Every resolution has one word carrying the argumentative weight. Find it, pin it, and most of the debate is already decided — because the side that gets its reading on the record has quietly set the win condition for both.
Links: Online Debates, Absolutes and Differentials — the inflation move with the motte pre-registered: a criterion defined broadly enough to be technically satisfied later while carrying the strong reading up front, The Negative’s Easy Burden (its sibling — that page is about burden, this one about terms), The Weighting Problem, Constructed ≠ Arbitrary, Word War series
Origin: not a transcript review — a method extracted from nine specimens, most of them from the Word War bracket, where the same instrument kept resolving rounds that the debaters themselves could not.
Chris: “often topics hinge on degree type words.. even in my own debate, we both crystallized that we needed to nail down what ‘sign’ meant, and when I got a concession, it made the rest of it much easier… ‘America should build more data centers’ … the point here is that ‘should’ is doing the heavy lifting, and this is typical of debate topics.”
A resolution is a sentence, but it is not evenly weighted. Almost always one term decides what would count as winning — and it is usually not the topic word. “Is masculinity in crisis?” is not about masculinity; it is about crisis. “Is the algorithm fair?” is not about algorithms; it is about fair. “Poverty in America is a sign of exploitation” is not about poverty or exploitation; it is about sign.
Two consequences:
The most common family and the most slippery, because they smuggle in a threshold nobody states. Too sensitive compared to what? Works by what measure?
Failure mode: a standard with no failure condition. Both sides bid a threshold that flatters their case and neither can be refuted, because nothing was specified that could turn out false.
The repair — operationalise against function. Name the purpose the thing exists to serve, then ask whether it still serves it. That converts an unwinnable vibes question into a checkable one, because it supplies the missing failure condition.
| Resolution | Term | Operationalisation |
|---|---|---|
| Is masculinity in crisis? | crisis | a system no longer functions for the purpose it was built |
| Are we too sensitive? | too much | past the point where it destroys what it exists to serve |
| Is the algorithm fair? | fair | the tool meets the goals of its owner |
| Does therapy work for men? | works | achieves its stated goal — not exceeds it, not produced some learning |
The therapy round shows both ways to get this wrong at once: the Aff set the bar at passive return (“money works for you” — the patient should have to do nothing), the Neg at any learning whatsoever. One standard nothing could meet; one nothing could fail.
Failure mode: the burdens are set by the word and both sides argue as if they were symmetric.
The repair — read the burden off the quantifier. A universal claim is refuted by one counterexample; the affirmative owes necessity across every case. In the wealth round the resolution was “it is impossible to get rich without luck” — so the Neg needed exactly one clean counterexample and never claimed the entitlement, while the Aff invoked a sample-of-one principle that does nothing for a universal.
Failure mode: an evidential claim gets argued as a causal one. “X is a sign of Y” is much weaker than “Y causes X” — and much easier to break, because signs have structural requirements that causes don’t.
The repair — a sign must (a) vary with what it signals and (b) mark a departure from a baseline. From Chris’s own poverty prep: the official poverty rate moved 1.5pp across four decades while consumption poverty collapsed, so the metric is insensitive by construction; and poverty is the universal human baseline, so wealth is the thing requiring explanation.
This is the page’s positive specimen. It is the one case where the pinning was done deliberately and early, and it worked exactly as advertised:
Chris: “we both crystallized that we needed to nail down what ‘sign’ meant, and when I got a concession, it made the rest of it much easier.”
Failure mode: a practical question is argued as a moral one, when the modal has no room to range.
The repair — ask whether the thing is a live choice at all, and if not, find the real decision variable. Chris’s affirmative prep on “America should build more data centers”:
Chris: “this is not a moral claim.. building data centers is a market force, they are going to get built anyway, all you can do is control where they get built.”
If the outcome is going to occur regardless of the vote, should cannot range over whether; it can only range over where, by whom, and under what constraints. The resolution names a decision variable that isn’t the one actually in play — and locating the real one is the whole affirmative case. Same instrument, entirely different family from a degree word.
