When I describe novel ideas to a general audience, I sometimes publish an AI-written companion piece that explains what I am saying in different words, so that more people can grasp the advance on offer.
Earlier today, I published an essay drawing parallels between my father’s experience as a safety-critical engineer in aviation, my own in telecommunications, and the ways in which law can silently fail, producing something more like counterfeit law even when those involved act with professional integrity.
Lexineering: the missing engineering discipline for law
I grew up with the sound of a garage door opening late at night.
It proposes a new engineering discipline, lexineering, to identify and manage the failure modes of law, treating the legal system itself as a safety-of-life operational activity subject to independent engineering assurance.
The idea is subtle because it is easily confused with existing quality-control and oversight mechanisms within the legal system: reviews, appeals, inspectorates and the like. But the proposal is more fundamental.
It asks whether the machinery that produces and enforces legal authority can itself be examined for structural integrity, rather than simply judged according to its own established procedures and assumptions.
In this companion piece, we take another look at the concept and what makes it novel.
I hope it inspires someone to see the legal system differently, and to recognise the gap that can open between the legal system and law itself. The distinction may sound like wordplay, but its consequences for people’s lives can be profound.
To the best of my knowledge, nobody has quite proposed this particular manoeuvre:
Making the production and exercise of legal authority itself the subject of a systematic engineering discipline, capable of identifying common failure modes across otherwise unrelated areas of law.
A packet of minced beef is not beef because the supermarket says so.
It is beef because it came from a cow.
The packaging may carry the correct label, the producer may possess every necessary certification, and the supermarket may have followed every applicable regulation. But if the contents turn out to be horsemeat, the paperwork cannot rescue the claim.
We have a way of finding out. We can test the meat.
Now imagine applying the same question to law.
A court issues an order. It bears the proper seal, cites recognised authorities and is enforceable by the machinery of the state. Nobody disputes that the order is authentic.
But authenticity is not the same as integrity.
A genuine court can produce a defective legal operation, just as a genuine manufacturer can produce a defective product. The difference is that the legal system is also substantially responsible for determining what counts as a valid legal product.
And the machinery responsible for recognising its failures may depend upon the very assumptions, classifications and records that caused those failures.
This creates a peculiar possibility.
The legal system can certify the correctness of an operation without ever testing the condition upon which its correctness depends.
If we can test whether beef is beef,
how do we test whether law is law?
That is the deceptively simple question behind Martin Geddes’s proposal for a new engineering discipline: lexineering.
The missing condition
Imagine a court exercising powers over the property of a bankrupt person.
It issues an impeccably reasoned order concerning a particular asset. The correct statutory provisions have been cited, the procedures followed and the order properly sealed.
There is just one problem: nobody established that the asset belonged to the bankrupt person.
The court may have reasoned correctly about the powers available over such property without adequately establishing the relationship that made those powers applicable to this particular asset.
This is a hypothetical illustration, not a claim about how a particular court would resolve such a dispute. But the logical problem is real.
Correctly applying a power conditional upon something being true is not the same as establishing that the condition is true.
The distinction sounds elementary. Yet sophisticated legal reasoning can conceal it.
Before an institution can decide what the law requires, it must recognise what kind of problem it is dealing with:
Who are the relevant actors?
In what capacities are they acting?
What relationships exist between them?
Which legal states have already been established?
These questions determine which rules apply, which powers become available and which further questions are considered relevant.
Once the classification has been made, an enormous apparatus of legal reasoning can operate upon it. And that apparatus may work perfectly. Every citation may be correct. Every procedural requirement may have been satisfied. The conclusion may follow logically from its premises.
Yet the entire exercise may depend upon a mistake made before the recognised legal question was even formulated.
The machinery can be correct about everything it has chosen to examine while being wrong about what it needed to examine in the first place.
Now consider what an engineer might notice.
The immediate legal question is whether the court made a mistake. But the engineer would ask whether the mistake belongs to a recognisable class of system failures.
An operation has been performed whose authority depends upon a condition. The operation may be perfectly executed, yet the condition has not been established.
That defect is not peculiar to bankruptcy law. It could arise whenever an institution attributes a legal status, exercises a power or transforms someone’s rights without establishing the conditions that permit it.
