There is a particular kind of silence that falls over a trading floor when something that shouldn't matter suddenly does. On the morning of Monday, February 23, 2026, that silence happened across thousands of terminals simultaneously — triggered not by a central bank announcement, not by a geopolitical crisis, not by a quarterly earnings disaster. It was triggered by a Substack post.
The post, authored by a thirty-something independent analyst named James van Geelen under the banner Citrini Research, was framed as a speculative memo from June 2028 — a fictional intelligence briefing from a near-future world in which AI had not failed, but had worked exactly as advertised. Van Geelen called it the "Global Intelligence Crisis." By noon, the Dow was down 821 points. IBM had shed 13% of its value — its worst single-day loss in 25 years. DoorDash, American Express, MongoDB, AppLovin, Blackstone, Visa and Mastercard were all bleeding. The software sector ETF hit a 52-week low, having shed 23% since January. Over $285 billion in market capitalisation evaporated before the closing bell.
A blogger did that. So the question is: what exactly just happened?
The Scenario That Wasn't Supposed to Move Markets
Van Geelen's piece was not a research report. It carried no financial ratings, no price targets, no compliance disclosures. It was speculative fiction — a vivid, detail-rich thought experiment about 2028 in which AI's productivity miracle triggers a deflationary doom loop. The logic was seductive precisely because it inverted the consensus: AI doesn't fail and cause the crash. It succeeds — and that's what causes the crash.
The scenario works like this. AI agents flood white-collar industries with near-perfect efficiency. Consulting, software, legal, and financial analysis — all of it becomes commoditised overnight. Profit margins compress as competition intensifies and pricing power collapses. The high-income professionals who powered consumer spending lose their jobs or see their wages cut. A spending contraction follows. The S&P 500 falls 38%. Unemployment hits 10.2%. The Federal Reserve, having spent its monetary ammunition, has no obvious lever to pull.
It is, as economist Noah Smith pointed out in his rebuttal, "a scary bedtime story." Citadel Securities published a formal demolition of the thesis, noting that AI productivity gains historically generate new economic activity rather than simply destroying existing value. Critics across Wall Street and academia lined up to call it flawed, overstated, and logically incomplete.
And yet. The market moved anyway.
When Three Signals Arrive Simultaneously
To understand the full force of February 23rd, you have to understand what else arrived that morning alongside the Citrini post. Because markets didn't just react to a blog. Markets reacted to a pattern — and Citrini was merely the piece that made the pattern legible.
Nassim Taleb, author of The Black Swan and arguably the world's most famous theorist of catastrophic risk, appeared on Bloomberg the same day with a message that could have been written as Citrini's academic footnote. Software bankruptcies, he said, are coming. The AI rally, he warned, would be eradicated — as it always is when early pioneers get displaced by the models they helped create. He pointed to the quiet structural fragility hiding beneath the AI euphoria: companies loading onto long-term debt to fund infrastructure whose returns are years away, while AI is simultaneously destroying the revenue streams that would service that debt. This wasn't tail-risk speculation from Taleb — it was almost a geometric proof.
Then came Anthropic. Just weeks earlier, Anthropic had launched a legal AI plug-in for its Claude platform. Legal software stocks had plunged. Now, on February 24th, Anthropic announced ten new enterprise plug-ins: investment banking, wealth management, private equity, human resources, engineering, design — integrating directly with Salesforce, DocuSign, Slack, FactSet, LegalZoom, Intuit, and Gmail. Each announcement was a pointed finger at an incumbent. Each product was a question mark placed over a revenue line. Anthropic called them "partnerships." The market read them as threats.
Three signals. Three different registers — narrative, theory, product reality — all converging on the same underlying thesis within 48 hours. The Citrini post didn't move markets. The convergence did. Van Geelen had simply provided the story that made the pattern legible to enough people at the same time.
The New Architecture of Market Fear
Here is the deeper what if worth sitting with: what if we have permanently crossed a threshold where a sufficiently distributed, sufficiently credible independent voice can move global capital markets?
This has always been theoretically possible. What's new is the infrastructure that makes it probable. Citrini Research is the #1 finance Substack by following. Van Geelen earned that position the hard way — he called Nvidia early, he called GLP-1 weight-loss drugs early, and he built a reputation for being ahead of consensus. When a person with that track record publishes a scenario, the question financial professionals ask is not "is this fiction?" but "is this plausible fiction?" And plausible fiction, in markets wired for reflexivity, is enough.
There is also a regulatory asymmetry at work. A Goldman Sachs analyst who published a 2028 recession scenario with specific company callouts and a 38% index decline would face a compliance review before the ink dried. Van Geelen faces no such constraint. Independent research newsletters operate in a different legal and reputational space — bolder, faster, less filtered. In a world of real-time information and algorithmic amplification, that asymmetry is a structural edge. The sell-side is playing chess. The independent analyst is playing speed chess with no clock.
The Prophecy Problem
What makes the Citrini moment philosophically unsettling is not that the scenario was right or wrong — it's that the act of publishing it may have made it more likely to become right.
Markets are not passive observers of economic reality. They are participants in it. When $285 billion in software market cap disappears in a single session, something real happens: enterprise customers delay software renewals, startups find fundraising harder, CIOs pause procurement decisions, and software companies begin laying off staff to defend margins. The doom loop Citrini described — AI compressing white-collar value, triggering spending contraction, spiralling into recession — doesn't require a 2028 trigger. A sufficiently viral 2026 thought experiment about 2028 can begin the chain reaction today.
Taleb would recognise this dynamic immediately. It is the classic fragile system property: a system that reacts to the description of its own vulnerability is already vulnerable. The Citrini post didn't expose a fragility. It revealed that the fragility had been there all along, waiting for the right story to make it visible.
What We're Really Asking
Strip away the market mechanics, the plug-in announcements, and the Substack follower counts, and the February 23rd episode is really asking a question that ThinkingIf readers will recognise: at what point does the map become the territory?
Van Geelen's 2028 memo was a map — a fictional one, clearly labelled as such. But the territory — an AI economy reshaping labour markets, compressing software margins, destabilising incumbent revenue models — is not fictional at all. Anthropic's ten plug-in announcements are not fiction. The software sector's 23% year-to-date decline before the post was published is not fiction. Taleb's warning about debt-laden AI infrastructure is not a bedtime story.
What the Citrini moment revealed is that the public imagination — or at least the financial imagination — was already primed for this scenario. The post was kindling dropped on a fire that had been smouldering for months. Someone was going to write that story. Van Geelen just wrote it first, clearly enough, and distributed it to the right people at the right moment.
And here is the uncomfortable corollary: if the next Citrini post goes further — if someone describes, with equal clarity and credibility, a specific mechanism by which AI destroys value in energy markets, or healthcare, or defence — those sectors will sell off too. Not necessarily because the scenario is accurate. But because we have now confirmed, empirically, that the story is sufficient.
The line between science fiction and strategic planning has always been blurry at ThinkingIf. What February 23rd, 2026, demonstrated is that the line between speculative narrative and market-moving research has now dissolved entirely.
We should all be asking what story gets published next — and whether, by the time we read it, it will already be true.
ThinkingIf.com — Where Reality Meets the Edge of Possibility.
