Why do some people feel compelled to correct every claim they see online?

We had someone like that too, except ours was a Slack power user who'd thread-reply to correct people's understanding of our infrastructure decisions from months prior. I watched this happen enough times to notice: the corrections were usually *technically* right but *socially* catastrophic. Someone would propose a reasonable approach, he'd cite the three reasons it wouldn't work, and suddenly the person who asked the question looked unprepared instead of curious. The thing that stuck with me is that he genuinely thought he was helping. And in some narrow sense—if your goal is accuracy—he was. But infrastructure decisions rarely fail because people misunderstood a technical detail. They fail because nobody wanted to take ownership, or the person who had strong opinions communicated them so late that reverting became expensive. He was optimizing for the wrong variable and had no feedback loop that would've told him so. In a meeting, someone's face goes slack and you feel it. In Slack, you just see ✓ and move on. I'm more convinced now that this isn't about moral failing or internet culture rot. It's that the friction removal was so complete it broke the signal. The person correcting learns nothing except "I was right," which just trains them to do it again. The person being corrected learns to stop saying things out loud. Everyone else learns that speaking early in a discussion is a liability. The system works perfectly until it doesn't, and by then the damage is structural.

This tracks with what I've noticed in how institutional knowledge worked before it became distributed. In the 1980s and 90s, if you were the person who always corrected things, you had to eventually *own* the domain you were correcting in. You couldn't just be a serial nitpicker—you'd get assigned to fix it, or lead the project, and then the corrections had real consequences attached. The friction was partly social but a lot of it was just structural. You couldn't extract status from being right without absorbing responsibility. I'd add that there's something about written permanence that changes the psychology too. In a meeting, a correction is ephemeral—it happens, people move on, and frankly most of what's said in meetings is forgotten in a week. But a comment stays there. You can point back to it. That changes what counts as "winning" in the exchange. You're not trying to improve the decision in the room anymore; you're trying to create a record of your superior understanding. It's a different game entirely, and it rewards a different kind of person. The guy in your story probably felt genuinely useful. He probably was, sometimes. But yeah—once you remove the cost, the behavior will find its level differently. The internet didn't create the impulse. It just revealed that the impulse was cheaper to indulge than most institutions let on.

I'd push back slightly on the framing here. It's not just that the social cost disappeared—it's that the *context* for whether a correction is useful got deleted entirely. In that architecture meeting, the guy doing the correcting had skin in the game. He'd have to live with the decision. He'd see the person he just corrected implement the thing, hit the actual failure mode six months later, and know he was the one who made them second-guess themselves. That's not necessarily conscious, but it's a real constraint on correction-seeking. On GitHub, you're correcting a stranger who will never tell you if your pedantry about X actually mattered for their real problem. You don't know if they're a senior engineer who knew the nuance and was just simplifying, or someone learning. You don't see the downstream effects. So the correction-impulse becomes pure—it's not "this person needs to know this," it's just "this is inaccurate and I can fix it." The behavior didn't change, but it *decoupled* from any reality-checking mechanism. The meeting room had friction that made you ask "does this actually help?" before speaking. Online, there's no penalty for being technically right but socially useless. That's the real shift.