A field book · argument as education

Arguendo

How the world learned to argue in mock courts — and what changes now the machine reads the brief too.

↓ turn the page

Prologue

For the sake of argument

A student rises in a borrowed courtroom, addresses three judges who are not judges, and argues a case that does not exist. Everyone in the room knows this. It is still one of the most serious things a law student ever does.

The dispute is invented. The state of Arkania has never sued the state of Novaris, because neither state is on any map. The satellites that burned up in the fictional atmosphere never flew. And yet the young woman on her feet has read every word of the treaties she is about to cite, has anticipated the question the judge is forming behind her folded hands, and will answer it — or fail to — in the next ninety seconds. The stakes are unreal and the pressure is not.

This is a moot court: a simulated hearing in which students argue both sides of a hypothetical legal problem, in writing and aloud, before a bench of practitioners, academics, and sometimes real judges. It is the oldest continuous method in legal education, older than the lecture, older than the case-book, older than the law school itself. And it is, right now, being quietly upended by a technology that can read the whole problem in a second and hand back the argument the student spent three weeks assembling.

This book is about both facts at once. It is a field guide to how mooting has been run across the world — its rituals, its great competitions, its craft — and an honest reckoning with what artificial intelligence does to a discipline whose entire purpose is to make a person struggle productively with a hard question. The two halves belong together. You cannot see clearly what the machine changes until you understand what the tradition was for.

The stakes are unreal and the pressure is not. That gap is where the learning happens.

We begin, as the moot itself began, a long way from any computer.

Part One

The Tradition

Six centuries of arguing to learn — the ritual, the circuit, and the case for it.

§ · §

Chapter I

Origins: the moot before the mock

The word is older than the courtroom it now describes. To understand the moot, start with an assembly of arguing men in a medieval hall, and a profession that decided the best way to teach law was to make you do it.

The English word moot comes from the Old English ġemōt — a meeting, an assembly, a gathering called to decide something. Long before it meant a practice trial, it meant the thing a community did when it came together to settle a matter by speech. That older sense survives in the adjective we still use loosely — a "moot point" is one fit for debate — and it is the right root to keep in mind. A moot was never a test you sat quietly. It was a meeting you argued in.

Its home was the Inns of Court in London — Gray's Inn, Lincoln's Inn, the Inner and Middle Temple — the guild-like societies that from the fourteenth century onward trained England's barristers. There was no examination in the modern sense. A student learned law by living in the Inn, listening to senior members deliver readings on a statute, and taking part in moots: formal exercises in which junior members argued points of law arising from a hypothetical case, while the benchers — the senior lawyers who governed the Inn, and from whom we still get the word "bench" — presided and corrected. Advancement through the profession was measured partly in moots argued. To be called to the Bar was, in part, to have argued your way there.

Moot. From Old English ġemōt, an assembly convened to decide. In legal education, a simulated argument on a hypothetical case, conducted for training rather than judgment. The related "mock trial" simulates a jury trial with witnesses and evidence; a moot argues law — usually an appeal — on agreed facts.

Two features of that origin have never left the practice, and both matter for what follows. First, the moot was participatory: you did not watch advocacy, you performed it, and were corrected in the doing. Second, it was oral before it was written. The medieval moot was a speaking exercise, a rehearsal for a profession that lived by its voice in open court. The written brief came later, as legal systems grew document-heavy, but the beating heart of the moot remained the person standing up to answer a question they could not fully predict.

From guild ritual to global classroom

When legal education migrated into universities — first in the common-law world, then far beyond it — the moot came along, and shed its guild trappings for a competitive, curricular form. The twentieth century turned it from an in-house exercise into an international sport of the mind. The founding of the great cross-border competitions after the Second World War (we will meet them in Chapter III) gave the moot a new stage: not the Inn's hall but the simulated international tribunal, with teams flying between continents to argue invented disputes under real judges.

