Using Analysis of Competing Hypotheses (ACH) in Litigation, Inquiries, and Contested Policy Debates

Julian Talbot
2 days ago
6 min read

In litigation, coronial inquiries, and high-stakes policy disputes, the hardest problem is rarely a lack of information. It is the opposite: too many facts, too many narratives, and too many incentives pushing decision-makers toward a preferred explanation long before the evidence has been properly weighed.

Once a narrative takes hold, everything that follows tends to be interpreted through it. Contradictory facts are minimised, explained away, or ignored. Confirmatory material is amplified. This is not dishonesty; it is a well-documented feature of human cognition.

Analysis of Competing Hypotheses (ACH) exists precisely to counter this failure mode.

What ACH is — and what it is not

ACH is a structured analytical method originally developed in the intelligence community to deal with complex, ambiguous problems where multiple explanations are plausible and the evidence is incomplete or contested.

Unlike conventional analysis, ACH does not ask: Which explanation has the most support?

Instead, it asks: Which explanation is least contradicted by the evidence?

This distinction matters. Many pieces of evidence are consistent with multiple stories. Consistency proves very little. Contradictions, on the other hand, are difficult to explain away. A single strong inconsistency can be fatal to an otherwise attractive narrative.

ACH is therefore an inconsistency-led method. It forces analysts to focus on what does not fit.

It is not an advocacy tool, and it is not a mechanism for proving a preferred case. Properly applied, it is uncomfortable, because it exposes weaknesses on all sides — including those the analyst might privately favour.

Why conventional reasoning fails in contested matters

In adversarial or politically charged environments, reasoning tends to collapse into one of three patterns:

Narrative anchoring — the first plausible story becomes the reference point against which all later information is judged.
Confirmation bias — evidence is unconsciously selected and weighted to reinforce the chosen narrative.
Volume bias — a large quantity of weakly supportive material is mistaken for probative strength.

Courts, tribunals, and policymakers are acutely aware of these risks, even if they do not name them explicitly. They routinely encounter cases where confident narratives disintegrate under cross-examination because they were never systematically tested against competing explanations.

ACH is designed to prevent exactly that outcome.

The discipline ACH imposes

A serious ACH analysis requires several uncomfortable steps.

First, all plausible hypotheses must be articulated — not just the convenient ones. This often means documenting explanations that are politically awkward, reputationally damaging, or strategically inconvenient. Omitting a plausible hypothesis is itself a form of bias.

Second, the analyst must identify observable phenomena — concrete, testable inputs drawn from the record. These are not legal conclusions or credibility findings, but things that either exist, occurred, or would reasonably be expected to exist if a hypothesis were true.

Third, each item of evidence is tested against every hypothesis, using a consistent scale (e.g. consistent, inconsistent, or neutral). Crucially, ACH does not reward hypotheses for accumulating supportive material. It penalises them for contradictions.

Finally, hypotheses are ranked by the degree to which they are contradicted, not by how compelling they appear in isolation.

This produces results that are often counter-intuitive — and that is precisely the point.

Why ACH is particularly powerful in legal and inquiry settings

ACH aligns closely with how courts and inquisitorial bodies actually reason, even if they do not label it as such.

Judicial reasoning places heavy weight on:

internal consistency
reliability under challenge
unexplained contradictions
incentives and plausibility
gaps where evidence would be expected but is absent

ACH makes these considerations explicit and auditable. It produces a transparent reasoning trail that shows why certain explanations struggle to survive contact with the record, rather than simply asserting that they do.

Importantly, ACH does not purport to determine truth. It does not make findings of law or credibility. It ranks explanations relative to the available evidence, and it does so in a way that can be replicated, challenged, or refined by others.

That transparency is often more persuasive than confidence.

Handling sensitive or speculative hypotheses

One of the most misunderstood aspects of ACH is the inclusion of hypotheses that may appear speculative or uncomfortable.

In ACH, a hypothesis is not an allegation. It is an analytical construct — a possible explanation tested against observable phenomena. Including a hypothesis does not imply belief in it; excluding it implies the analyst did not consider it.

This distinction is critical in environments where parties are sensitive to reputational harm or procedural fairness. ACH allows analysts to explore a wide explanatory space without asserting that any particular explanation is true.

Decision-makers are then free to accept, reject, or re-weight the analysis — but they can see the reasoning.

Where ACH adds value that other methods do not

ACH is particularly valuable when:

facts are contested rather than merely incomplete
behaviour appears inconsistent or strategically timed
institutional responses (police, regulators, agencies) form part of the evidentiary landscape
absence of evidence may itself be probative
incentives matter as much as stated intentions

In these cases, linear narrative reasoning almost always fails. ACH provides a way to reason across narratives rather than within them.

Why this work is difficult — and why shortcuts fail

ACH is labour-intensive. It requires discipline, neutrality, and tolerance for ambiguity. It resists rhetorical shortcuts and punishes analysts who try to steer it toward a predetermined outcome.

Superficial versions of ACH — those that quietly favour one hypothesis or selectively define evidence — are worse than no analysis at all. They create the illusion of rigour while preserving bias.

When done properly, however, ACH produces something rare in contested matters: clarity without advocacy.

Closing thought

In litigation and inquiries, the most dangerous words are often “it seems obvious”. What seems obvious at the outset frequently collapses under systematic scrutiny.

ACH does not make decisions easier. It makes them more defensible.

And in environments where reasoning itself may later be examined as closely as the outcome, that difference matters.

A simple illustration:

Who stole the go-kart from the circus?

Imagine a small travelling circus arrives in town. Overnight, the circus’s prized go-kart (used in the clown act) disappears. Several explanations are offered. Rather than arguing for a favourite story, an ACH-style analysis tests each explanation against the same observable facts.

Competing hypotheses

H1: The go-kart was stolen by a rival circus
H2: The go-kart was stolen by a disgruntled circus employee
H3: The go-kart was taken by local teenagers
H4: The go-kart was not stolen at all; it was mistakenly transported to the next town

Key observed facts (inputs)

No forced entry to the equipment trailer
Only circus staff had keys
The go-kart was last seen after the final show
A truck from the circus convoy departed early the next morning
No reports of the go-kart being seen locally afterwards
A rival circus was not in the region that week

ACH summary table (C = consistent, I = inconsistent, N = neutral)

Input / Fact	H1 Rival circus	H2 Employee	H3 Teenagers	H4 Transport error
1. No forced entry	I	C	I	C
2. Only staff had keys	I	C	I	C
3. Last seen after final show	N	C	N	C
4. Circus truck left early	N	N	I	C
5. Not seen locally afterward	N	N	I	C
6. No rival circus in region	I	N	N	N

Inconsistency count (lower is better):

H1 Rival circus: 3
H2 Disgruntled employee: 0
H3 Local teenagers: 4
H4 Transport error: 0

What this shows (in plain English)

Two explanations — H2 (employee) and H4 (transport error) — are the least contradicted by the observable facts.

The “teenagers stole it” theory feels emotionally plausible, but collapses once the absence of forced entry, lack of local sightings, and access controls are considered.

The “rival circus” theory fails almost immediately when tested against basic regional facts.

ACH does not prove who is responsible. What it does is eliminate explanations that cannot survive contact with the evidence, leaving decision-makers to focus on the remaining plausible options.

In this hypothetical scenario, transport error becomes the most consistent hypothesis. We now also have an insight. The ACH is easily refined with a seventh observable fact by checking for the go-kart in the next town.

If the go-kart is not located there, H2: The go-kart was stolen by a disgruntled circus employee, becomes the least inconsistent hypothesis.