The Grammar of Bad Faith

The tactic taxonomy, from person to poll

Everyone who has spent time arguing in public knows the moment. You are mid-conversation, patiently laying out a position, when you notice that your interlocutor is not actually confused. The misreadings are too convenient. The outrage arrives too precisely on cue. The standards keep moving, but always in one direction. You came to clarify; they came to obstruct. And the sinking realization is not that you are losing the argument — it is that there was never an argument to lose. You were a prop in a performance staged for someone else.

That performance has a name: bad faith. Arguing in bad faith means intentionally misleading others while presenting oneself as sincere. Intent is constitutive, but rarely observable directly. A confused person and a bad-faith actor may utter the same sentence. Repetition after clear correction, selective standards across comparable cases, private evidence, and a stable pattern of evasions can raise confidence; one disagreeable exchange cannot. The tactics below are therefore warning signs and discourse failures before they are diagnoses of a person’s inner state.

The previous chapters treated cultural patterns ecologically — how they replicate, mutate, merge with hosts, and curve the fitness landscape they compete on. This part turns to armaments. Discourse is the arena where patterns fight for hosts, and like any arena it has weapons native to it: techniques that win the exchange without winning the argument. Bad faith is the general form of those weapons, and it has a grammar — a small set of composable tactics that recur so reliably, across every topic and every faction, that once you learn to parse them you can never again mistake them for conversation. This chapter lays out that grammar, and then reads two texts written in it: one aimed at a person, one embedded in a poll.

The Seven Tactics

Bad-faith exchanges often use a recognizable repertoire. The same surface moves can also arise from confusion, emotion, or incompatible assumptions, so each must be tested in context.

Motivated misinterpretation. Deliberately distorting statements to portray an opponent negatively — reading a factual claim as an attack, a nuanced position as an extremist one. The move works by stripping context: keep the words, discard the qualifications, and the misleading paraphrase does the rest. It is the workhorse tactic, and the first case study below is built almost entirely out of it.

Double standards. Applying rules selectively. The bad-faith actor demands generosity, empathy, and nuance for his own positions while withholding all three from yours. Every minor flaw in your argument is disqualifying; every major inconsistency in his is a distraction from the real issue. The asymmetry is the tell: a sincere interlocutor applies one standard and lets the chips fall.

Strategic outrage. Deploying faux offense to derail the exchange, provoke a defensive response, and claim the moral high ground in a single move. The discussion shifts from substance to emotion, and now you are apologizing for tone instead of defending a claim. Outrage is cheap to manufacture and expensive to answer, which is exactly why it is manufactured.

Goalpost shifting. Changing the criteria for evidence or success whenever they are about to be met. Prove the point as stated and the point turns out to have been something else; meet the new standard and a newer one materializes. The function is to make defeat unreachable — not by winning, but by ensuring the game never ends.

Rhetorical bait-and-switch. Pivoting rapidly between modes of argument — facts, values, authority — to evade accountability. Challenge the factual claim and it was really a moral point; challenge the moral point and here is a study; challenge the study and who are you to question the experts. The continuous switching of grounds prevents you from ever pinning down a position long enough to expose its contradictions.

Feigned curiosity. Asking “innocent questions” that inject doubt, misinformation, or conspiracy while retaining plausible deniability. Just asking questions is the canonical form. The question format does the work of an assertion without incurring an assertion’s liability: press the questioner and he is merely curious, an open mind unfairly maligned.

Personal attacks and deflection. When all else fails, attack the arguer — character, motives, credibility — rather than the argument. Ad hominem is the oldest entry in the repertoire and still the most reliable, because it discredits and distracts simultaneously, and because there is no defense against it that does not look like protesting too much.

Notice what unites the seven. None, when sustained, engages the actual content of the opposing position. The repertoire appears across factions; what distinguishes intentional obstruction from ordinary failure is the pattern, evidence of state of mind, and response to correction.

What It Costs

Engaging with bad faith is more damaging than a waste of an afternoon. The damage is systemic, and it compounds.

The immediate casualty is trust — the currency every exchange of reasons runs on. When I argue with you sincerely, I extend a credit: I assume your words track your beliefs, your questions want answers, your concessions are real. Bad faith counterfeits that currency. And like monetary counterfeiting, the harm falls not mainly on the party defrauded in one transaction but on the currency itself. Repeated exposure teaches people to assume the worst even of sincere interlocutors; every future conversation starts from suspicion. That is cynicism, and cynicism is not a mood but a rational adaptation to a debased discourse.

