Nothing New Happened. The Market Moved Anyway.

Based on the research ofYin, Jia, Xu and Zhang, "Bounded Rationality in the Digital Age: How Salient Negative Secondhand Information on Knowledge-Sharing Platforms Triggers Investor Mis-Reactions," Information Systems Research, 2026

Researchers Zhitao Yin, Ning Jia, Sean Xin Xu, and Xiaoquan Zhang ran a randomized field experiment on Wikipedia: between June 15 and July 13, 2013, they added previously disclosed, factually accurate litigation news to the Wikipedia pages of selected public firms, compared outcomes against matched control firms whose pages were left alone, and included a placebo group whose pages instead received routine operations-news edits. Nothing about the litigation was new. It had already happened, already been reported, and already, in theory, been priced into the stock. And yet, relative to controls, the treated firms saw Wikipedia pageviews rise, retail trading volume climb 15.7%, and bid-ask spreads narrow by 6.9 basis points. The placebo edits, cosmetically identical in form, produced no comparable effect. The only thing that changed was where an old fact sat on a page people already trust.

An Experiment Built to Answer One Question

The efficient market hypothesis says none of this should happen. Old information, once public, is supposed to be baked into a stock's price the moment it appears; repeating it later should be a non-event, like reading yesterday's newspaper twice. The paper's finding is not that Wikipedia spreads misinformation. Every fact the researchers added was true and already on the record. The finding is that the platform's act of resurfacing was itself a behaviorally potent event, indistinguishable, to the investors who acted on it, from real news.

The effects were not uniform, and the pattern is the interesting part. They were larger when the litigation was more severe or more recent, when the firm operated in a more visible or more positively regarded information environment, and when the underlying case signaled a governance weakness rather than an isolated dispute. That pattern lines up with representativeness-based judgment: investors were not calmly re-verifying a known fact, they were pattern-matching on a vivid, negative story and reacting as though a new one had just landed. A mediation analysis in the paper points to heightened attention as the specific channel: the Wikipedia edit did not add information, it added eyeballs, and the eyeballs did the rest. That distinction, between attention and information, is the entire mechanism, and it is what makes the result hard to wave away as market noise.

Google Already Ran a Version of This, by Accident

Wikipedia is not the only knowledge platform capable of turning an old fact into a fresh shock, and it did not take a research team to prove it. In September 2008, a staffer at a financial research firm was searching Google for recent bankruptcy news and got, as a top hit, a story about United Airlines filing for Chapter 11. The staffer posted a summary to the Bloomberg terminal network, and within minutes UAL's stock lost nearly all of its value before Nasdaq halted trading. The story was real. It was also six years old: United had filed for bankruptcy in December 2002 and emerged from it in 2006. Google's index had simply resurfaced a Chicago Tribune and South Florida Sun-Sentinel story that a search algorithm, not an editor, decided was worth returning at the top of the page. United's stock later recovered once the timeline became clear, but for a few minutes a fact everyone with access to a newspaper archive already "knew" moved billions of dollars because a search box put it back in front of the right person at the right moment.

The mechanism is identical to the Wikipedia experiment, minus the human editor. Nobody lied. Nobody disclosed anything new. A retrieval system, whether a research team's edit or a ranking algorithm, decided an old fact deserved a fresh position, and traders responded to the position, not the age of the fact underneath it.

The litigation was old, the lawsuit was old, the bankruptcy was old. The only new thing, in every one of these cases, was that somebody looked.

The Filing Was Always There. Now It's Searchable.

There is a slower, more structural version of the same dynamic sitting inside financial regulation itself. The SEC's EDGAR system lets anyone run a full-text search across the complete content, including exhibits and attachments, of every filing submitted electronically since 2001. A disclosure buried in a 10-K from 2004 is exactly as retrievable today, by keyword, as a filing submitted this morning. That is a genuine public good: it makes corporate history harder to bury. It is also a resurfacing machine on a fifteen-plus-year time horizon. A short seller, a journalist, or an amateur forum poster who runs a well-chosen keyword search can pull a years-old, already-disclosed fact into the same visual and emotional register as breaking news, with no algorithm or edit war required, just a search box that treats 2004 and this morning as equally reachable. The Yin, Jia, Xu, and Zhang result is the behavioral proof that this kind of retrieval event, not disclosure, is enough to move volume and liquidity on its own.

This is also where the finding connects to earlier work the same authors' field cites. A 2013 study in Scientific Reports found that shifts in how often finance-related Wikipedia pages were viewed carried early signals of subsequent stock market moves, evidence that pageview attention on the platform is not a passive byproduct but a leading indicator worth watching in its own right. And a separate 2013 MIS Quarterly study by two of the same authors found that Wikipedia's information aggregation changes how quickly managers disclose bad news and how sharply investors react to it once they do. Read together with the new experiment, the throughline is consistent: Wikipedia is not a neutral mirror of what companies have already said. It is an active participant in when and how hard the market notices.

What Every Knowledge Platform Is Actually Holding

None of this requires bad actors. Wikipedia's editors who added the litigation content in the experiment were not manipulating the market, they were doing exactly what an encyclopedia editor is supposed to do: keeping a page complete and accurate. Google's ranking algorithm in 2008 was not trying to crash an airline's stock, it was doing exactly what a search engine is supposed to do: surfacing the most relevant match for a query. EDGAR's full-text search is a transparency tool functioning as designed. In every case, the platform behaved correctly by its own stated purpose, and a correct, boring action, resurfacing a true, old fact, was still enough to move retail trading volume by double digits and shift the price at which a stock trades hands.

That is the operating rule for anyone who runs, moderates, or studies a platform where facts about public companies live: your archive is not inert. Every page you host is a live wire connecting a static fact to an audience that may not check its date. The moment your indexing, editing, or search infrastructure changes what gets seen, you have created a market event, whether or not anything in the world actually changed. Wikipedia found this out under carefully controlled, randomized conditions. Google found it out live, on a Monday morning, at the cost of a few chaotic minutes and a company's stock chart. Any platform sitting on years of accurate, disclosed, currently unread information about publicly traded firms is holding the same lever, and does not need to add a single new fact to pull it.