By the time the practice reached the law schools of India, Africa, continental Europe, Latin America, and East Asia, it had become something the medieval bencher would half-recognise and half-marvel at: a worldwide culture of students, most of whom will never practise the exact law they argue, training their reasoning against one another across borders and legal traditions. The ritual survived every change of setting because the thing it teaches never went out of demand.

14th–15th c.

The Inns of Court train barristers through readings and moots; the benchers preside. "Moot" and "bench" enter the vocabulary of advocacy.

19th–20th c.

Legal education moves into universities; mooting becomes a curricular and competitive exercise rather than a guild rite.

1960

The Jessup competition is founded — the moot goes international, simulating the International Court of Justice.

1990s–2000s

An explosion of specialised international moots and a deep national moot culture in India and elsewhere. The circuit becomes a career on-ramp.

2023 →

Generative AI enters the workflow. Competitions begin writing rules for a tool that can read the whole problem in seconds.

Chapter II

Anatomy of a moot

Strip away the variations between competitions and every moot has the same skeleton: a problem, two written cases, and an oral contest before a bench. Learn the skeleton and you can read any moot in the world.

Every moot begins with the problem — a document, released weeks or months ahead, that sets out an invented dispute and the questions the tribunal must decide. In an international-law moot styled on the International Court of Justice, it takes the form of a compromis or special agreement: the two fictional states jointly agree to submit their quarrel, and the document lays down the facts they have agreed and the issues left for the court. Everything the competitor may argue lives inside this document and its annexes. Nothing outside it is fair game. The problem is a closed world, and that closedness is the single most important fact about a moot — we will return to it more than once.

The compromis (or "the problem", "the proposition"). The agreed statement of a moot's invented dispute: the facts the parties concede, the annexed documents that form the record, and the legal issues referred for decision. In a special-agreement case, whatever the parties have jointly stated cannot be contested; everything else can. A competitor's first task is to mark which is which.

The written round: memorials

Before anyone speaks, each team files its written submissions — usually called memorials (in international moots) or briefs, factums, or memoranda elsewhere. Crucially, most competitions require a team to write for both sides: the Applicant and the Respondent, the Claimant and the Defendant. You do not get to believe in your case. You must construct the strongest version of each, which is the first hard lesson mooting teaches — that a good argument is not the same as a true belief, and that understanding your opponent's best case is the price of beating it.

A memorial is a formal document with a fixed anatomy: a cover page, a table of contents, an index of authorities (every treaty, case, and text you cite, listed to a citation standard — the Bluebook, OSCOLA, or a competition's own rules), a statement of jurisdiction, a statement of facts, the issues, a summary of argument, the body of argument, and a prayer for relief. It is marked to a rubric that rewards knowledge of the law, structure and clarity, the depth and use of authority, and — worth real marks — correct format and citation. A brilliant argument in the wrong citation format loses points, on purpose: the profession the moot is training runs on precision.

You do not get to believe in your case. You must build the strongest version of each side.

The oral round: the bench

Then the competition proper: teams meet in successive rounds — preliminary, then knockout brackets of quarter- and semi-finals, then a grand final — and argue aloud. Each side fields two speakers (oralists), with a third team member, the researcher, present but silent, passing notes and marshalling authorities. A draw of lots assigns sides. A clock runs. And the bench — the panel judging the round — does the thing that makes oral advocacy unlike any written exercise: it interrupts.

A "hot bench" is one that questions relentlessly, and learning to be interrupted is the craft's steepest climb. The speaker must yield to the question, answer it directly, and return to the thread without losing the time or the argument. The best oralists seem to welcome the hardest question, because they have already heard a worse version of it in practice. The round is scored on knowledge, responsiveness to questions, structure, and courtroom manner — poise, courtesy, the ability to concede a point gracefully and still win the war.