Different legal disputes may conceal the same engineering defect.
If such defects can be described independently of the particular controversy, they can potentially be tested for, traced through subsequent decisions and designed against.
That is the leap from identifying legal mistakes to developing an engineering discipline for legal integrity.
The self-certifying machine
The difficulty deepens because law does not merely describe an independently existing world. It can also create authoritative legal states.
A court determines a status. The determination enters the official record. Another institution relies upon it. Further decisions are made on that basis.
There are good reasons why legal decisions sometimes become final despite possible mistakes. People need certainty, and disputes cannot continue indefinitely.
But finality and correctness are different things.
The authority of a final decision is not independent confirmation that the premises supporting it were true.
Suppose the original classification or exercise of power was defective.
The resulting legal state may nevertheless acquire institutional authority. Subsequent decisions can rely upon it, and those decisions may reinforce the apparent legitimacy of the original state.
What began as a contested proposition becomes an authoritative premise. That premise supports further conclusions, which enter the official record.
With each iteration, the machinery may become more internally consistent while the original defect becomes harder to recognise.
This is where law differs from minced beef.
A supermarket cannot turn horsemeat into beef by changing its label. But an authorised legal institution can genuinely create legal relationships and statuses through its decisions.
That power is necessary to law. It also complicates assurance, because the system’s outputs can become authoritative inputs to its subsequent operations.
The machinery can incorporate a defective operation into the legal reality against which later operations are judged.
Adding another procedure or appeal does not necessarily solve the problem if the additional mechanism inherits the same assumptions and classifications.
More scrutiny within the wrong frame can
produce greater confidence in the wrong answer.
This does not make appeals or legal safeguards useless. It raises a question about whether they can detect the failures they are supposed to correct.
In aviation, the physical condition of an aircraft provides a reference against which its certification can be tested. A damaged component does not become sound because the maintenance records say otherwise.
A defective legal operation, by contrast, may leave behind precisely the same outward evidence as a successful one. The order is authentic. The procedures were completed. The result is enforceable. The case is closed.
Yet a necessary condition for exercising the authority may never have been established.
This is the lex crash: a failure of legal integrity that may not register as a failure of legal operations.
And the consequences are not distributed equally. An institution may absorb a defective decision as one event among thousands. The person subjected to it may lose a home, a livelihood, liberty or a relationship with their children.
The system can survive an error that the individual cannot absorb.
That is why this belongs in the territory of safety engineering.
From legal arguments to engineering tests
The possibility of recurring structural defects changes what we might look for.
An unsupported ownership attribution in a property dispute and an unauthorised change of legal status in a family dispute may appear to have little in common. They involve different statutes, procedures and bodies of precedent.
But beneath those differences may lie the same structural failure: an institution has performed an authoritative operation without establishing a necessary condition for it.
Once we recognise that common structure, we can begin asking engineering questions:
Where did an authoritative proposition originate?
What evidence supports it?
Who possessed the power to change a legal status, and what conditions governed that power?
Has an allegation quietly become an established fact?
Has an institutional assertion acquired authority without its underlying warrant being examined?
And if a premise is later corrected, which subsequent decisions depend upon it?
These questions suggest methods for examining the integrity of legal operations:
We might test whether the actor, object and relationship are of the kinds to which a claimed power applies.
We might trace an authoritative proposition back to its source, checking whether its meaning and evidential status have been preserved.
We might examine what permitted one legal state to become another, identify the conclusions dependent upon a disputed premise, and test whether correction mechanisms can actually repair the defects the system is capable of producing.
In engineering terms, these are questions of type correctness, provenance, permissible transitions, dependencies and corrigibility.
None requires pretending that law is a computer program or that every legal question has a mechanically decidable answer.
Indeed, one of the most revealing possibilities is that an engineer might be unable to determine who owns a disputed asset, while still being able to demonstrate that a particular exercise of authority depended upon ownership having been established when it had not.
Lexineering need not know the correct answer to every legal dispute to identify that the machinery has not warranted the answer it has produced.
That is a different contribution from offering another interpretation of the law. And it points towards something larger.
Engineering disciplines do not merely identify individual mistakes. They develop reusable descriptions of failure.
