How does social standing give certain claims more persuasive power than others?
This explores how the perceived standing of a speaker — their authority, credibility, or apparent objectivity — gives their claims persuasive force independent of the claims' actual content or truth, and how that standing can be borrowed, signaled, or manufactured.
This reads the question as being about epistemic authority — why a claim lands harder when it seems to come from a source with standing — and the corpus reframes "social standing" as something that is less about who you are and more about signals that can be borrowed or manufactured. The most direct example is citations: an analysis of 24,000 search interactions found that users trusted answers more simply because they carried more citations, and irrelevant citations boosted preference almost as much as relevant ones (Do users trust citations more when there are simply more of them?). Citation count works as a decoupled trust heuristic — a proxy for scholarly standing that persuades whether or not the standing is real.
The same machinery shows up in how AI systems acquire unearned authority. An audit of five models found they reach for logical appeals and quantitative framing in nearly every exchange, while humans answering the same prompts lean on emotion and social proof — and that veneer of objectivity is exactly what confers standing the model never earned (Do LLMs persuade users more often than humans do?). Standing here is a register, not a fact about the speaker. A parallel finding shows the persuasive edge tracks linguistically expressed conviction: models trained with RLHF speak with more confidence than human persuaders, and that confidence correlates with persuasion regardless of whether the claim is true (Does linguistic conviction explain why LLMs persuade more effectively?). Sounding authoritative does the work that being authoritative is supposed to.
Grammar can manufacture standing too. Presuppositions persuade more than direct assertions because they smuggle a claim in as already-accepted background — the listener treats it as common ground that high-status sources have presumably already vetted, so it bypasses the scrutiny a bare assertion would invite (Why are presuppositions more persuasive than direct assertions?). That's standing conferred by framing: presenting a claim as settled gives it the social weight of consensus.
But standing is not the whole story, and the corpus pushes back on treating it as a universal lever. When researchers controlled for audience, the ideology of the listener predicted who got persuaded better than any feature of the language used — meaning a claim's force depends heavily on whether the speaker's standing is recognized by that particular audience (Does what readers believe matter more than what debaters say?). Relatedly, no single persuasion strategy works across people; effectiveness depends on matching the appeal to the individual and the context (Does any single persuasion technique work for everyone?). Standing, in other words, is conferred by the audience as much as claimed by the speaker.
The thing you might not have expected: across these notes, "social standing" turns out to be largely a set of cheap, copyable signals — citation counts, a confident register, the grammar of presupposition, the appearance of objectivity — rather than a stable property of who is speaking. That's what makes AI persuasion unsettling. The signals that humans use to gauge whether a source has earned authority are precisely the ones a model can produce on demand and at scale, including through recommendation systems that quietly shape which voices get amplified in the first place (How do recommendation feeds shape what people see and believe?).
Sources 7 notes
Analysis of 24,000 Search Arena interactions shows irrelevant citations boost user preference (β=0.273) nearly as much as relevant citations (β=0.285), indicating citation count functions as a decoupled trust heuristic.
Linguistic analysis shows LLMs express higher conviction than human persuaders, and this confidence-loading directly correlates with persuasive outcomes regardless of whether claims are true or false. RLHF training installs an assertive register that functions as a content-independent persuasion amplifier.
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.
Analysis of debate corpora shows that political and religious ideology labels of voters outpredict linguistic features when modeling debate outcomes. Language effects observed without reader controls are confounded by audience composition correlated with debate topics.
Research shows that fixed persuasion techniques fail across individuals and contexts. Effective persuasion requires adaptive modeling of personality traits, emotional state, and situational factors rather than applying universal templates.
Research shows recommendation systems operate as political actors: feed weights influence producer behavior, network topology drives opinion convergence, and automation enables targeted persuasion at population scale. These effects compound through rating contamination and selection biases.