The roles

Two who speak, one who knows where everything is

A standard team is three: two speakers, who argue, and a researcher, who does not — but who has read more of the record than anyone, prepares the authorities, and during a round finds in seconds the paragraph the speaker needs. Many competitions award a separate prize for Best Researcher, and some set a written research test. The researcher is the moot's unglamorous engine, and its best training for the reality of practice, where the person who can find the right authority fastest is quietly indispensable.

Woven through all of it is a set of integrity rules that the competitor ignores at their peril: strict anonymity (memorials identify the team only by a code, never by name or institution, so judges cannot favour a school), page and word limits enforced by penalty, and plagiarism thresholds. These are not bureaucratic fussiness. They are the guarantees that make the contest a contest — and, as we will see, they are exactly the rules that generative AI strains hardest.

Chapter III

The global circuit

A handful of great competitions turned a classroom exercise into a world championship of legal reasoning — each simulating a different tribunal, each with its own culture. And beneath them, national ecosystems that feed thousands of students a year.

If the moot has a capital city, it is wherever the Jessup is being argued that spring. The Philip C. Jessup International Law Moot Court Competition, founded in 1960 and run by the International Law Students Association, is the largest moot in the world: teams from scores of countries argue a public-international-law dispute before a simulated International Court of Justice, winning through national and international rounds to a global final. The Jessup is where most law students first meet the compromis, the memorial, and the hot bench in their international form, and its problem each year is a genuine event in the field — a barometer of which questions of international law feel live.

Around it orbits a constellation of specialised competitions, each mimicking a different forum:

The great competitions

A tribunal for every kind of argument

The Willem C. Vis Moot (Vienna, from 1994, with a parallel Vis East in Hong Kong) simulates an international commercial arbitration under the CISG — the culture of the arbitration bar, less about a hot bench than about the subtle management of a tribunal.

The Manfred Lachs Space Law Moot Court Competition, named for the great ICJ judge, argues the law of outer space — treaty-thin, doctrinally young, and argued largely by analogy to general international law.

The ICC Moot (The Hague) simulates the International Criminal Court; the Nuremberg Moot takes on international criminal law in the city of the trials; the Telders is Europe's Jessup-analogue; the Price Media Law moot argues freedom of expression; the Frankfurt Investment Arbitration Moot takes on investor-state disputes.

Each competition breeds a distinct temperament. Arbitration moots reward a quieter, more procedural cunning; a criminal-law moot demands moral seriousness; a space-law moot rewards a competitor comfortable arguing at the frontier, where the treaty runs out and everything is analogy. Coaches speak of "a Vis person" versus "a Jessup person" the way conservatories speak of a chamber musician versus a soloist.

The national ecosystems

Below the international marquee lies something larger in sheer volume: national moot cultures. India's is among the most intense in the world. Its national law universities and law faculties treat mooting as a central extracurricular discipline; a dense domestic calendar of competitions runs year-round, hosted by law schools and sponsored by firms and chambers; and success on that circuit is a visible line on a graduate's résumé, read by recruiters as a proxy for exactly the skills the profession is hungry for. A strong national circuit is a feeder to the international one — the pipeline that carries a first-year who argued a college moot to, three years later, a podium at an international final.

Coaches speak of "a Vis person" versus "a Jessup person" the way musicians distinguish a chamber player from a soloist.

What every rung of the circuit shares is the closed-world problem, the both-sides memorial, and the bench. A student in Bengaluru and a student in Vienna are doing recognisably the same thing, judged on recognisably the same virtues. That universality is why a change to the method — a new tool that alters how the work is done — is not a local event. It reaches every room on the circuit at once.

Chapter IV

What mooting builds

Ask a hiring partner why they light up at a Jessup finalist on a CV, and the answer is never "they know international space law." It is everything the arguing of it forced the student to become.

The subject matter of a moot is, in a hard sense, disposable. The finalist will likely never litigate the law of atmospheric re-entry or the CISG's rules on conforming goods. What survives the competition is not the doctrine but the disciplines the doctrine was a pretext to build — and those transfer to any area of law, and to a great deal beyond it.

