Model Foundations

Why This Is Not a Stochastic Model:
The Judicial Architecture Behind the Probability Framework

The AIOOJ model is built on the same logical structure that courts are required to apply. This page explains that correspondence — and why it matters for the credibility of every probability assessment in the model. The near-certain overall win probability is not optimism. It is the correct output of applying to this case the same closed, rule-governed decision method that the NSW Court of Appeal has confirmed courts use.


Part I — The Foundational Distinction

Stochastic vs Juridical Probability

Most probability models assess uncontrolled variability — the future is genuinely uncertain, and the model estimates the distribution of possible outcomes. A 70% probability in a stochastic model means there is a genuine 30% chance of the opposite result, because external, uncontrollable variables could produce it.

The AIOOJ model is not that kind of model. It is a juridical probability model. The difference is not a matter of degree — it is a difference in kind.

Stochastic — Sport / Markets / Insurance

Reflects genuine future uncertainty

A 70% favourite retains a genuine 30% chance of loss. External, uncontrollable variables — weather, form, sentiment, market movement — can produce any outcome. The future is genuinely open and the model distributes probability across that openness.

Juridical — Courts

Reflects pathway availability

A court determines whether a legally cognisable pathway to a different outcome exists. Once all pathways are mapped and the viable ones identified, what remains is not “likely” — it is legally determined. The system is closed and rule-governed, not open and variable.

The AIOOJ model’s near-certain overall win probability (~99.997%) is not optimism — it reflects the absence of any identified, legally cognisable pathway by which all three causes of action simultaneously fail. The near-certainty is structural: three independent OR-gated streams, each with individually high base probabilities. These are different statements, and only one correctly characterises this proceeding.


Part II — The Judicial Warrant

The Court’s Own Method — Confirmed in Its Own Words

The methodological foundation of this model was independently confirmed by the NSW Court of Appeal panel at the NSW Bar Association Advanced Legal Writing Seminar on 20 November 2020 — in their own words, describing how courts actually decide cases.

“After you’ve done all the preparation of the case, have you actually sat down and said: what are the ultimate propositions of fact or law that I am contending for, in summary form, in propositional form, which if the court accepts I will win the case? A huge amount of work is required in producing what in the end will be only half a page — but for judges of any court to get that sort of material would be very helpful in preparing for the hearing and then ultimately deciding the case.”
Justin Gleeson SC — Former Solicitor-General of Australia  ·  NSW Bar Seminar, 20 November 2020

This is precisely what the AIOOJ model does. Each of the nine probability inputs (P1–P9) corresponds to a discrete ultimate proposition — a legal issue that the court must decide, which if resolved in Harrison’s favour produces recovery. The inputs were not chosen to optimise the output. They were chosen because they are the issues the court must decide.

The Key Principle
The model is structured by the court’s own decision path — not by the plaintiff’s preferred framing. A model that formalises the requirement of repeatable, rule-governed decision-making is not imposing a foreign framework on the law. It is expressing what the law already requires, in terms that make the underlying logical structure visible.

Part III — The Closed System

Five Mechanisms That Constrain Legal Outcomes

The legal system is a closed, rule-governed system. This is a structural observation about how variability in legal outcomes is constrained — not a claim about judicial perfection. The NSW Court of Appeal panel at the 20 November 2020 seminar confirmed five interlocking mechanisms that define the decision space, each operating as a hard constraint on the outcomes a court can lawfully reach.

01
Binding Precedent

The hierarchy of courts requires lower courts to follow higher court decisions. A High Court authority is not a preference to be balanced against novel arguments — it is binding. The court cannot invent a new pathway around Wardley Australia or any other binding authority. Gleeson SC: “Do you have a duty to follow another court unless it’s plainly wrong?”

02
Statutory Frameworks

Parliament fixes governing rules. A statutory provision does not bend because its application is commercially inconvenient. Section 16(1)(b) Limitation Act 1969 (NSW) — 12-year specialty limitation for the Deed claim — does not vary with the defendant’s preferences. Bell (President): applicable rules are not procedural options for advocates.

03
Pleading Discipline

Issues for determination are defined by the pleadings. The court adjudicates the issues raised — not issues that might theoretically exist. Failure to identify ultimate propositions is appellable error. Bell (President): identification of challenged facts is “critical in delineating the Court’s task.”

