A Claimed Title Beats a Verified Badge. Every Single Time.

Based on the research ofPuthineedi and Jha, "Where the Ball Starts Rolling? An Empirical Investigation into Initial Opinion Formation on Social Media Platforms," Information Systems Research, 2026

A stranger who types "Dr." in front of their name beats a verified account with a hundred thousand followers, and the decision gets made before anyone checks a single fact.

That is the headline result from a new paper by Venu Puthineedi of NEOMA Business School and Ashish Kumar Jha of Trinity College Dublin, published in Information Systems Research. Across three controlled experiments using Instagram-style feeds, 185, 246, and 179 participants respectively, the authors find that people do not evaluate accuracy first and form an opinion second. They form the opinion first, often within a handful of exposures, and evaluate accuracy, if at all, later and half-heartedly. The paper calls this early plateau the "Point of Critical Information": the moment a bounded run of consistent posts, roughly five in the reported experiments, is enough to stabilize what a user believes. After that point, posts that agree get absorbed easily and posts that disagree get waved off, regardless of which set is actually true.

The detail that should worry every platform trust and safety team is the third experiment. Profiles that displayed an unverified professional title, the study uses "Dr." as its example, generated more trust and stronger engagement than verified influencer accounts posting the identical content. Users leaned on a claimed credential harder than they leaned on the very features platforms built specifically to signal legitimacy. Badges, checkmarks, and follower counts are supposed to be the trust layer. This paper finds a stranger's unverified job title can out-rank all of them.

The Five Posts That Decide It

Puthineedi and Jha are not describing slow-building conviction. They are describing a fast, largely automatic process that resembles scrolling behavior far more than deliberation. Study 1 in the paper shows opinions stabilizing after a bounded volume of consistent exposure, a pattern the authors call a "sufficiency" effect. Study 2 shows that engagement with a piece of information barely tracks whether it is actually true during that early window, because users are relying on familiarity and narrative fit rather than fact-checking. Study 3 adds the identity layer: once a user has decided a source feels credible, usually from a title or framing rather than a badge, later posts that reinforce that framing get an easier ride and later posts that contradict it get discounted.

This lines up with a broader pattern in persuasion research going back decades, that people process most everyday information heuristically, on cues like source authority and repetition, rather than by weighing evidence line by line. What is new here is the timing. Puthineedi and Jha put a number on how little repetition it takes on a real feed format, and they show the identity cue that does the heaviest lifting is not the one platforms spent a decade engineering.

When "Duke PA Student" Was Enough

The mechanism is not hypothetical. In August 2026, a Netflix reality-show alum named Angel Victoria Murdock was accused of posting nearly two years of content presenting herself as a physician assistant student in Duke University's PA program, complete with scrubs, staged clinical videos, and homemade Duke Hospital badges, according to reporting by Inc. Duke told Inc. it had no record of her ever applying to or attending the program. She had built an audience of hundreds of thousands of followers before classmates in the real 2026 graduating class noticed her cap and gown were the wrong color and started asking questions.

Nothing about that account was platform-verified. No blue checkmark, no institutional confirmation, no credentialing check of any kind stood between Murdock and a large, trusting audience for two years. What did the work was the costume: scrubs, a hospital-adjacent setting, and a repeated, consistent claim to a professional identity. That is close to a field demonstration of Puthineedi and Jha's Study 3 finding, that a claimed professional identity can out-trust actual verification apparatus, because most viewers never got past the early, low-effort impression the persona was built to produce.

The identity cue that does the heaviest lifting on a feed is not the one platforms spent a decade engineering into badges and follower counts.

Platforms Bet Heavily on the Wrong Signal

This finding lands at an awkward moment for the badge itself. Twitter, now X, spent years building the blue checkmark into a scarce, editorially-vetted signal of "active, notable, and authentic" identity. Then, starting April 1, 2023, the company wound that legacy program down and turned the checkmark into a benefit of the paid X Premium subscription, according to X's own help center. Anyone with a phone number and a subscription can now carry the mark that used to require an internal review. In late 2025, the European Commission went further, ruling that X's checkmark misleads users under the EU's Digital Services Act, and X has stopped even calling Premium subscribers "verified" in the European Economic Area, per the same X help page. The signal platforms spent years training people to trust has been visibly devalued from the inside, twice, in the space of three years.

LinkedIn took the opposite design path, and it still is not enough. In April 2023, the company rolled out free identity verification through a partnership with the ID service CLEAR, plus workplace verification tied to a user's actual company email, according to TechCrunch's reporting at the time, explicitly without a paid subscription or checkmark model. LinkedIn's bet is that verifying the underlying fact, this person really works where they say they work, is worth more than a badge that anyone can buy. Puthineedi and Jha's data suggests that bet is only half the battle. Even a rigorously verified employment badge is competing against a cognitive habit that responds more to a confidently claimed title than to any icon next to a name. The platform did the harder, more honest engineering work, and the paper implies users still may not process it the way its designers hoped, because the credibility judgment often resolves before the badge is even consciously registered.

Corrections Arrive After the Decision Is Made

The paper's most consequential implication involves timing, and here it recreates a much older experimental literature rather than contradicting it. Stephan Lewandowsky, Ullrich Ecker, Colleen Seifert, Norbert Schwarz, and John Cook's widely cited review in Psychological Science in the Public Interest documents the "continued influence effect": corrections reliably reduce reliance on debunked information, but reliably fail to eliminate it, even when people say they believe the correction. The false impression keeps leaking into judgments after the fact has been formally retracted.

Puthineedi and Jha's contribution is to show why the correction was fighting an uphill battle before it even arrived. If an opinion stabilizes after a bounded run of consistent posts, often built on an unverified identity cue rather than a factual check, then a correction is not landing on a blank slate. It is landing on an evaluative framework that already treats aligned information as easy to accept and challenging information as easy to wave off, the exact asymmetry the paper documents in Study 3. A fact-check published a day, a week, or a news cycle later is arguing against an impression that was never built to update on evidence in the first place. Fact-checks almost always arrive after content has already circulated, and by then an initial evaluative framework, one that was not really assessing accuracy to begin with, is already doing the work of filtering what comes next.

The Design Rule

Put the two literatures together and the sequencing problem becomes obvious. The window in which an opinion is genuinely up for grabs is short, on the order of a handful of posts, and it typically closes before any correction, warning, or fact-check has a chance to appear. Everything platforms currently spend on post-hoc correction, labels, fact-check partnerships, community-notes-style crowd verification, is aimed at a moment that, per this paper, mostly comes too late to matter as much as designers hope. Everything platforms spend on verification badges is aimed at a signal that a claimed professional title can simply outrun.

None of this means correction infrastructure or verification badges are worthless. It means they are solving the wrong part of the timeline if the goal is preventing durable false belief, not just documenting it after the fact. The leverage point this paper identifies sits earlier: at the first few exposures to an unfamiliar, high-stakes claim, especially one carrying a confident professional-sounding identity cue. That is a genuinely uncomfortable design target, because it asks platforms to intervene before there is any way to know whether a claim is true, based only on the pattern of how it is being presented. It is also, according to this research, the only point in the sequence where intervention still has real leverage over what someone ends up believing.