Four disciplines

Research under a closed universe. A moot teaches a student to take a bounded set of materials — a problem, a record, a finite body of authority — and know it completely: not to skim it, but to hold the whole of it and see how each piece bears on each issue. This is the researcher's discipline, and it is rarer and more valuable than it sounds.

Written advocacy. The memorial teaches structure under constraint: a fixed page limit, a required skeleton, a citation standard enforced to the mark. A student learns that clarity is a form of respect for the reader, and that the argument which is easiest to follow usually wins.

Oral advocacy and thinking on your feet. The bench teaches the thing no written exercise can: to hold a line of argument in your head while a hostile, intelligent interrupter tries to dismantle it, and to answer the question actually asked rather than the one you prepared. This composure is the moot's signature gift.

Arguing what you do not believe. Writing both sides, and being assigned your side by lot, teaches intellectual honesty of an unusual kind — the ability to see a dispute from inside each party's strongest case. Lawyers call it issue-spotting; philosophers call it the principle of charity; it is the foundation of both good advocacy and fair judgment.

Why the profession cares

The moot as an on-ramp

A moot record is read by recruiters as a bundle of proven abilities — research, drafting, composure, teamwork under deadline — demonstrated competitively and under scrutiny, which a transcript cannot show. For many students, especially those without family connections in the law, the moot is the most legible signal of ability available to them: a meritocratic stage where a first-generation lawyer from a lesser-known college can out-argue a favoured team and be seen doing it.

Hold that last point. It is the moot's quiet social function — a ladder that rewards preparation and reasoning over pedigree — and it is precisely the thing that the arrival of a powerful, unevenly distributed tool puts under strain. To that arrival we now turn. But keep the four disciplines in mind, because the sharpest question about AI in mooting is not "can it help?" It plainly can. The question is which of these four it strengthens, which it hollows out, and which it cannot touch.

Part Two

The Craft

How it is actually done — from the competitor's desk and from the organiser's chair.

§ · §

Chapter V

To participate: the competitor's art

A moot is won in the weeks before anyone speaks. Here is the sequence a strong team moves through — and the places where competitions are decided.

The work begins with the problem, read not once but until it is memorised. Experienced mooters read the compromis with two coloured pens: one for facts the parties have admitted — which are now unassailable and can be built upon as bedrock — and one for facts left contested, which are the terrain of the fight. Getting this division right is the whole foundation. A team that argues a contested fact as though it were conceded, or wastes pages proving something already admitted, has lost marks before it has found a single case.

Admitted — build on it

  • The parties' identities and treaty memberships
  • The sequence of events both sides accept
  • Anything the special agreement jointly states

Contested — argue it

  • Causation, and the weight of the evidence
  • Attribution, knowledge, and intent
  • The characterisation the tribunal must choose

The authority hunt

With the issues fixed, the team builds its ratio — the chain of authority for each proposition it must establish. In an international moot this means the treaties, the resolutions and soft-law standards that carry weight in a thin field, the national practice that evidences custom, and the decisions of international courts imported by analogy. Each authority is pinned to a pinpoint — the exact article or paragraph — because a citation without a pinpoint is a bluff a good bench will call. The strongest teams build the opposing chain with equal care, because the argument you have already defeated in preparation is the one that cannot surprise you at the podium.

Drafting the memorial

The memorial is then written to the skeleton and the citation standard, both sides, under the page limit. The discipline here is subtractive: the second draft is shorter than the first. The index of authorities is assembled as the team goes, formatted to rule, because a scramble to build it the night before the deadline is how avoidable marks are lost. And the whole is kept scrupulously anonymous — no team name, no institution, no tell.

The argument you have already defeated in preparation is the one that cannot surprise you at the podium.

