There is a particular kind of comment that has become the lingua franca of the internet. It surfaces under articles about public transit, insisting that the writer clearly doesn’t understand because my commute is different. It materializes in response to scientific studies, armed with a cousin’s anecdote that proves data wrong. “Well, that wasn’t my experience,” the comment begins, as if personal experience were a skeleton key capable of unlocking any empirical door.

Welcome to the era of the Bean Soup Theory, where every piece of content is filtered through the self-centered lens of “what about me?” and personal anecdotes masquerade as empirical truth. The bean soup problem is more than an online quirk; it’s a symptom of a widespread confusion about what constitutes evidence, what defines a category, and how to participate in collective reasoning about anything at all.

The “Bean Soup Theory” originated from a viral TikTok moment — one simple recipe video, tens of thousands of comments demanding bean-free alternatives. The recipe’s purpose was clear, but the response was not: viewers projected personal preferences onto content, derailing the discussion and demanding a universe that catered to their specific needs. In miniature, it’s a case study in main character syndrome, algorithmically supercharged.

But the phenomenon is not limited to soup. It’s everywhere. Parenting videos devolve into critiques of sleep methods, product demos become battlegrounds for ingredient substitutions, and every general statement online is met with a chorus of “but what about me?” The original intent of content — context, nuance, collective relevance — is flattened by a relentless focus on the individual.

In the algorithmic agora of the internet, personal stories have become the currency of credibility. Social media platforms reward relatability, not rigor. The more a story “feels true” to someone, the more likely it is to be amplified — regardless of whether it actually is true. As a result, anecdotal evidence is elevated to the status of data, and the line between perspective and proof grows even fainter.

Cognitive science has a name for this: the anecdotal fallacy. It’s the ancient error of treating a personal experience as if it constitutes fact for a general conclusion. We are built to respond to narrative. We are not built to remember that response is not the same as reasoning.

Your grandmother’s experience with a medication is not a clinical trial. Your uncle’s financial success does not disprove economic inequity. Yet online, the plural of anecdote has somehow become “proof.” The distinction matters: a personal story can humanize or suggest avenues for inquiry, but it cannot, by itself, validate a generalization.

The numbers are sobering. According to a Gallup analysis of data from the U.S. Department of Education’s Programme for the International Assessment of Adult Competencies (PIAAC), approximately 130 million American adults — 54 percent of those between the ages of 16 and 74 — read below a sixth grade level. Nearly one in five reads below a third grade level. Unsurprisingly, the Barbara Bush Foundation for Family Literacy notes that counties with the lowest literacy rates often experience the highest rates of poverty, poor healthcare, and limited economic mobility. The estimated economic cost? $2.2 trillion annually in lost productivity.

But the implications extend beyond economics into the realm of cognition itself. Jean Piaget, the Swiss developmental psychologist, describes the “formal operational stage” — the cognitive milestone, typically beginning around age 11, where individuals develop the capacity for abstract thought. This is when children begin to reason about the implausibility of Santa Claus, not because someone at school told them the truth, but because they’ve developed the cognitive architecture to weigh evidence against belief.

What Piaget’s research revealed — and what subsequent studies have confirmed — is that formal operational thinking is not guaranteed. Many adults never develop or consistently apply formal abstract reasoning, or they use it only in domains where they have significant experience. As one academic overview notes, this stage “requires shared definitions, sustained attention, and the ability to follow an argument past your own ego.” Not everyone practices these skills. And now, on the internet, millions of people are attempting to engage in discourse without them.

What happens when you attempt to discuss statistical patterns with someone operating at the concrete operational stage — the developmental level typically associated with children ages seven to eleven — is that every general statement gets derailed. “Most X does not mean all X,” they remind you, as if you had claimed otherwise. Definitions become negotiable. Discussions collapse into “well, that’s not what happened to me.” The cognitive mode is not malicious, it is, in the clinical sense, underdeveloped. And it has found, in social media, its perfect habitat.

