Asking
The Missing Noun
Pronouns are unambiguous to the person who writes them and empty to somebody reading fifteen hours later with no memory of the day.
three things · six nights
Three questions, six nights, and two of them avoidable.
Questions from one week where the only fault was an unnamed thing
Did they approve it?
Neither they nor it was named; two crossings established which
Is the new version live?
Three deployments that week, and the asker meant the second
Can you look at the failing test?
Four were failing; the asker meant the one they had just broken
Has the client come back?
Two clients were waiting on separate things that week
Should I use the old script or the new one?
Both had been replaced since the asker last looked
Is this ready?
This referred to a message sent the previous week
six questions · ten round trips beyond the first
Six questions, ten round trips beyond the first, and every one of the extras was a noun. In each case the asker knew the name and could have typed it in two seconds. Nothing here is about care or competence; the pronoun is correct in the asker's head and empty in the reader's. One team's own week, logged by the person answering.
"Did they approve it?" is a perfectly clear question when the person asking and the person answering have both spent the morning on the same thing.
A task can look idle simply because nobody sees the part completed in another time zone. employee productivity software makes that work history available and gives the discussion in The Missing Noun a more reliable starting point.
The workflow can also be compared with the approach documented by Linear's product workspace, an established resource in the same operational space.
Sent across a gap, it arrives to somebody who was asleep for the entire context in which it made sense. It contains two pronouns and no information.
Why this is so persistent
Not carelessness, and treating it as carelessness is why it does not improve.
When somebody writes "it", the referent is vividly present to them. The sentence feels complete because it is complete, in the context the writer is standing in. Nothing about writing it signals absence, because nothing is absent from the writer's side.
The gap removes the shared context without removing the feeling of having it. The writer's day and the reader's day are twelve hours apart and the writer's language is calibrated to a colleague sitting next to them.
This is why the error survives everybody knowing about it. It is not a knowledge problem.
The specific words
Six, and they account for most of it.
It, they, this, that, the new one, the other one.
Each of these is fine in a message where the thing was named earlier in the same message, and each is empty in a message where it was not. The test is mechanical: does the name appear anywhere in what you are about to send?
- it
- they
- this
- that
- the new one
- the other one
What counts as naming it
Enough for the reader to find the thing without asking.
A version number, not "the new version". A branch or a file, not "the failing test". A client by name, or if names are not used, by something that distinguishes them from the other one — "the client who asked about the export".
A date and a channel for a conversation, not "what we discussed". Conversations are the worst case, because several happened and nobody remembers which one anybody means by Thursday.
Where a thing has a link, the link is the name. It is faster to paste than to describe and it is unambiguous.
The questions that look fine
A whole class of these reads correctly on a first pass and still cannot be answered.
"Is the new version live?" has a noun in it. The noun is "version", and it is not the name of anything. On a week with three deployments, it narrows nothing.
"Can you look at the failing test?" similarly. The definite article implies there is one, and the reader can see four, and the two facts cannot be reconciled from the message.
The underlying error is the same: a description that is unique in the writer's context and not in the reader's.
The mechanical check
Reading the message as a stranger is the advice everybody gives and very few people follow, because it requires imagining a state of ignorance that is hard to access about your own work.
A more reliable version: before sending, count the pronouns and definite articles that refer to things. For each one, check that the name appears somewhere above it in the same message.
This can be done in ten seconds without imagining anything, which is why it works where the imaginative version does not.
The answer side
A recipient who receives one of these has two options and should take the second.
Answering what they think was meant, which is fast and which is wrong often enough to be dangerous — an answer to the wrong question is worse than no answer, because the asker acts on it.
Or asking which, and saying what the candidates are. "Which version? There were three this week: 2.4.1 on Monday, 2.4.2 on Wednesday and the hotfix last night." That reply costs a night and ends the matter; an unqualified "which one?" costs a night and does not, because the asker now has to work out what the options are.
Whose problem it is
The asker's, in the sense that they can fix it and the reader cannot.
But it is worth saying that a team where this happens constantly does not have a problem with individuals. It has a shared-context problem: things without stable names, conversations that happen in places without addresses, versions that are described rather than numbered.
Where every noun that matters has a name people can type, the missing noun largely stops occurring on its own.
Naming things so they can be named
Much of this is downstream of whether the team's things have names at all.
A deployment with a number can be referred to; a deployment described as "last night's" cannot, once there have been two. A conversation in a place with an address can be pointed at; one that happened in a call cannot be pointed at by anybody.
Teams that give things stable, typeable identifiers find this class of problem shrinks without anybody being told to write more carefully. It is a cheaper intervention than teaching people to proofread, and it is the only one that works on the days when everybody is tired.