The oral rounds

Preparation for the bench is rehearsal against interruption. Teams run practice rounds — "moots" in the older sense — before coaches and alumni who play the hostile judge, and they build, for each issue, a mental map: the one-sentence proposition, the two or three cases that carry it, the record cite that grounds it, and the answer to the question they most dread. Speakers prepare a roadmap to open with, learn to signpost ("I turn to my second submission"), and rehearse the graceful concession — conceding the small point to hold the large one. Rebuttal, the brief reply after the opponent has spoken, is prepared in outline but written in the room, because it must answer what was actually said.

A working example

Arkania v. Novaris: reading a real problem

Consider a recent space-law problem: a fleet of satellites is de-orbited over thirteen days, burns up in the atmosphere, and acidic rainfall follows over a neighbouring state — against a background of that state's own decades of industrial pollution. The proposition asks, in effect, who carries the risk when the science is inconclusive?

Read closely, it is barely about space law at all. It is a problem about causation under scientific uncertainty and the burden of proof, dressed in a space-law fact pattern. The winning insight is not to know more space treaties; it is to see what the problem is really testing, and to argue the burden of proof — because whoever the court says must prove the uncertain link is the party that loses. That is the reading skill a moot is built to teach, and no amount of retrieved authority substitutes for it.

Chapter VI

To run: the organiser's burden

Behind every moot is a committee that wrote the problem, recruited the bench, and now must defend the fairness of the whole thing. Running a moot well is a distinct craft, and mostly invisible when it is done right.

The organiser's first and hardest task is the problem itself. A good moot problem is a small work of art: balanced so that neither side is obviously stronger, rich enough to sustain several distinct issues, precise enough that its agreed facts are genuinely agreed, and open enough that the law it turns on is genuinely contestable. It must reward the team that reasons best, not the team that guessed the drafter's intent. Competitions often publish clarifications in the weeks after release, answering teams' questions about ambiguities — a sign both of care and of how hard perfect drafting is.

The rulebook

Around the problem the committee builds a rulebook: team composition, the memorial's required sections and format, the citation standard, page and word limits with their penalties, the schedule of rounds, the time allocation for each speaker, and the scoring rubrics for memorial and oral performance. Much of this reads as bureaucracy and all of it is load-bearing. The page limit forces concision; the anonymity rule protects against favouritism; the citation standard makes the written contest comparable across teams. Remove any of them and the competition stops measuring what it claims to measure.

The scoring

What the marks actually reward

A memorial is typically marked out of a hundred, weighted across knowledge of the facts and law, analysis and organisation, the extent and use of authority, correct format and citation, and grammar and style — with deductions for exceeding limits or missing required sections. Oral rounds score knowledge, responsiveness to the bench, structure, and courtroom manner. Two features are worth noting: citation and format alone can be a sixth of a memorial's marks, and the researcher may be tested and prized separately. Both facts matter enormously once a machine enters the room.

The bench and the integrity of the result

A moot is only as good as its judges, and recruiting them — practitioners, academics, sitting judges, alumni — is a perennial labour. Benches must be briefed, calibrated so that a lenient panel in one room does not disadvantage a team against a harsh panel in another, and rotated to avoid conflicts with particular schools. And the committee is the guardian of integrity: it enforces anonymity, runs plagiarism checks against the similarity threshold, polices the ban on outside assistance during rounds, and adjudicates disputes. When a moot's result is trusted, it is because an unseen committee spent months making it trustworthy.

Remove any one of the rules and the competition stops measuring what it claims to measure.

Hold that sentence too. Every rule in the book encodes an assumption about how the work will be done — how long research takes, what a student can plausibly produce alone, what counts as their own work. Generative AI breaks several of those assumptions at once, and the organiser is the person who must decide, rule by rule, what to do about it. That is the subject of the rest of this book.

Part Three

The Machine in the Courtroom

A tool arrives that can read the whole problem in a second. What collapses, what breaks, and what it cannot touch.

