The engineer who never got the credit
There is a familiar character in almost every team. She is the one who, three weeks after a decision goes sideways, turns out to have flagged the exact failure in the meeting where it was made. She said it once, quietly, in a single sentence, and the room moved on. Meanwhile the person everyone remembers as the sharp one — the person who got the stretch project, the nod from leadership, the reputation for judgment — spent the meeting talking almost continuously, and was wrong about the thing that mattered.
We like to believe meetings reward the best thinking. What they often reward is the most thinking out loud. And the gap between those two things has a name in the research literature, and it is not flattering: the babble hypothesis.
What the babble hypothesis actually says
The babble hypothesis is the finding that the sheer quantity of a person's speech predicts whether the group sees them as a leader — largely independent of the quality of what they say. In studies of leaderless small groups, researchers measured how much each person talked and then asked the group to rate who had emerged as the leader. Total speaking time was a strong predictor of those ratings. The content of the contributions, and even measures of the speaker's intelligence and personality, mattered far less than the raw volume of airtime.
Sit with how strange that is. The thing that most shapes who we anoint as capable in a room is not what they contribute but how many seconds they hold the floor. The person who talks twice as much doesn't need to be twice as right, or right at all. They need to be present, audible, and continuous.
This isn't a claim that talkers are frauds. Some of the most talkative people in your meetings are also your best thinkers. The point is narrower and more unsettling: even when talk quantity and talk quality come apart, the group tends to follow quantity. We are running a heuristic, and the heuristic is loud.
Why the brain uses volume as a proxy for competence
The reason this happens is that judging competence in real time is genuinely hard, and the brain reaches for cheaper signals. Assessing whether an idea is actually correct requires domain knowledge, patience, and the willingness to be uncertain. Noticing who is talking a lot requires nothing. So we substitute the easy signal for the hard one, usually without knowing we've done it.
A few well-documented mechanisms stack on top of each other here.
The first is the confidence heuristic. People use the confidence others project as a cue to how accurate they are, because in many everyday situations confidence and accuracy genuinely correlate. But the correlation is loose, and it is easy to fake. Someone who speaks in fluent, unhedged sentences reads as competent even when the underlying claim is shaky.
The second is mere exposure. The more often an idea and a face are paired in front of us, the more familiar and therefore the more agreeable both become. A person who restates their position four times isn't just persuading — they're becoming the ambient default of the conversation.
The third is simple airtime capture. Meeting time is finite. Every minute one person talks is a minute no one else does. A dominant speaker doesn't only add their own signal; they subtract everyone else's. The quiet engineer's one correct sentence has to compete with twenty minutes of confident narration, and it loses on volume alone.
The cost isn't just fairness — it's worse decisions
It would be tempting to file this under office politics: annoying, but survivable. The deeper problem is that the babble hypothesis quietly corrupts the decisions themselves.
Groups are supposed to be smart because they pool many independent perspectives. That advantage only holds if the perspectives are actually independent and actually surface. When perceived leadership tracks talking time, the loudest voice becomes an anchor that the rest of the conversation forms around, and the room starts optimizing for agreement with that voice rather than for the truth. The people most likely to be overlooked — the reserved, the junior, the ones who think before they speak, the ones translating from a second language — are often carrying exactly the disconfirming information the group needs.
There's a cruel efficiency to it. The person most rewarded for talking has the least incentive to stop, and the person whose dissent would be most valuable has just watched the room reward volume over accuracy. So they say their one sentence, get no traction, and stop. The meeting confuses the resulting silence for consensus.
How to interrupt the pattern
The useful thing about the babble hypothesis is that once you can name the bias, you can design against it. None of these fixes require anyone to become a different personality.
Separate idea generation from discussion. Before opening the floor, have everyone write their position or their top concern silently for two minutes, then collect them. This is sometimes called brainwriting, and it works because it captures the quiet engineer's sentence before airtime capture can bury it. The best idea enters the room without having to be the loudest.
Watch airtime as a metric, not a vibe. If you run meetings, you can literally track who has spoken and who hasn't. A round-robin at a decision point — going person by person, briefly — forces the distribution of talk to flatten. You will be surprised how often the person who has said the least has been sitting on the objection that changes the call.
Ask for the case against. Confidence and volume both collapse a little when a room is explicitly invited to argue the other side. Assigning someone to make the strongest case against the emerging decision converts dissent from a social risk into an assigned duty, which is the only way many people will voice it.
Decouple the record from the performance. Judge decisions on what was actually proposed and by whom, in writing, after the fact. When the reasoning is written down and revisited, the person who was quietly right becomes visible, and the person who was confidently wrong stops getting graded on delivery.
The throughline in all of these is the same: stop letting real-time impressions do the work that a durable record should do. In the moment, volume wins. On the page, accuracy has a chance.
The quiet sentence deserves a second hearing
That last point is where a meeting postmortem earns its keep. Most of what goes wrong in the babble hypothesis is invisible while it's happening and obvious in hindsight — the flagged risk everyone forgot, the confident claim that didn't survive contact with reality, the person whose airtime and whose accuracy pointed in opposite directions. meetingmortem exists to make that hindsight routine: to capture what was actually decided and by whom, so the record reflects who was right rather than who was loudest, and so the quiet sentence gets a second hearing before it becomes next quarter's expensive mistake.
If your best thinking keeps losing to your loudest talking, that's not a personality problem to fix — it's a process to design. You can start by seeing it clearly: meetingmortem.lumenlabs.works.