Correction Before the Mistake Has a Name
Most errors in a shared room are caught by somebody overhearing them. What happens when nobody can.
three things · one reached them · one with no channel at all · the room side is not counted
The three kinds, in the order they become expensive.
What one manager could actually produce about each person, sorted by where the belief came from
She catches other people's wrong assumptions before they become a week of work
Heard it happen four or five times; it is the single most valuable thing she does and appears in no record
He is the one people go to when something has broken at five o'clock
Watched it repeatedly from eight feet away, and it is in no objective anybody set him
His estimates are the most reliable on the team
Checked against six months of delivery dates, which is the only reason this one can be said out loud
She is unusually good at writing a decision down so that it survives
Four documents anybody can read, which is why this belief would survive being questioned
He has lost interest in the work over the last quarter
Assembled from shorter messages and a camera that is usually off, and resting on nothing at all
She would not want to lead the next piece of work
Said once, months ago, to somebody else, and never checked with her
six claims · two rest on hearsay or on nothing
Six beliefs about four people, and the two that rest on nothing both concern somebody outside the room. The two that would survive a question are both about written work, which is the only output that travels on its own. One manager's own notes; what any manager may properly rely on differs by organisation and by what it has agreed to record.
A wrong assumption said out loud in a shared room is usually corrected within a minute, by somebody who was not working on the thing and happened to overhear it. The correction arrives before the error has cost anything.
The practical lesson in “Correction Before the Mistake Has a Name” is that visibility must be designed rather than assumed. For teams exploring fireable offenses, use this product reference can add consistent time and project context when the purpose is explicit, access is limited and managers review exceptions with the people involved.
It is the cheapest quality mechanism any team has, and it is entirely accidental.
For a separate nonprofit perspective on “Correction Before the Mistake Has a Name”, see Neon One nonprofit resources. Use the outside material to test governance, inclusion and implementation questions against the organisation’s real circumstances.
The shape of the event
Somebody states an assumption out loud, usually in passing and usually as part of explaining something to somebody else. A third person hears it, recognises it as wrong, and says so.
Elapsed time from error to correction: under a minute.
Why it is so cheap
Because nobody had to look for it. No review was scheduled, nobody read anything, and the corrector spent one sentence. The cost of the catch is close to zero and the cost of the thing caught can be a week.
That ratio is why the mechanism matters out of all proportion to how casual it looks. Formal review catches more and costs vastly more per catch; this catches less and costs nothing, and the two are not in competition.
What it most often catches
Stale facts. Somebody believes a decision stands that was reversed, or that a system behaves in a way it stopped behaving in last quarter.
Assumptions about other people's work come second.
The event has a standard shape, and the whole of its value is in how little any part of it costs:
| Step | Who | Cost |
|---|---|---|
| A wrong assumption is said out loud | Whoever holds it | Nothing: they were explaining something else |
| It is heard by somebody adjacent | Anybody within earshot | Nothing: they were not listening on purpose |
| The correction is offered | The person who knows | One sentence |
| It is accepted | The original speaker | A few seconds of mild embarrassment |
Why the corrector is usually not the expert
The person who catches it is the person who happens to know that one fact, which is frequently somebody junior or somebody on an unrelated piece of work. Expertise is not what is being deployed; adjacency is.
This matters because it means the mechanism cannot be concentrated into a reviewer. The whole value lies in the correction coming from an unpredictable direction, and any attempt to designate who does it removes exactly the property that makes it work.
The cost of a late catch
The same error found two weeks later costs the work done on top of it, the rework, and a conversation about why nobody noticed.
Multiply by however often it happens.
What the person outside gets instead
Their assumptions go uncorrected until they produce something. The first correction arrives at review, which is the expensive point.
Nobody is at fault for this; the correction simply has nowhere to happen.
The compounding effect
An uncorrected assumption does not sit still. It becomes the basis for the next decision, which becomes the basis for the one after that.
By the time it surfaces, unpicking it means unpicking everything built on it, which is why the same error costs an order of magnitude more at a distance. The person outside is not making more mistakes; they are making the same mistakes and carrying them further.
Thinking out loud as a mechanism
The reason correction works is that people in rooms narrate. They say what they are about to do before doing it, to nobody in particular.
Nobody narrates to an empty room, so the material the mechanism feeds on is absent.
What can be built instead
A habit of stating assumptions in writing before acting on them, in a place where somebody might read them. It is slower and it catches less, and it is the closest available substitute.
The phrase that does the work is "I am assuming X — shout if not".
Why that phrasing matters
It gives the reader permission to object cheaply. Most people will not volunteer a correction to a finished-looking statement, because objecting to finished work is socially expensive.
Marking something as an assumption lowers that cost to almost nothing, which is the property the room had for free.
Who should do the narrating
Everybody, but the manager first and most visibly. A team adopts this from example and not from instruction, and the person who can make it ordinary is the one whose assumptions it is most awkward to correct.
If you want it to happen, be corrected in public and thank the person.
What this will not achieve
Parity. A written assumption is read by whoever happens to look, at whatever time they look, and the correction arrives in hours rather than in seconds. The compounding is slowed rather than stopped.
It also only covers assumptions somebody thought to write, which is a small fraction of the assumptions anybody actually holds. The room catches the ones nobody knew they had, and nothing proposed here reaches those at all. The realistic claim is that the worst category — a wrong belief that quietly underpins a fortnight of work — becomes less common, and that is worth the minute a day it costs.