— ⟡ —

Chapter VII

What collapses

Start with the honest part: much of what a moot used to demand in weeks, a well-built AI system now does in minutes. Pretending otherwise helps no one. The question is what that speed is worth, and what it costs.

The closed universe of a moot — one problem, a finite record, a bounded body of authority — is, it turns out, the ideal shape for a modern retrieval system. A general search tool over the whole of the law can only ever be approximately helpful; a system pointed at a single problem can be complete. It can hold the entire world of the dispute and work it the way a diligent junior would. That is not a marginal convenience. It changes the arithmetic of preparation.

Retrieval by the fact pattern

The first thing to collapse is the authority hunt. Where a researcher once spent days learning which treaty article and which imported case governed each issue, a semantic search over the instruments can return the decisive provision from a plain description of the problem — "who is liable for a burnt-up satellite" — without the student first knowing the case's name. The provision arrives pinpointed, with its binding status, and the full instrument open beside it. Days become minutes.

The record, read for you

The second is knowledge of the record. A system that has indexed the problem and its annexes answers "where did the telemetry go?" or "what did the board know, and when?" with the exact paragraph, tagged by issue and marked admitted or contested. The researcher's discipline — holding the whole record in your head — is precisely the discipline a machine most easily imitates.

Scaffolding, counters, and a practice bench

The third is the drafting scaffold. A grounded system can produce a research memo for an issue and a side: the chain of authority with pinpoints, the strongest counter-arguments, and an honest note of where the argument is weakest — not a memorial to file, but the research a student writes the memorial from. It can generate the opposing side's memo at a click, so a team hears the worst version of its case before the bench does. And it can play a tireless, record-literate practice judge, asking the hardest questions on an issue at two in the morning when no coach is awake.

A general tool over all of law can only be approximately helpful. A system pointed at one problem can be complete.

The equity cut

The ladder, and who now stands on it

Recall the moot's quiet social function — a ladder that rewarded preparation over pedigree. AI cuts both ways here, and honestly. On one hand, a good tool levels: the first-generation student at a lesser-known college, with no alumni network of senior advocates to run practice rounds, can now rehearse against a machine bench and research like a well-resourced team. On the other, if the best tools sit behind a price, they become a new form of the old advantage — the pedigree of the paid subscription. Which of these two futures arrives depends less on the technology than on how it is distributed and how the rules respond.

Chapter VIII

What breaks

Now the harder part. The same capabilities that compress weeks into minutes strain the rules that make a moot a moot — and one failure mode is not an inconvenience but a professional catastrophe in miniature.

Begin with the failure that has already escaped the classroom and reached real courtrooms: the invented citation. A general-purpose language model, asked for authority, will sometimes produce a case that does not exist — correct in form, plausible in name, and entirely fabricated. Real lawyers have been sanctioned for filing briefs built on such phantoms. In a moot, a hallucinated citation is worse than a weak argument: it is a bluff that a knowledgeable bench will call, and calling it — "counsel, I cannot find this case" — ends the argument then and there. The tool that was meant to save time has manufactured a trap.

The integrity rules under strain

The moot's guarantees of fairness were written for a world in which a student's work was, unmistakably, their own. Generative AI unsettles three at once:

The AI-content cap. Many competitions now limit the proportion of a memorial that may be AI-generated, and some ban it outright, with disqualification in play. This is enforceable only imperfectly — detection is unreliable — so it functions partly as an honour rule. It also draws a bright and useful line: the tool that helps you research is on one side of it; the tool that writes your memorial is on the other.

Plagiarism and authorship. The similarity threshold assumed that copying meant copying from another human text. A model that generates fresh prose defeats a plagiarism checker while raising the deeper question the checker was a proxy for: whose analysis is this? A memorial the student did not think through is not plagiarised in the old sense, and is still not the student's work in the sense the moot cares about.

