readable or machine-executable.
The OECD's 2026 Law as Code consultation uses authoritative legal text as the legally binding source.
Its model keeps machine-executable representations linked to that text. It also keeps interpretation, discretion, and evaluation visible.
This is a careful design choice.
This article asks an earlier question:
Why must authoritative legal text be the starting point?
The question does not reject Law as Code. It asks whether the root representation must always be written natural language.
6. Can We Ask the Law?
A person can search the law. A person can read the law. A person can ask a lawyer about the law.
A public authority can apply the law. These actions are different from a direct query to the legal system.
Consider a system with a current authoritative form of law.
The system receives legally relevant facts. It identifies applicable legal sources, exceptions, missing facts, interpretations, discretion, and uncertainty.
The system then gives a legal result or states that one result is not determined.
The result shows the applicable legal sources. It also shows the relevant facts, assumptions, interpretation, and uncertainty.
This would be different from a normal chatbot. A chatbot can give an opinion about law.
An authoritative query system would give an official legal result from authoritative legal sources.
Such a system must also show its limits. It must be able to give this result:
The available facts and law do not determine one legal result.
This is not necessarily a system failure. It can be a valid legal condition.
A useful system must show uncertainty. It must not hide uncertainty.
7. The Citizen and the Authority
Law applies to citizens and public authorities.
The same legally relevant facts should not produce different legal rules for each side.
A citizen and an authority can disagree about facts. They can disagree about evidence.
They can disagree about legal interpretation when the law permits interpretation.
They should not use different hidden versions of the law.
This gives a design principle:
A citizen and a public authority should have access to the same authoritative legal sources. They should also see the same interpretation layer.
They should also be able to submit the same facts to the same legal system.
If the facts are the same, the system should identify the same applicable sources.
If the legal result differs, the system should show why.
The reason can be a different fact. The reason can be a different permitted interpretation.
The reason can be an exercise of legal discretion.
The reason should not be an invisible difference between two legal systems.
This principle does not remove lawyers, courts, or human judgment.
It reduces one specific problem.
It reduces differences that result only from unequal access to law and its operational interpretation.
This gives a stronger question:
Can a citizen and a public authority use the same authoritative legal system?
8. If We Designed Law Today
Assume that written law does not exist. A society must create a legal system.
The system must contain rights, duties, permissions, prohibitions, powers, procedures, conditions, and consequences.
The society has computers, networks, databases, cryptography, and machine learning.
It can store structured information. It can link rules to sources. It can record changes and preserve old versions.
It can test rules against cases. It can show dependencies, interpretation, and uncertainty.
It can provide different interfaces to the same legal information.
Now ask a design question:
Would natural-language documents be the native form of this legal system?
Perhaps they would. Natural language can express context, uncertainty, values, and judgment.
These properties are important. However, they do not prove that every part of law must use the same form.
Some legal rules are highly structured. Other legal rules need interpretation.
A legal system could use different forms for these different functions.
Human-readable text could support human understanding. Structured rules could support computation.
Test cases could support verification. Version records could support review.
A visible interpretation layer could show where judgment enters the system.
These forms would need clear rules for authority and conflict.
The important point is not that legal text should disappear.
The important point is that its role should be a deliberate design decision.
9. A Question Before a Proposal
This article does not propose an AI judge. It does not propose machine-made law.
It does not propose the removal of courts. It does not propose the removal of legal text.
It asks an earlier question:
What makes a form of law authoritative?
Legal theory does not give one agreed answer.
Some theories connect legal validity to recognized social sources and practices.
Other theories give legal interpretation and principles a larger role in the content of law.
This article does not try to resolve that debate.
It asks whether written natural language must be part of the answer.
A future computational legal system would create a second question.
Such a system can need transparency, review, correction, and a method for challenge.
These are design requirements for a system that people can inspect and contest.
They are not necessarily conditions for the existence of law.
The two questions must remain separate:
What gives something legal authority?
and:
What safeguards should govern an authoritative computational legal system?
This gives the final question:
Is written language necessary for legal authority, or is it one technology through which legal authority is expressed and interpreted?
Before we replace written law, we must know what properties we need to preserve.
Before we assume that written law must remain, we should know why.
— Dennis Hedegreen, trying to see the structure
Sources and notes
-
ASD-STE100, Issue 9, 15 January 2025. The standard defines a controlled natural language for technical documentation.
ASD-STE100 official website -
OECD, Consultation on the digital provision of law: Towards a shared reference framework for Law as Code, 2026.
The consultation keeps authoritative legal text as the legally binding source. It also keeps interpretation and discretion visible.
OECD consultation -
Star Trek: Voyager, “Tinker, Tenor, Doctor, Spy.” Janeway reads the Federation Judicial Code and reviews legal precedent for holograms.
Voyager episode transcript -
StarTrek.com material on The Measure of a Man. The episode uses a legal hearing to examine Data's status and rights.
StarTrek.com article