An aircraft accident may reveal a design vulnerability relevant to thousands of other aircraft. A software defect may expose an entire class of unsafe operations. A structural failure may reveal a missing constraint in the engineering of buildings.
Lexineering proposes that legal authority may possess
comparable integrity conditions across different areas of law.
If an unsupported attribution, an unauthorised change of status or a failure of correction can be described independently of the particular controversy, lessons from one legal domain may inform assurance in another.
This remains a research proposition to be demonstrated, not a finished technology.
But it marks the difference between diagnosing individual legal errors and developing a systematic means of discovering, classifying and preventing them.
Making law itself the object of assurance
There is a reason this possibility has been difficult to see.
Law is ordinarily the framework within which we examine almost everything else.
A dispute about land becomes a question of property law. A disagreement between businesses becomes a question of contract. A challenge to government becomes a question of administrative or constitutional law.
Even when we question a legal decision, we generally do so through recognised legal categories, procedures and authorities.
There is nothing inherently wrong with this. Legal systems need internal mechanisms for determining and correcting errors.
But those mechanisms are also part of the machinery whose integrity we may need to investigate.
Lexineering makes an unexpected move.
The production and exercise of legal authority itself
becomes the object of engineering assurance.
Instead of asking only whether a conclusion follows from recognised legal premises, we examine whether the entire chain of operations producing that conclusion was warranted.
That includes recognition, attribution, authority, permissible transformations, consequences and correction.
A failure at any of these boundaries can leave the downstream reasoning apparently sound. And a valid-looking conclusion is no substitute for a warranted chain of operations producing it.
Jurisprudence already investigates the nature and foundations of law. Lawyers examine whether particular acts comply with recognised legal authority. Computer scientists can verify whether formal systems behave according to their specifications.
Lexineering proposes to connect such inquiries around a distinct engineering objective:
Assuring the warranted production and exercise of legal authority under actual operating conditions.
A philosopher may investigate the general criteria for legitimate authority. Engineering asks how we could demonstrate that those criteria were satisfied in a particular operation.
This does not mean engineers can impose their own definition of law, or that every mistaken judgment is somehow counterfeit law.
Legal validity, factual correctness, procedural regularity, legitimacy and justice are distinct, sometimes contested properties. Some defects may make an act void. Others may make it appealable, correctable or legally effective despite its shortcomings.
An assurance discipline must distinguish them, make its assumptions explicit and remain capable of finding that the machinery is sound.
Otherwise it would become another means of certifying predetermined conclusions.
The novelty is not that nobody has previously questioned legal legitimacy or institutional failure. It is that the production of law itself has not become a comparably established object of engineering assurance.
And the innovation lies in the possibility of developing reusable tests for structural integrity across legal domains, rather than treating every defect as an isolated controversy to be resolved entirely within its own specialism.
What must the stamp answer to?
Martin’s father was a licensed aircraft maintenance engineer at Heathrow. His little ink stamp certified that an aircraft could return to service.
The stamp did not make the aircraft airworthy.
It represented an assurance that the necessary conditions had been satisfied.
In law, the corresponding problem is more difficult because authorised institutional acts can themselves create legal states. There is no single external laboratory test that resolves every question about legal authority.
But that does not mean there is nothing to test.
We can examine whether a power was conferred, whether its conditions were satisfied, whether the relevant facts and relationships were established, whether a change of legal status was authorised, and whether correction mechanisms can detect and remedy defects.
We can also look for the same structural failure appearing elsewhere, concealed beneath entirely different legal language.
That is where the proposal begins to resemble an engineering discipline rather than a new vocabulary for legal criticism.
The legal system cannot be its own sufficient evidence of correctness simply because it has produced an authoritative result.
The stamp must answer to something beyond the stamp.
We know how to ask whether an action complies with the law. Lexineering asks whether we can establish that the machinery determining compliance has itself acted with warranted authority.
And it asks whether the failures of that machinery can be made sufficiently explicit, traceable and repeatable to be engineered against.
If we can discover common structural defects beneath apparently unrelated legal disputes, then law itself becomes something we can begin to engineer for integrity—not merely argue about after it fails.
That is what lexineering reveals.