Once you know which term is load-bearing, both dirty moves become legible. They are the same operation in opposite directions: adjust the term’s extension until the resolution can’t be lost.
| Move | Direction | Effect | Specimens |
|---|---|---|---|
| Inflation | Widen until no counterexample survives | The resolution becomes trivially true and stops meaning anything — you win the sentence and lose the argument | luck → “essentially anything that happens” (wealth); boomer → a disposition, “spiritually boomer” (boomers) |
| Severance | Narrow until the opponent’s evidence falls outside | You concede the substance and win on scope — every example is granted as damning and ruled off-topic | masculinity severed from biological sex (masculinity); algorithm severed from human moderation (algorithm); culture severed from circumstances (fragility); therapy culture severed from CBT (therapy culture) |
| Loading | Leave the extension alone; build the conclusion into the definition | Accepting the term concedes the resolution — the proposition becomes true by stipulation rather than by argument | therapy culture := the overuse of therapy (therapy culture) — “over” already asserts excess, which is the whole question |
The tell for inflation: substitute the offered definition back into the resolution and read it aloud. “It is impossible to get rich without anything happening” is unfalsifiable and worthless — and its rhetorical force is being borrowed from a narrower reading the speaker is no longer defending.
The tell for loading: the proposed definition contains an evaluative word the resolution was supposed to establish — overuse, excessive, proper, genuine, real. Ouedrago’s refusal is the model counter, and it takes one sentence: “He’s loading a lot of normative language there. Overuse implies that it’s over — means it’s not good… so obviously I won’t accept the definition because it by definition has a negative connotation.” Then supply a descriptive substitute rather than merely objecting; he offered “the generalisation of therapeutic ideas in society,” and it was accepted. Note this is the cheapest of the three moves to defeat and the easiest to miss, because a loaded definition sounds like a definition rather than an argument.
The tell for severance: the opponent starts agreeing with your evidence while denying its relevance. “I would agree that that’s unfair, but that’s not part of the idea of an algorithm.” The counter is never to accept the boundary and argue inside it — it is to attack the boundary directly.
In both cases the fix is the same: get the definition on the record before the evidence is presented, so the boundary can’t be redrawn once it becomes inconvenient.
The caveat that keeps this honest:
Chris: “often it is hard to know what the ‘winning argument’ really is during the debate, and the better debaters come in with a strategy based on this.”
Identifying the load-bearing word looks obvious from a transcript and is genuinely hard in the room. That is precisely why this is preparation work. The contenders who look sharp are not thinking faster live; they arrived having already decided which hill they were fighting on. Two rounds in the bracket were lost by debaters who named the winning frame in their first few minutes and then abandoned it — not from carelessness, but because nothing had told them in advance that it was the one that mattered.
| Debate | Load-bearing word | Family | What happened |
|---|---|---|---|
| Poverty is a sign of exploitation | sign | evidential | Pinned deliberately; concession won. The positive case |
| America should build more data centers | should | modal | Prep only — “they get built anyway; the variable is where” |
| Impossible to get rich without luck | luck / impossible | degree + quantifier | Inflated to vacuity; burdens never claimed |
| Are we too sensitive? | too | degree | Never pinned → a verbal dispute with no proposition in contention |
| Is modern culture making people fragile? | fragile | degree | The contrast case. Same question as the row above, pinned mid-round — and a real debate follows. Chris’s best round in the bracket |
| Does therapy work for men? | works | degree | Two defective standards, opposite directions |
| Is the algorithm fair? | fair | degree | Severance; “impartial” offered to define the contested term |
| Is masculinity in crisis? | crisis | degree | Severance; 58 minutes at layer 1 |
| Are boomers responsible? | responsible | degree (culpability vs. causality) | Moderator forced the split; Aff then inflated boomer |
| Is feminism helping relationships? | helping | degree | Defined feminism in 90 seconds, never defined helping → layer-3 death |
| How immigration replaced the woke left | “the third world” | referential (who is the actor?) | The first non-debate specimen — and the speaker pins it against himself. “They’re not homogeneous, but they have enough aligned interest to be considered one political block for the time being” — the aggregation is conceded aloud, then never defended, and it is what makes a single-actor thesis possible at all. Unpin it (Hindu vs. Muslim UK voting divergence) and the actor dissolves into groups with conflicting interests |