Social media platforms, by design, optimize for engagement. The algorithms that curate our feeds analyze every click, like, and share to construct what Eli Pariser called “filter bubbles” — personalized information ecosystems that surface content aligned with our existing preferences. Research published in PNAS examining more than 100 million pieces of content found that homophilic interactions — i.e., engagement with like-minded others — dominate online behavior, particularly on Facebook and Twitter (now X).

The effect is recursive. Algorithms identify what you already believe and serve you more of it. You engage, reinforcing the algorithms’ model of your preferences. Your feed becomes increasingly narrow, your interactions increasingly homogenous. What emerges is not dialogue but a series of monologues, each participant convinced of their centrality, each comment section a hall of mirrors reflecting individual grievances back upon themselves.

The bean soup theory describes, in miniature, what this looks like at scale. Every piece of content is interpreted through the lens of personal relevance. A recipe is not merely a recipe; it is an opportunity to declare that you, specifically, cannot eat this thing. An Article about a general phenomenon is not information to be evaluated; it is an occasion to announce your exception to it. The algorithmic architecture of social media trains users to see themselves as the protagonist of every interaction, which trains them to respond to collective reasoning as if it were a personal attack.

This is not a matter of political affiliation or educational credentials. Studies suggest that even highly educated individuals fail to appropriately discount flawed research when it aligns with their prior beliefs. A 2021 paper in Cognitive Research: Principles and Implications found that participants could often identify methodological weaknesses in studies while still electing to act on the flawed conclusions — a dissociation between recognition and application that suggests the problem is not simply one of knowledge but of practice. We know, in the abstract, that anecdotes are not data. We forget this, in the specific, when the anecdote is our own.

When every comment section is a digital focus group for one (read: hellscape), creators are left to navigate endless demands for personalization. This isn’t just exhausting — it’s unsustainable. Over 50% of content creators report burnout due to the impossible task of appeasing every viewer. Platforms reward engagement, not nuance; creators face disproportionate criticism for failing to cater to niche preferences.

There is a tendency, when discussing these phenomena, to reach for technological solutions. Perhaps better algorithms. Perhaps media literacy curricula. Perhaps, as some researchers have suggested, interventions that deliberately expose users to ideologically diverse content. And yet the literature offers limited reassurance. A field experiment conducted on Twitter found that exposing users to opposing viewpoints actually increased political polarization — a finding consistent with the idea that the problem is not simply exposure but the cognitive infrastructure for processing disagreement.

What Piaget understood, and what the architects of social media perhaps did not, is that abstract reasoning is a skill that must be developed and practiced. It required the ability to hold contradictory ideas in mind, to distinguish between categories and examples, to recognize that general patterns and individual exceptions can coexist without mutual annihilation. The formal operational thinker can understand that “bean soup is a dish containing beans” and “you prefer dishes without beans” are both true statements that require no reconciliation. The concrete operational thinker experiences the latter as a refutation of the former.

The comment sections of the internet are not debates. They are diagnostic instruments. They reveal, in aggregate, the distribution of cognitive modes across a population — how many people have learned to separate their feelings from their arguments, how many have practiced following a logical chain past their initial emotional reaction, how many understand that an appeal to personal experience is not a counterargument but a category error. The numbers are not encouraging.

The bean soup people are not an aberration. They are a demographic.

The bean soup theory is, finally, a theory about the limits of collective reasoning in an age of algorithmic individualism. When every feed is personalized, when every user is the protagonist, when every comments section rewards the assertion of personal exception over the acknowledgement of general pattern, the conditions for shared understanding erode. What remains is not a public town square but a series of private grievances, each shouting into a void that has been optimized, with exquisite precision, to shout back.

The beans were never the point. But they were the catalyst to the question: can we still talk about anything at all?

Editor’s Note: Full access to The Sixteenth will soon require a subscription. More info to come.