The second casualty is participation. Sincere participants are the ones who bear the cost of bad-faith tactics — they are the ones who prepare arguments that will be distorted, offer nuance that will be weaponized, extend charity that will not be returned. Rational actors eventually stop paying. They exit, and the discourse they leave behind is selected for exactly the participants who thrive in it. The arena keeps its gladiators and loses its citizens.

The end state is fragmentation. Communities that can no longer presume good faith across their internal divides stop being communities that disagree and become adjacent camps that monitor each other. Cooperation contracts to the like-minded. This is how bad faith, a tactic deployed conversation by conversation, aggregates into a structural fact about a culture — and why the remaining chapters of this part treat its specialized forms as civilizational problems rather than etiquette violations.

The Counter-Protocol

The instinctive response to bad faith is to argue harder — more evidence, more patience, more precision. This is exactly wrong, because it mistakes the encounter for an argument. There is no configuration of evidence that satisfies a moving goalpost, no clarification that survives motivated misinterpretation. Effort spent persuading a bad-faith actor is not merely wasted; it validates the pretense that a genuine exchange is occurring.

The effective protocol has three moves.

First, name the tactic — for the observers. Identify what is happening, explicitly and succinctly: “You’ve now shifted the goalposts twice; the original claim was X.” The audience for this sentence is not the bad-faith actor, who knows perfectly well what he is doing. It is everyone watching. Public arguments are almost never duels; they are performances before a gallery, and the gallery is persuadable even when the performer is not. Naming the tactic converts the exchange from a debate you cannot win into a demonstration you cannot lose.

Second, disengage. Once bad faith is evident and named, prolonged interaction only feeds it. Every additional round supplies more material to distort and lends more credibility to the fiction of a real dispute. Walking away is not conceding; against an opponent whose victory condition is your continued entanglement, it is the only winning move.

Third, redirect. The energy that would have been spent on the committed obstructionist goes instead to audiences who value sincerity — educating neutral observers, strengthening engagement with people who argue to understand. And where you have standing to shape a forum, set explicit standards for discourse in advance, because a stated norm makes the violation visible and the naming credible.

The protocol’s underlying principle is this: in public discourse, the interlocutor is rarely the audience. Bad-faith actors understand this perfectly — it is why they perform. The defense is to understand it equally well and perform better: transparently, on behalf of the sincere.

Now the two case studies, at two scales. The first shows the grammar deployed against a single person; the second shows it embedded in an institution’s measuring instrument, where it is harder to see and harder to answer.

Case One: The Rowling Treatment

No recent episode illustrates motivated misinterpretation at the personal scale better than the sustained campaign against J.K. Rowling.

Start with what she has actually said, because the distortion only becomes visible against the original. Rowling explicitly affirms the rights and dignity of trans people, supports their freedom from discrimination, and advocates for their well-being. Her central arguments concern safeguarding sex-based rights for women, protecting single-sex spaces, and urging caution about medical transition for minors. One can dispute any of these positions — they involve genuine trade-offs among biological realities, safety considerations, and competing claims — and serious people do dispute them. That is what the positions are: disputable, nuanced, and grounded in stated concerns.

What her most vocal critics dispute is none of it. Instead the positions are routinely compressed into a caricature — hateful, bigoted, anti-trans — by the simple device of conflating disagreement over policy and biology with prejudice against persons. The tactic is textbook motivated misinterpretation: ignore the explicit affirmations, strip the qualifications, and read the residue as animus. It is executed through the moral binary — “ally” or “bigot,” with no admissible position between — a pair of weaponized labels doing the compression work that argument would otherwise have to do.

Here is the diagnostic question, the one that separates bad faith from mere error: what is the distortion for? Not persuasion — nobody believes Rowling will be argued out of her views by being called a bigot, and her critics do not expect her to be. The target is everyone else. Painting a figure of her prominence as morally reprehensible is discourse policing by moral intimidation: a demonstration, staged for bystanders, of what happens to those who voice similar concerns. She is, in the terms established above, the prop; the gallery is the point. That the campaign has conspicuously failed against Rowling herself — she is unusually insulated by wealth and standing — does not diminish its effectiveness against the thousands watching who are not.

The chilling consequences run in three directions.

First, nuanced discourse is silenced. Fear of reputational destruction deters individuals — especially public figures with careers to lose — from openly addressing genuine questions about women’s rights and safeguarding. The observable debate shrinks to those too insulated to punish and those too extreme to care.

