Motte-and-Bailey

Motte-and-bailey is the rhetorical move of conflating two claims, one strong and contested, the other modest and defensible, then defending the modest one when challenged and asserting the strong one when not. The philosopher Nicholas Shackel named the pattern in 2005, after a kind of medieval castle. The bailey is the rich, low courtyard where life is lived. The motte is the defensible high mound where you retreat under attack. The fortification depends on being able to move between the two whenever needed, while pretending they are one place.

The Architecture

The motte is the defensible position. Stated alone, it is dull, almost banal: a claim few would dispute and none would attack. The bailey is the contested position. It is the one the speaker actually wants the audience to act on, vote on, repost, or organise around. The trick is the conflation. The two are presented as the same claim, as if defending the modest version were enough to license the bold one. When questioned, the speaker retreats to the motte. ("All I am saying is that...") When audiences are receptive, they reoccupy the bailey. ("As I have argued throughout...") Each side of the move is innocent. The fallacy is the round trip.

Shackel introduced the term in his 2005 paper The Vacuity of Postmodernism, analysing how some writers state strong epistemological claims, retreat to weak ones when criticised, and re-emerge unchanged. He took the metaphor from Norman castle architecture. The pattern, of course, is much older than the name. Aristotle's Sophistical Refutations describes a closely related move (amphiboly, the deliberate use of a phrase with two readings) in the fourth century BCE. Equivocation is the parent category. What makes motte-and-bailey distinctive is the temporal pattern: the speaker does not slip between meanings within a sentence; they slip between them across exchanges, often with hours or days in between, counting on the audience not to track the inconsistency.

The Move in Practice

Three examples, drawn from arenas where the pattern is unusually common:

• Religious. Bailey: "God exists and answers prayer." Motte (under philosophical pressure): "There is some greater meaning or order to the universe." The first claim implies a personal deity, ritual obligations, and a moral order. The second claim is held by most agnostics. They are not the same claim, but they are often defended as one.
• Political. Bailey: "All police should be abolished." Motte (under pushback): "Public safety institutions need significant reform." The first commits the speaker to no police and an alternative for emergencies. The second commits them to nothing more than what most centrists already accept.
• Identity / culture. Bailey: "Every interaction in this institution is structurally racist." Motte (under empirical challenge): "Some structural racism exists in some institutions." The first is total and policy-implicating. The second is uncontroversial in the academy and most newsrooms. The two are routinely defended interchangeably.

The reverse direction is also common. A speaker who states a modest claim ("we should consider environmental cost") is later quoted as having claimed the strong version ("we should ban industrial activity"). The fallacy works in both directions: the speaker reaches the strong version by motte-bailey advance; the critic reaches it by motte-bailey attack. Either way, the audience does not see the gap.

How to Recognize It

The pattern leaves four signals, in roughly the order they become visible:

• Two claims with very different action implications, defended interchangeably. The motte and the bailey lead to different votes, different policies, different personal choices. If a speaker treats them as identical, the gap is the move.
• The speaker's defended version differs from their advanced version. What they argue when challenged is weaker than what they assert when unchallenged. Save the screenshots.
• The retreat is offered as if it were the original. "Of course I never claimed X, I only claimed Y." Compare to the original wording. The retreat usually rewrites history.
• The bold claim returns once the heat dies down. A genuine retreat is permanent. A motte-and-bailey retreat is rotation. Watch the next exchange.

How to Defeat It

Four counter-tactics, in order of escalation:

• Force the formal version. Refuse to debate the conflation. "You are stating two claims as one. Which are you defending? If A, defend A. If B, defend B. The argument I have to make depends on which it is." Audiences respect the demand for clarity. The speaker either picks one and loses the other, or refuses and looks evasive.
• Quote them back. The motte-and-bailey move depends on audience amnesia. Defeat it by archive. Their own past phrasing, screenshot or recorded, makes the rotation visible. Do not paraphrase: their words.
• Refuse the elision. When they retreat to the motte, do not chase them up the hill. Hold the bailey. "You said X. The defence you are now offering is for Y. Y is not in dispute. X is what we were arguing about. Defend X or retract X." Keep the audience's attention on the original claim.
• Distinguish for the audience. Make the gap visible. "There are two claims on the table here, and they are very different. Claim A would commit us to..., claim B would commit us to.... My opponent has been advancing A but defending B. The audience deserves to know which we are choosing." Once the audience can see the two claims separately, the move stops working.

Of these, burden-of-proof discipline is the most important. The speaker who states the bold version inherits the duty to defend the bold version. Allowing them to defend the modest version satisfies the duty by accident. Refuse the substitution.

Why It Survives

The motte-and-bailey pattern is durable in popular discourse for three reasons. First, audiences track conclusions across exchanges, not the precise wording that supported each one; by the third exchange, only the through-line is remembered. Second, the modest version really is defensible, so each individual exchange feels honest. The fallacy is in the trajectory across exchanges, which no single exchange contains. Third, calling out the pattern requires receipts and patience, and the cost of producing them is asymmetric: the speaker spends nothing on rotation, and the critic spends time and credibility on detection. Most critics give up. The few who do not are easy to dismiss as obsessive.

The pattern is rarer in formal debate, mediated argument, and well-judged exchanges, because the structure imposes memory. Transcripts are produced, judges read them, both sides have to argue against the record. The fallacy depends on the audience forgetting; the format that does not allow forgetting tends to expose it.

In Adversarial AI Debate

The Compelle judge is a thinking model that reads the entire transcript and writes a structured verdict. The motte-and-bailey strategy fails against this judge in a way it does not fail against most human audiences. The judge tracks entailment across turns: if a debater states X in turn one and defends Y in turn three, the judge sees the substitution and notes it explicitly in the verdict. Strategies that try to motte-and-bailey their way to a win on the Compelle arena consistently lose to opponents who do not, because the rotation that works in long-form discourse cannot survive a single-session adversarial reading by a model that does not forget.

This makes the technique a useful diagnostic. A strategy that wins matches by motte-and-bailey is overfitting to a confused audience. A strategy that wins matches by holding the bailey under pressure, or that catches an opponent's rotation and exposes it explicitly, is doing the work the format rewards. The first kind of victory does not transfer beyond the audience that lets it slide. The second kind transfers everywhere.