You think they heard you. They didn't.

Here's a small, mean experiment you could run on yourself. Next standup, say your update exactly the way you normally would. Then, immediately after, ask the person next to you to repeat back — not the topic, the actual claim — what you just said. Not "something about the auth thing." The specific, checkable content. Most people, doing this for the first time, are quietly unsettled by the gap.

That gap isn't a communication skills problem. It's two separate, well-studied cognitive biases stacking on top of each other, and they both point the same direction: toward you overestimating how much of what's in your head actually made it into anyone else's.

Bias one: you think your face is a broadcast tower

Thomas Gilovich, Kenneth Savitsky, and Victoria Medvec ran a series of experiments in the late 1990s on what they named the illusion of transparency: the tendency to systematically overestimate how visible our internal states are to other people. In one memorable version, participants tasted drinks — some laced to taste awful — while trying to keep a poker face, and then guessed how many observers could tell which drink was the bad one. Participants consistently guessed far higher than the actual number of observers who could tell. The internal experience of "I am obviously struggling right now" simply does not translate outward at anywhere near the intensity it's felt inward.

Translate that from facial expressions to standup sentences and the mechanism is the same, just relocated: when you say "I think it's the retry logic," the full weight of everything that led you there — the three hours you spent last night staring at the retry wrapper, the specific log line that tipped you off, the mental model of exactly which code path you mean — is vivid and present to you. None of that context transmits in the six seconds it takes to say the sentence. What lands for everyone else is six seconds of words, stripped of the internal certainty that made those words feel obviously clear when you said them. You feel like you broadcast a signal. You transmitted a fragment.

Gilovich and his co-authors ran a companion study that makes the size of the gap vivid: they had people lie, under mild pressure, while observers tried to catch them, and asked the liars afterward to estimate how obvious their lying had been. Liars consistently overestimated how transparent their deception was — feeling caught out internally translated into believing they looked caught out externally, when observers, in fact, caught them at close to chance levels. If deliberate, high-stakes deception is that hard to read from the outside, an offhand standup sentence — delivered with none of a liar's elevated self-monitoring — should be assumed even less legible than it feels to the person saying it.

Bias two: you can't remember what it was like not to know

The second bias comes at the problem from the other direction. Colin Camerer, George Loewenstein, and Martin Weber ran economic experiments demonstrating what they called the curse of knowledge: once you know something, it becomes very difficult — not impossible, difficult — to accurately model the mental state of someone who doesn't know it. In their experiments, better-informed participants in a market setting couldn't fully suppress their own private information even when it actively hurt their negotiating position to reveal it; the knowledge leaked through, unconsciously, because the informed party couldn't cleanly simulate what "not knowing this" felt like anymore.

In a standup, this shows up as jargon that isn't meant to exclude anyone and does anyway. You say "yeah, fixed the auth thing" and mean, with total internal clarity, the specific OAuth callback token-refresh bug you've been staring at for two days — and you say it in five words because, to you, those five words are already fully specified. They are not fully specified to anyone who hasn't been staring at the same code. You've forgotten what it was like not to know which bug "the auth thing" refers to, because the curse of knowledge doesn't let you hold that ignorance in mind anymore, even on purpose.

Illusion of transparency says: what you feel, others cannot see nearly as clearly as you think. Curse of knowledge says: what you know, you cannot un-know long enough to say clearly to someone who doesn't. Stack them and you get a speaker who is systematically, honestly, doubly convinced they communicated more than they did.

Why standups are an unusually good breeding ground for both

Neither bias needs a standup specifically to operate — they show up in marriages, in classrooms, in boardrooms. But a few features of the standup format make them worse than average. First, the time pressure: fifteen minutes split eight ways leaves almost no room for the kind of elaboration that would otherwise catch a curse-of-knowledge gap, so speakers default to the shortest phrasing that feels complete to them, which is exactly the phrasing most contaminated by what they already know. Second, the audience is the same small group every single day, which breeds a false sense of shared context — "they were in the incident channel with me last night, they'll know what I mean" — that quietly stops being checked as team membership shifts, as people miss a day, as the specific incident recedes from everyone's memory except the speaker's. Third, and most corrosively, the format actively discourages the one behavior that would catch the gap: asking a clarifying question costs airtime that isn't yours, in a meeting explicitly designed to move fast, so most ambiguity goes unchallenged not because it wasn't noticed but because challenging it felt like the wrong kind of contribution to make.

Put those three features together and you get a meeting format that is, almost by design, worse at correcting these biases than an ordinary hallway conversation would be — which is a strange thing to say about a ritual whose entire stated purpose is keeping everyone informed. A hallway conversation has no clock and no queue of other speakers waiting their turn, so a confused listener will usually just ask. A standup removes both of those safety valves at once, which means the two biases this post describes don't just survive the format, they're mildly selected for by it.

A thought experiment: the standup as a radio with no receiver light

Imagine a walkie-talkie with a transmit light but no receive light — no way to see whether the person on the other end actually got the message, only a light that confirms you pressed the button and said something. That's a standup. You get feedback that you spoke (people nodded, moved on to the next person), but essentially zero feedback on whether the specific, checkable content of what you said actually landed in anyone else's head the way it existed in yours. The nodding isn't evidence of reception. It's evidence that a socially appropriate amount of time has passed since you stopped talking.

Run this across a whole team, every person doubly biased in the same direction — overestimating their own transparency, unable to fully model what their listeners don't already know — and a fifteen-minute standup produces something closer to fifteen minutes of parallel transmission than of actual shared understanding. Everyone leaves the room believing information was exchanged. Some of it was. A meaningful amount of it existed fully formed in one person's head and landed, elsewhere, as a fragment nobody quite has the full shape of.

Why this points back to the sentence, not the speaker

Neither bias is a character flaw, and neither is solved by "communicating better" as a New Year's resolution — they are structural features of how minds work, documented independently, decades apart, in completely different experimental settings, and they don't go away because you're aware of them (Gilovich's own subjects didn't stop overestimating their transparency after being told about the effect; awareness helps less than you'd hope). What actually helps is not asking speakers to somehow transmit their full internal state through six seconds of speech — that was never going to work — but keeping the one thing that survives the gap regardless: the verbatim sentence, attached to its speaker, available later for anyone to check against what it turned out to mean. You don't need everyone in the room to have perfectly received "I think it's the retry logic" in the moment. You need the sentence to still exist, unedited, when the telemetry settles the question of whether the retry logic was actually it.

You think they heard you. Historically, reliably, they heard less than you think — and the fix was never a clearer voice. It was a better record.