Assistance during rounds. Most rulebooks forbid outside help once a round begins, and confiscate devices at the courtroom door. This rule, at least, the tradition defends easily — the oral round remains a sealed human contest — which is exactly why, as we will see, the oral round is where the moot's future is safest.

A memorial the student did not think through is not plagiarised in the old sense — and is still not the student's work.

The deeper break: hollowing the pedagogy

The gravest risk is not cheating. It is that a student who lets the machine do the struggling never builds the four disciplines the struggle was for. If the authority hunt is outsourced, the research muscle never forms. If the memorial is generated, the drafting muscle never forms. The competition can still be won, and the education can still be lost — and because the two used to be the same thing, no one may notice until the student reaches practice and the tool is taken away, or the stakes are real. A moot that measures who has the best AI has stopped measuring what it was built to measure.

Chapter IX

What endures

For every capability the machine collapses, there is one it cannot touch — and they turn out to be the ones that mattered most. The moot's core is more defensible than the panic suggests.

Return to the four disciplines. AI presses hardest on the first two — research and drafting scaffolds — and is largely helpless before the second two. That asymmetry is the whole of the good news, and it points to where the tradition should plant its flag.

The sealed room

The oral round is the moot's fortress. When the devices go to the clerk and the speaker rises before a live bench, no model can help. What is tested there is not retrievable: the composure to be interrupted and stay on the line; the judgment to know which of the bench's questions is the dangerous one; the instinct to concede the small point and hold the large; the human read of a room. A student who has genuinely done the preparation carries it into that room in their own head, and a student who outsourced it arrives hollow and is found out in the first exchange. The oral round does not merely survive AI. It becomes the part of the moot that proves the rest was real.

The oral round does not merely survive the machine. It becomes the part of the moot that proves the rest was real.

Judgment, voice, and the ethical sense

Beyond the sealed room, three things endure everywhere. Judgment — the reading of what a problem is really about, the space-law problem that is secretly a burden-of-proof problem — is a synthetic act the machine can assist but not perform, because it requires knowing which of a hundred true things is the one that decides the case. Voice — the particular, persuasive shape of a human argument — is the thing a generated memorial most conspicuously lacks, and benches learn to hear its absence. And the ethical sense — the advocate's duty of candour, the line between zealous argument and misleading the court — is not a capability to be automated but a character to be formed, and forming it is part of what the moot is for.

The reframing

Research is not the artefact; thinking is

The deepest thing AI clarifies is what the valuable artefact of a moot was all along. It was never the pile of paragraphs. It was the thinking — the chain of authority understood, the counter anticipated, the weak link found and honestly faced. A tool that hands a student research to think about — rather than prose to hand in — is aligned with the pedagogy, not against it. The machine, used this way, does not replace the struggle. It moves the struggle up a level, from finding the authorities to reasoning with them.

Chapter X

Rewriting the rules

The organiser from Chapter VI now has a decision to make, rule by rule. Prohibition is tempting and mostly unenforceable. Here is a more durable settlement — one that bends the tool toward the pedagogy instead of pretending it away.

The first instinct of many committees was to ban generative AI outright. It is the wrong instinct, for the plain reason that it cannot be enforced: detection is unreliable, and a rule that can be broken invisibly punishes only the honest. Worse, a blanket ban forfeits the tool's genuine good — the leveling of the research field for students without well-resourced coaching. The better path treats AI the way the profession itself is learning to: not banned, not ignored, but governed.

Five principles for a durable rulebook

1. Draw the line at authorship, not assistance. Permit AI for research — finding authority, testing arguments, rehearsing against a practice bench — and forbid it for the memorial's prose. The line the AI-content cap already gropes toward is the right one: the student may be helped to think; the student must do the writing.

2. Require verification, and punish the phantom. Since the catastrophic failure is the invented citation, make the competitor responsible for every authority they cite, and let a fabricated citation carry a real penalty. This is not new severity; it is the professional standard, taught early.

