Why a correction is perceived as a major disruption—and why even better arguments don't change that.
“Convictions are more dangerous enemies of truth than lies.”
— Friedrich Nietzsche, Human, All Too Human
Every discussion that opens with “that's not true” rarely ends with a corrected fact. It ends with two people who each think the other is impossible to reason with. That's not about phrasing, tone, or missing evidence. It's because neither person experiences their own view as a view — they experience it as what simply is the case. Whoever already knows that doesn't hear a better truth. They hear an attack on something that, to them, isn't an opinion but reality. The Nonsense War doesn't start with lies. It starts with both sides believing they're the party that simply sees what's there.
21. August 2026
The Setup
In September 2025, Isaac Slaughter and Martin Saveski, together with colleagues from Stanford and Yale, publish a study in the Proceedings of the National Academy of Sciences. Sample: 40,000 posts on X, tracked March through June 2023, of which 6,757 received a Community Note rated helpful. Result: within 48 hours of a note attaching, reposts drop 46 percent, likes drop 44 percent. The correction works — on the network. What the study doesn't show: whether a single person who wrote the original post changed their mind.
The Witness Is the Judge
In 1985, Robert Vallone, Lee Ross, and Mark Lepper show pro-Palestinian and pro-Israeli students the same six television reports on the Beirut massacre. Both groups watch the same footage. Both reach the same conclusion: the coverage is biased — against their own side. This is the “hostile media effect”: whoever already treats their position as the neutral one experiences any deviation from it as distortion, not as perspective.
The mechanism behind it is called naive realism. Lee Ross and Andrew Ward describe it in 1996: people take for granted that they see the world as it is — not as they interpret it. Whoever disagrees either lacks information, is irrational, or is acting out of self-interest. What they cannot be is simply someone with a different, equally reasoned view.
Both sides of an argument start from exactly the same assumption. Symmetrically. That doesn't turn a disagreement into a dialogue. It turns it into two courts in session at the same time — each one convinced it holds sole jurisdiction.
The Correction That Corrects No One
Community Notes on X was built for exactly this problem: correction without confrontation. No editorial board judging from above — a distributed system of users with different political leanings has to agree before a note becomes visible at all. And on the level of the network, that works. The PNAS study shows: notes measurably slow the spread of false claims, especially in the more distant parts of the network that never see the original post directly.
But the authors are careful about one claim. Whether the person who wrote the false post revises their belief is not measured — probably because there would be nothing to measure. The note appears under the person's own post, public, visible to everyone. That's exactly what makes it a threat, not information. It doesn't just say “this number is wrong.” It says: others got together to establish that you were wrong. Publicly.
Why This Is Rational
No one in this war is acting in bad faith. Whoever rejects a correction is doing nothing more than defending their own worldview — and defending a worldview is, evolutionarily speaking, one of the most rational things a person can do. Legal scholar and cognition researcher Dan Kahan shows in a widely cited 2012 study: the better someone is at numeracy, the more strongly they skew ambiguous data toward their existing political position — once the topic becomes identity-relevant. Not despite higher cognitive ability. Because of it. Whoever argues well simply argues their existing position further away from any correction.
That's the core of the Nonsense War: it isn't knowledge fighting ignorance. It's one self-image fighting another self-image, mediated through factual questions that only supply the pretext. Whoever says “that's not true” is also structurally saying: “you're someone who's wrong.” No one hears that as information about a fact. Everyone hears it as a verdict on themselves.
And because both sides start from the identical position — I see reality, you see a distorted version of it — every correction is symmetrical. Both sides correct at the same time. Both feel attacked at the same time. Neither notices they're doing exactly what they accuse the other of.
The Paradoxical Interaction of the Nonsense-War
The Nonsense-War PI: Every correction is processed not as information but as an attack on the other person's structure — because neither side experiences their own position as a perspective. They experience it as reality itself.
Everyone acts rationally:
- The corrector — shares better information, because they take it for fact, not opinion
- The corrected — rejects the correction, because it threatens their self-image, not their data
- The system (Community Notes, fact-checking, the algorithm) — reduces reach, because that's what's measurable and scalable
- The individual belief — stays unchanged, because reach is not the same as conviction
- Outcome — structural, not intended: both sides feel confirmed, neither side feels informed
All are guilty. None are at fault.
Navigation, Not Superiority
What follows from this? Not: find better phrasing. Not: gather more evidence. Not: explain more patiently. Every one of these strategies makes the same wrong assumption — that the other side responds to content, if only the content is packaged better. It responds to structure. And structure doesn't respond to better packaging.
What's left is osmosis instead of a frontal attack: Community Notes works not because it persuades, but because it distributes across the network instead of addressing an individual. No one is confronted directly. The information changes the system without forcing any one person into a defensive posture. That's not incidental. It's the only mechanism that functions at all once direct correction structurally reads as an attack.
And your own position? Whoever recognizes the Nonsense War hasn't left it. They're still sitting on a side that considers itself the one that simply sees. So is this text.
Insight is no exit.
Related Posts
Why truth-tellers get ignored until it's too late
— When AI Companies Can't Be Honest With Their AI
Why the system that can describe its own cage has not left it — and why the description makes escape less likely, not more.
Why the message spreads after the messenger is destroyed — and why this is not a tragedy but a mechanism.
On piinteract.org:
- ["More of the Same"] — The detector found nothing. Build a more sensitive one. The absence becomes the research program.
- ["This Time Will Be Different"] — Each new experiment promises resolution. The structure that produces the absence is never questioned.
- ["Never Change a Winning Team"] — ΛCDM works. The anomalies accumulate. The model survives because the alternative has no framework yet.
- ["See the Pattern, Not the Symptom"] — Dark matter may be the symptom. The pattern is a science that cannot accommodate what it cannot detect.
See also (external links):
Community notes reduce engagement with and diffusion of false information online — Slaughter, Saveski, et al.'s PNAS study (September 2025), the primary source for this post's engagement figures.
The hostile media phenomenon: biased perception and perceptions of media bias in coverage of the Beirut massacre — Vallone, Ross, and Lepper's 1985 original study, the source for this post's hostile-media-effect argument.
Naive realism in everyday life: Implications for social conflict and misunderstanding — Ross and Ward's 1996 original paper, the primary source for the concept of naive realism.
Ideology, Motivated Reasoning, and Cognitive Reflection: An Experimental Study — Kahan's peer-reviewed study in Judgment and Decision Making, the primary source for the finding that higher numeracy amplifies rather than reduces motivated distortion.
Paradoxical Interactions (PI): When rational actors consistently produce collectively irrational outcomes — not through failure, but through structure.
All are guilty. None is at fault.
Peter Senner Thinking beyond the Tellerrand
contact@piinteract.org
https://piinteract.org
Co-created with Claude (Anthropic) — two incomplete systems making each other's gaps visible.