Ask a table of six people whether the Great Wall of China is visible from space and you will get at least one confident yes. It is one of the most widely believed facts in the world, taught in classrooms for decades, and it is false. Astronauts have looked. The wall is long but narrow, rarely more than six or seven meters wide, and made of stone that matches the terrain around it. From low orbit you can pick out airport runways and highways, which are wider and brighter, while the wall itself dissolves into the landscape. Chinese astronaut Yang Liwei came home in 2003 and said plainly that he could not see it, which caused enough of a stir that China revised its textbooks.
Now put that claim on a dinner table and watch what happens next.
Fifteen years ago, someone would have pushed back, and the table would have been busy for half an hour. How high is space, exactly, and does it count if you can see it from a plane. Why would a wall be visible when a highway is not. Somebody would work out, roughly and wrongly, how wide something needs to be to be seen from four hundred kilometers up. Someone else would ask why everyone believes this in the first place, and the table would stumble onto the real story, which is that the claim predates spaceflight entirely, appearing in print as early as the 1930s, a guess that hardened into a fact before anyone could check it. Nobody would arrive at a clean answer. Everybody would arrive somewhere.
Today a phone comes out. Eleven seconds later a box at the top of the results says: no, it is not visible to the naked eye from orbit. The table nods. The subject closes.
The verdict is correct and nearly useless, because the interesting material was never the yes or the no. It was the reasoning underneath: what visibility even means, how belief outruns evidence, why a fact nobody verified survived seventy years of schooling. None of that fits in a box.
I want to name the thing being lost, because it is not conversation in general and it is not attention span, which is where this observation usually gets parked. What disappears is the interval between a claim and its resolution. Call it Verdict Latency: the elapsed time from the moment a disagreement is stated to the moment some authority settles it. For most of human history that number was measured in hours, days, or never. It is now measured in seconds, and it is still falling.
The interval was doing real work. Two people holding incompatible positions have to reach for supporting material, and reaching is where the discovery happens. You go looking for evidence and find three adjacent things you were not looking for. You state your position out loud and hear, for the first time, that it does not survive being said. Your friend brings up a case you had never considered, which forces you to narrow your claim or drop it. None of that requires anyone to be right. It only requires that the matter stay open long enough for both minds to move.
This is the road trip, except the comparison needs to be pushed further than navigation. Imagine Google Maps offered a vortex: you type the destination, press go, and you are simply there. Every driver on earth would use it, and every driver would lose the same things. The roadside diner you only stopped at because the tank was low. The bakery on the corner of the street you turned down by mistake. The argument about which route was faster that taught you how the city fits together. The vortex is not a better trip. It is the end of the trip as a category, arrival with the traveling removed. A search box is that vortex for conversation: you press go on a disagreement and materialize at the verdict. Detour Yield is the value produced by everything a conversation touches on its way to a conclusion it may never reach, and the vortex sets it to zero by design, invisibly, because no accounting of efficiency has a column for the bakery you never found.
Takeaway: the destination was never the point of the argument. The argument was a search process, and search processes generate value in proportion to how much ground they cover, not how fast they terminate.
Your brain reasons better with an opponent
Two cognitive scientists, Hugo Mercier and Dan Sperber, spent years asking an awkward question about human reasoning: if it evolved to help us find the truth, why is it so bad at that job when we reason alone? Their answer, laid out in The Enigma of Reason, is that truth-finding was never the primary function. Reasoning evolved to help us justify our positions to others and to evaluate the positions others push at us. It is social equipment. That is why a person reasoning alone drifts comfortably toward whatever they already believed, while a group containing real disagreement routinely outperforms its smartest member. The machinery was built to run with someone across from it. Remove that person and the machinery still runs, but it stops correcting.
The learning research points the same direction from a completely different starting point. Robert Bjork, a memory researcher at UCLA, spent his career studying what he called desirable difficulties: conditions that slow learners down and make performance look worse in the moment, yet produce stronger retention weeks later. Generating an answer yourself, even a wrong one, beats reading the correct one. Guessing before being told beats being told. Effortful retrieval strengthens memory in a way that fluent reception never does, which is why the material you struggled with in school is the material you still have.
An argument between two friends is retrieval practice with social stakes attached. You are pulling knowledge out of storage under pressure, in public, with someone waiting to find the hole in it. Very few learning conditions are that good. The vortex dissolves all of it and swaps an effortful reconstruction for an effortless arrival.
