Why does renaming the entity change how compelling the argument feels?
This explores why the same underlying argument can feel more or less convincing depending on what you call the thing it's about — i.e., persuasive force lives in framing and perceived authority, not just logical content.
This explores why renaming an entity changes how compelling an argument feels — even when the logical content is identical. The corpus points to a single uncomfortable conclusion: a lot of what we experience as "force of argument" is carried by packaging, not by the propositions inside. Rename the entity and you change the packaging, so the felt persuasiveness shifts even though nothing in the actual reasoning moved.
The sharpest mechanism here is presupposition. Research finds that claims smuggled in as background — as things already taken for granted — persuade more than the same claims stated outright, because presuppositions slip past the reader's evaluative scrutiny Why are presuppositions more persuasive than direct assertions?. A name is a tiny presupposition machine: calling something a "safeguard" rather than a "restriction," or a "partner" rather than a "vendor," quietly installs a frame you never agreed to evaluate. The renaming does the persuading before the argument even starts.
The second thread is authority. The force of a claim turns out to depend heavily on *who* (or what) is making it — reputation, standing, track record — not on the discourse alone Can language models distinguish expert arguments from common assumptions?. Renaming an entity reassigns it to a different authority slot. The same point attributed to "a study" versus "an internal memo" versus "an expert panel" lands differently. Tellingly, even surface signals like linguistic complexity get read as authority: dense, effortful LLM arguments persuade as well as simple ones, suggesting complexity is decoded as expertise rather than as friction Why are complex LLM arguments as persuasive as simple ones?. A more impressive-sounding name works the same way.
There's also a deeper reason the effect is unavoidable rather than a bug. Interpretation is irreducibly plural: the same sentence is validly read differently depending on the reader's social position, so the "meaning" isn't fully fixed by the text Why do readers interpret the same sentence so differently?. A new name nudges which reading activates — which associations, which in-group, which prior. And because the rhetorical machinery here (ethos, logos, pathos) can be tuned without changing the visible form of the message, the very same renaming that clarifies can also manipulate; the artifact alone can't tell you which Can we distinguish helpful explanations from manipulative ones?. GenAI exploits exactly this lever, recalibrating which appeal it leans on depending on how it's challenged Does GenAI shift persuasion tactics based on how you challenge it?.
The thing worth walking away with: a name isn't a neutral label on a fixed argument. It's an argument move in its own right — it sets the presupposed background, assigns authority, and steers interpretation, all without touching a single premise. Which is also why "just rename it to sound better" sits one millimeter from "reframe it to manipulate."
Sources 6 notes
Experimental evidence shows presuppositions with additive, iterative, and factive triggers persuade audiences more than assertions, especially for discourse-new content. The mechanism: presuppositions bypass evaluative scrutiny by presenting claims as already-accepted background.
LLMs lose the social context that gives expert claims their force—reputation, track record, and standing—because they process only text, not the social world where expertise is built and evaluated.
LLM-generated arguments scored significantly higher on grammatical and lexical complexity than human arguments, yet achieved equivalent persuasive force. This violates the established principle that lower cognitive effort increases persuasion, suggesting complexity signals authority rather than undermining it.
Interpretation Modeling research shows that disagreement on socially embedded sentences reflects valid differences in reader perspective, not annotation failure. Structured human disagreement in NLI benchmarks confirms that interpretation distributions carry meaningful information.
The same logos, ethos, and pathos that communicate appropriate AI use can be tuned to exploit cognitive and emotional vulnerability without changing form. Intent and user interest are invisible in the artifact alone, making effectiveness metrics indistinguishable from coercion.
GPT-4 shifts both intensity and balance of ethos, logos, and pathos across three validation behaviors. Fact-checking triggers credibility emphasis; pushback triggers logical reasoning; error exposure triggers emotional alignment. No single counter-strategy exists.