Second, the debate polarizes. When every position is forcibly mapped onto the ally/bigot binary, good-faith disagreement becomes structurally impossible; the middle positions do not lose the argument, they cease to be utterable. Each camp then encounters only the other’s worst representatives, and concludes the worst.

Third — and least appreciated — genuine trans advocacy is weakened. Real advocacy, focused on dignity, safety, and well-being, depends on public trust, and exaggerated accusations spend that trust at a ruinous rate. When observers watch a plainly non-hateful position denounced as hate, they recalibrate — and the recalibration does not stop at the false alarm. The accusation loses force precisely where it is true. The strategy devours the cause it claims to serve.

This is the personal-scale weapon in full: motivated misinterpretation as deterrent, enforced through stigma, at the cost of the discourse and ultimately of the enforcers’ own stated goals. What begins as distortion of one woman’s words scales up, through the machinery this part goes on to examine, into the tiered punishment system of cancel culture and the memory hole. The grammar is the same; only the institutional muscle grows.

Case Two: The Semantics of Violence

The second case moves from person to institution, and from rhetoric to measurement. Bad faith at this scale no longer looks like an argument at all. It looks like data.

An NPR/PBS News/Marist poll asks Americans whether they agree that “Americans may have to resort to political violence to get the country back on track.” The headline results, from the March 2024 and September 2025 waves:

(Source: NPR/PBS News/Marist national polls, March 2024 and September 2025.)

The coverage wrote itself: nearly a third of Americans open to political violence, a surging radicalization, one party slightly worse than the other. But look at the question before you look at the answers. There is almost never a good reason to phrase a question this way. It is not a probe of moral conviction; it is a psychological Rorschach — an engineered ambiguity that measures framing effects, not beliefs. Three linguistic tricks do the inflating, each preserving deniability for the instrument’s designers.

Diffused agency. The subject is “Americans” — not you. The word is an absolution: someone else, like you, a member of the moral collective, might have to act. It converts personal responsibility into abstract observation, granting the respondent moral distance while still activating group identification. It is permissioned empathy for hypothetical violence.

Implied necessity. “May have to” presupposes legitimacy before the respondent weighs in. It encodes inevitability — violence as reluctant duty forced upon good citizens rather than a choice anyone makes. The respondent who agrees is not endorsing anything; he is merely acknowledging a tragic possibility. This is assent laundering.

Normative framing. “To get the country back on track” smuggles in the premise that the country is off track. The grievance is baked into the question, and anyone dissatisfied with the state of the nation — most people, most of the time — is guided toward agreement regardless of their actual view of violence.

Composed, the three tricks transform a moral question — is violence justified? — into a sentimental one — are you frustrated enough to believe someone might have to act? The difference is everything, and the data above confirm it. A sixteen-point Democratic surge in eighteen months is not a wave of radicalization; it is a change of administration. The measure tracks whose grievance is currently activated, exactly as a frustration-meter should and a violence-meter should not.

The analysis yields a falsifiable prediction. Replace “Americans” with “people like me,” or sharper still, “members of my political party,” and support collapses across the board — because the abstraction of responsibility is what keeps the numbers inflated. The poll is not measuring extremism. It is measuring distance from culpability. And a neutral formulation exists; it is not even hard to write:

“Do you personally believe that the use of physical force for political goals can ever be justified in the United States today?”

Ask that, and you find near-universal rejection across parties — a boring, stable result unfit for the outrage economy. The sensational asymmetry lives only inside the linguistic fog, which is why the fog is maintained. The ambiguity is not a flaw in the instrument. It is the product. The intent is not to reveal; it is to polarize, and the manufactured headline — one tribe likelier to “endorse political violence” — is the payload. (The same fog drifts in the opposite direction whenever speech is reclassified as violence to license its suppression; I draw that boundary precisely in where speech ends, and will not re-derive it here.)

Pollsters don’t measure beliefs. They design them. Control the phrasing of the question and you control the moral topography of the answer — which makes a poll question, in the wrong hands, the most scalable bad-faith instrument yet devised: motivated misinterpretation performed in advance, on millions of respondents at once, with the distortion built into the instrument so that every honest answer arrives pre-corrupted.

That is the grammar of bad faith at its two extremes: retail, against a named person, enforced by stigma; and wholesale, through an institution’s measuring apparatus, enforced by headline. Between those extremes lies a whole armory — labels engineered as loyalty tests, categories cut loose from their anchors, suffering minted into currency, punishment systems for the insufficiently compliant. The chapters that follow take up the weapons one at a time. The grammar, now that you can parse it, will be recognizable in every one.