3. Lean on the sealed room. The oral round already defeats AI, so weight it. A memorial that a student cannot defend on their feet fails the only test that cannot be gamed. The more the result turns on the bench, the less any tool can distort it.

4. Disclose, don't hide. Require teams to state what tools they used, as journals now require of authors. Disclosure normalises legitimate use, isolates the illegitimate, and teaches the habit of candour the profession will soon demand.

5. Close the equity gap deliberately. If a tool levels the field only for those who can pay for it, the committee should treat access as part of fairness — providing or endorsing tools available to every team, so the moot's ladder is not quietly re-tilted toward the well-resourced.

A tool built to the line

What "governed, not banned" looks like in practice

The principle has a concrete shape. A research system built for a moot can be designed to sit on the right side of the authorship line by construction: it searches and quotes only instruments it hosts, so every citation is real and one click from its source — no phantoms; it produces research — a chain of authority, counters, weak links — and never memorial prose, so nothing it emits belongs in the filed document; and its most valuable feature, a practice bench for the oral round, strengthens exactly the part of the moot no tool can corrupt. A tool shaped like this does not fight the rules. It embodies them.

A rule that can be broken invisibly punishes only the honest.

Epilogue

The associate, not the oracle

There are two ways to put a powerful machine into a discipline built on human struggle. One hollows it out. The other deepens it. The difference is a choice, and it is still open.

Picture the young advocate from the prologue, on her feet in the borrowed courtroom, a year from now. Behind her preparation stands a machine — but of a particular kind. It did not write the words she is speaking; it could not, because they are hers, shaped by her voice and defended in her own head. What it did was earlier and humbler: it found the authorities so she could spend her nights reasoning with them rather than hunting them; it played the hostile judge so the bench's hardest question was not the first time she had heard it; it showed her, honestly, where her argument was weakest, so she could meet the weakness instead of being ambushed by it. It behaved, in other words, like a good associate — the diligent junior who does the legwork so the advocate can think — and not like an oracle that hands down the answer and leaves nothing to learn.

That distinction is the whole of it. An oracle replaces the struggle and, with it, the education. An associate relocates the struggle upward — from the mechanical to the intellectual, from finding to reasoning, from the paragraph to the argument. The tradition we traced from the Inns of Court was never really about the paragraphs. It was about making a person struggle productively with a hard question in front of other people, and become, in the struggling, someone who can be trusted to argue. Nothing about a machine forecloses that. A great deal about a machine could serve it.

An oracle replaces the struggle. An associate relocates it upward — from finding to reasoning.

Six centuries ago, in a hall at the Inns, a junior rose to argue a case that did not exist before benchers who would tell him, without mercy, where he was wrong. He was building something the profession has needed in every generation since, and needs no less now that a machine can read the brief. The moot endures because the thing it makes is not a document. It is a mind that can stand up, under a hard question, and answer for itself.

Keep the machine in the room. Keep it an associate. And keep the student on her feet.

↑ Return to the beginning

A note on this book

Written as a field guide to a living practice at a moment of change. The history — the Inns of Court, the founding of the Jessup in 1960, the specialised international competitions and the national circuits — is drawn from the general record of legal education; the account of the craft follows the standard shape of international moot competitions (the compromis, the both-sides memorial, the researcher's role, the scoring rubrics, the integrity rules). The working example is a space-law fact pattern of the kind these competitions set. Names of competitions are used descriptively; specifics such as exact rules, prizes, and dates vary year to year and by organiser, and should be confirmed against a competition's own documents before they are relied on.

The argument of Part Three is exactly that — an argument, offered in good faith about a technology whose effects are still unfolding. Its central claim is falsifiable and worth testing: that a moot which weights the sealed oral room, punishes the invented citation, and treats AI as an associate rather than an oracle will keep measuring what it was built to measure. Time, and the committees now rewriting their rulebooks, will tell.

Arguendo · for the sake of argument