I have written before about Phantom Fluency, the state of producing outputs you cannot explain, defend, or iterate on. What happens at that dinner table is its conversational cousin. The table now knows the wall is not visible from orbit. Not one person there could explain why a runway is and the wall is not, and none of them will remember the exchange by Thursday, because nothing in it required them to build anything.
Difficulty is the mechanism
Most of my work puts friction in the denominator. Value rises when the cost of acting on something falls, and the formula I keep returning to holds up:
Value ∝ (Existing Demand × Relevance) / Friction
Conversation breaks that relationship, and it took me a while to accept it. In a debate, a certain amount of friction belongs in the numerator, because the difficulty is the mechanism rather than the obstacle.
Insight ∝ (Divergence × Verdict Latency) / Perceived Finality
You read this as: how much two people come away understanding rises with how far apart their starting positions were and how long the disagreement stayed open, and falls the more absolutely final the settling source feels. Two friends who half-agree, resolve it in nine seconds, and treat the result as beyond appeal produce almost nothing. Two friends who start far apart, sit in it for twenty minutes, and end with a provisional answer they both suspect is incomplete will still be chewing on it in the car.
Perceived finality is the term most people skip past. A search result carries a tone of settledness that a knowledgeable friend does not. Your friend can be interrogated. The box at the top of the page arrives already dressed as a conclusion, and it closes the matter even when the underlying question is contested, badly specified, or has four defensible answers depending on how you frame it.
There is a second cost, slower and harder to notice. The capacity itself fades with disuse, and it fades the way muscle does, fastest at the start. Someone who has not had a real disagreement in two years does not simply argue less often. They begin to find disagreement itself intolerable, which is a different and worse condition, and it arrives without announcing itself.
Watch a team argue about why one product holds its customers through the first year while another loses a third of them. For ten minutes the discussion is genuinely productive, because nobody knows, and the guesses on the table range from pricing to onboarding to something in the emotional register of the brand. Then someone opens a laptop and puts the retention curve on the screen, and the meeting ends. The curve answered what. Nobody got to why, and strategy lives almost entirely in why.
Resolution Reflex is the automatic reach for the device the instant a claim is contested, and inside organizations it works as a discussion terminator that everyone in attendance experiences as diligence. The person who pulls the number looks like the most rigorous person at the table. They have in fact just ended the only part of the meeting that was producing anything new.
At national scale it looks like public debate where both sides arrive carrying links. Nobody is reasoning. Everyone is presenting exhibits. The exchange runs until one party produces a source the other cannot immediately discredit, which resolves nothing, because neither person built their position and so neither can be moved by an attack on it. Beliefs held without construction cannot be revised. They can only be swapped.
Takeaway: a society that has outsourced verification has not become better informed. It has become better supplied, which is a different thing, and it has lost the practice of changing its mind in front of another person.
Where speed genuinely wins
I do not want to romanticize ignorance, and there is a version of this argument that curdles into nostalgia for being wrong together. Instant answers are a spectacular improvement for a large class of questions. Dosage. Flight times. Whether the shop is open. Which artery the surgeon is looking at. Anything where a wrong answer carries a real cost and a right answer requires no further thought, take the vortex, and take it before you open your mouth.
The loss concentrates in questions where the reasoning matters more than the fact. Why audiences behave the way they do. Whether a decision was defensible. What a historical event means for the one in front of us. How much of a person’s success came from skill. Those questions have retrievable pieces inside them, and pulling a piece early kills the part that was worth doing, the way reading the last page ruins a novel while checking a train schedule ruins nothing.
The practical move is smaller than it sounds. Guess before you look. Say what you think the answer is and why, let the other person say theirs, and only then reach for the phone. Prediction before feedback is one of the most reliably effective learning conditions ever measured, and it costs about forty seconds. It also converts the search from an ending into a scoring event, which keeps the conversation alive past the verdict, because now there is a discrepancy to account for.
Two friends who both guessed wrong about the wall have another twenty minutes of material, and they will end up somewhere near the question of how many other things they believe that nobody ever checked, which is a good question with no bottom to it. Two friends who looked it up first have nothing left, and one of them will reach for the phone again in four minutes, for a different reason, and the evening will go that way.
If you found this useful, subscribe to STRATEX by Naz for more on how people, organizations, and markets actually work.