04
Rules of Evidence — Including Mandatory Inferences

Factual inputs are controlled. The rules governing admissibility are fixed. Critically, adverse inference rules (Jones v Dunkel) are mandatory obligations on the court, not discretionary tools. Gleeson SC: “There are some principles of law which the court must comply with, such as where appropriate Jones v Dunkel.”

05
Appellate Oversight

Error-correction mechanisms ensure that deviations from required reasoning are identified and corrected. Unarticulated pathways — arguments the defendant did not advance — do not enter the decision space. Gleeson SC: the court must identify issues “because that will be an appellable error.”

Together, these five mechanisms ensure that outcomes are generated by defined rules applied to defined facts. External, uncontrolled variables — the kind that dominate stochastic systems — do not enter the decision space. The AIOOJ model encodes each of these constraints as the structure within which the nine inputs operate.


Part IV — Mandatory Court Rules

Two Rules the Court Must Apply When Data Is Withheld

When a defendant withholds data that determines the answer, two court rules operate automatically. These are not analytical preferences built into the model. They are rules the court is required to apply — they are incorporated in the probability assessments because the court will apply them, not because the model designer sought a higher number.

Mandatory Adverse Inference

Jones v Dunkel (1959) 101 CLR 298

Where a party fails to call evidence or produce documents within their control that would resolve a disputed issue, the court must draw the inference that the withheld evidence would not have assisted that party. This is not a discretion. It is a mandatory obligation triggered by established fact.

Confirmed by Gleeson SC (20 November 2020): “There are some principles of law which the court must comply with, such as where appropriate Jones v Dunkel.”

Applied to P3, P4, P7 — data refusal components

Defendant-Created Uncertainty

Chaplin v Hicks [1911] 2 KB 786

Uncertainty created by a defendant’s own conduct does not defeat a damages claim. The court applies best available evidence and makes a reasonable assessment. A defendant cannot create the evidentiary gap that prevents precise proof and then rely on that gap to resist liability.

Applied consistently in NSW Commercial List practice for defendant-controlled data withheld in breach of contractual verification obligations.

Applied to P3, P6, P7 — quantum determination

Why Both Rules Apply Simultaneously Here

Aegon initially denied that the premium data existed. Aegon subsequently admitted it does exist (D3.70, 2 August 2024 — written admission by the General Manager). Aegon has still refused to produce or verify it notwithstanding that admission. That sequence — denial, admission, refusal — activates Jones v Dunkel (withheld evidence would not assist) and Chaplin v Hicks (defendant-created uncertainty) simultaneously. The court does not have a discretion about whether to draw the inference. It has a duty.


Part V — The Legitimacy Conclusion

What the Probability Assessments Actually Mean

The five mechanisms in Part III, the judicial warrant in Part II, and the mandatory rules in Part IV together produce a single conclusion about the nature of the AIOOJ model’s probability assessments. They are not predictions about what a court might do in an uncertain world. They are statements about what a court applying the law correctly is required to do.

A legal system that produced random outcomes would be indistinguishable from arbitrary power. Its decisions could not be explained, challenged on appeal, or trusted by those subject to them. The justice system is constructed — through the five mechanisms above — to ensure that like cases are decided alike, that outcomes follow reasoned application of law to fact, and that decisions are capable of scrutiny, explanation, and replication. These are not aspirational statements. They are the core operating requirements of judicial legitimacy.

The Legitimacy Conclusion

The AIOOJ model is not advocacy dressed as mathematics. It is an honest expression of what the legal system requires. A court that properly applies Wardley, s.16(1)(b), Jones v Dunkel, Chaplin v Hicks, and the documentary record in this case will reach the same conclusions that this model reaches. The probability assessments are not predictions about what a court might do. They are statements about what a court applying the law correctly is required to do.


Source Documents

Where to Find the Full Analysis

The judicial architecture described on this page is fully documented in the Series 11 analytical instruments. This page is the accessible summary. The documents below are the auditable foundation.

S0-07 v6 — Juridical Probability Framework (internal, not for filing) S11-02 — Probability Framework Justification v2 S10-MASTER — Part A: Issues & Ultimate Propositions S11-01 — Litigation Outcome Model Narrative v2 S11-03 — AIOOJ Model v9.1.4 (Excel)

For the detailed rationale behind each P input — including how each maps to a specific court issue — see P1–9 Rationale. The Judicial Architecture accordion on that page contains the issues-to-inputs mapping table and OR-gate explanation.