The philosopher Antony Flew gave the fallacy its name with a small story. A man reads that a Scotsman committed some grisly crime and says, "No Scotsman would do such a thing." Told that one did exactly that, he replies, "No true Scotsman would do such a thing." The category got smaller the instant a counterexample threatened it. Nothing was argued. The claim was simply protected.
The move never went away. It shows up most in identity-based arguments, the kind that fill threads about politics, faith, fandom, and which jobs count as the real version. Whenever a group gets a bad data point, a defender can redefine the group to exclude it. The category survives, the claim about it survives, and the counterexample that actually mattered gets quietly waved off. This is no true Scotsman, and here is how to spot it, five modern examples held in balance, and the definitional disputes that look identical but aren't.
The fallacy isn't that definitions exist
Categories can be defined. A vegan eats no animal products. A licensed electrician has passed a specific exam. These are real criteria, and someone can fail them. Pointing that out is not a fallacy. It's just checking the definition.
The fallacy is about timing. The definition in a no-true-Scotsman move appears only after the counterexample lands, for the sole purpose of excluding it. Before you raised the inconvenient case, "Scotsman" meant anyone from Scotland. After you raised it, "true Scotsman" means anyone from Scotland who behaves the way the speaker needs them to. The category was reverse-engineered from the evidence it had to dodge.
There's a clean diagnostic. Ask: "What's your definition of an X, the one you held before I gave you this example?" If the speaker can state criteria that existed independently of your counterexample, there's no fallacy, just a disagreement about who qualifies. If the only definition they can produce is "a real X wouldn't do what you just described," the move is exposed. The definition is doing no work except shielding the claim.
Five places it shows up now
The pattern cuts across every line people draw around themselves. A spread:
Software work. "Real programmers don't lean on AI assistants." Someone points out that plenty of senior engineers use them daily. The reply: "Then they're not real programmers when they do." The word "real" did all the lifting. "Programmer" went from a description of what someone builds to a loyalty test about their tooling, the instant a counterexample arrived.
Religion. "No real member of our faith would support that." A congregation member who plainly does is named. The response: "Well, no true believer would." The faith has now narrowed to "people who already agree with the speaker." Doctrine has actual content, but this answer cited none of it. It just drew the circle tighter around the inconvenient name.
A claim from the left. "No real progressive would back that surveillance bill." Then several elected progressives are shown voting for it. The reply: "They've abandoned progressivism, so they don't count." "Progressive" stopped describing a coalition and started describing a purity score, calibrated after the votes were counted.
A claim from the right. "No real conservative would have voted for that spending package." Then several long-serving conservatives are shown voting yes. The reply: "Real conservatives don't spend like that, so they're RINOs." Same machine, mirrored. "Conservative" was redefined on the spot to exclude the ones who did the thing.
Sports, to reset the temperature. "A true fan never misses a home game." Your friend with a newborn missed three. "Then they're not a true fan." Lower stakes, identical structure. The badge of "true fan" got rewritten the second a counterexample with a good excuse walked in.
In all five, the giveaway is the same. The speaker met a real example and answered it by adjusting the category instead of the claim. The move belongs to no team. Anyone defending a generalization under pressure can reach for it, and most people have.
The disputes that look like this but aren't
This is where the move gets interesting, because not all definitional pushback is a dodge. Some of it is the opposite of a dodge. Three honest cases that get mistaken for the fallacy:
Definitions that predate the argument. "Journalism" has criteria that existed long before any specific bad actor. If someone publishes unverified rumor with no sourcing and you say "that isn't journalism," you're applying a standard the field already held. You didn't invent it to exclude one writer. The test holds: the definition came first, the counterexample second.
Self-identification that contradicts observable practice. Someone says, "I'm vegan," and you note they eat fish on weekends. Replying "then you're not actually vegan" isn't redefining vegan to win. Vegan has a fixed, public meaning, and the person's own behavior fails it. The category didn't move. The person stepped outside it.
Words whose meaning is genuinely contested. Sometimes a term really is being fought over in good faith, with both sides arguing about what it should mean. That's a definitional debate, not a fallacy, as long as nobody is quietly retrofitting the meaning to dodge a single example. The line is whether the criteria are argued openly or smuggled in to protect a claim.
The same question settles all three: was the definition articulated before the counterexample, or reverse-engineered from it? A definition that materializes the instant a counterexample appears, and exists only to exclude it, is the fallacy.
One caution for the person doing the citing. A pre-existing definition can still be lazy. "A true journalist verifies sources" is a standard. "A true journalist has integrity" is close to circular, and leans on circular reasoning the same way the fallacy does. The honest version names something testable.
If you catch yourself reaching for it
Speaker first, because that's who lands here most. If you've claimed "no real X would do this" and someone produces an X who did, you have two honest exits.
The first is to take the counterexample and revise the claim. "Fair, some members of this group do that. The version I stated was too strong." This costs less than it feels like it will, and it leaves you with a claim that can survive contact with evidence.
The second is to state a definition with criteria that existed before this argument. If those criteria are real and predate the example, the counterexample legitimately doesn't apply, and you can say so without redrawing anything. The catch is honesty about which case you're in. If the criteria only occurred to you when the inconvenient example showed up, you're not defining a category. You're protecting a claim.
The durable fix is upstream. Get comfortable with claims that have visible failure conditions. "Most members of X do Y" can take a counterexample and stay standing. "All real members of X do Y" cannot, which is why it keeps tempting people into redefining "real" mid-sentence. The weaker-sounding claim is usually the one you actually meant, and it's far harder to knock over. Reaching for "no true X" is often a sign you over-claimed at the start, the same overreach behind a hasty generalization.
When someone else is doing it
The diagnostic question, out loud and friendly: "What's the definition, written down, that applies to everyone equally?"
If the answer loops back on itself, "a true X is someone who would never do this," the move is on the table for everyone to see. The definition and the claim are the same sentence wearing two hats.
If the answer is a coherent prior definition, there's no fallacy, and the gracious move is to say so and engage with it. Then check one more thing: is that definition doing real work, or just guarding the claim? A speaker who keeps two definitions ready, a defensible narrow one and an expansive one they actually use, is running motte and bailey, retreating to the strict definition when pressed and spreading back out when the pressure lifts. The fix is the same either way: pin the definition down once, in public, and hold everyone to it. For more on flagging a pattern without turning the thread into a brawl, see using fallacy.is in a discussion.
Categories without definitions can't carry real claims. If the definition only shows up when a counterexample threatens it, the claim was never about the category. It was about the speaker's commitment to it. Spotted the move in a thread? fallacy.is/no-true-scotsman or fallacy.is/ntsm does the labeling for you.