"Don't make the specific unnecessarily vague."

Reached out to client to check in about proposal. That sentence is in a system somewhere right now, hundreds of times over, and it has been the example I use for this rule for as long as I have had the rule. It is not a lie, it is not lazy, and the person who wrote it did a real thing that day and recorded it. Every word of it is true.

Three weeks later somebody asks what happened with that proposal.

Reached out how? Email, text, voicemail, or standing in the lobby. Those are four different events with four different expectations attached, and the next move depends entirely on which one it was. A voicemail that went unreturned means one thing. An email that got a reply saying they were taking it to their board means something else, and if you call them now you look like you were not listening.

Which client? The company is the client. The company did not take a call. A person did, and there are four people on that account, and one of them signs and three of them do not. If you reached the coordinator, the proposal has not been seen. If you reached the one who signs, it has.

And checked in about what? Asked whether they wanted to proceed. Asked whether they had questions. Confirmed they got it. Asked about timing. Those produce four different states of the world, and the difference between confirmed receipt and asked whether they want to proceed is the difference between a deal that has not started and a deal waiting on an answer.

One sentence, three unknowns, and the writer knew all three when they wrote it. The information existed. It was in their head, it took no research to produce, and it cost maybe eleven extra words to record. It was not withheld. It evaporated, in the ordinary way, because at the moment of writing the whole thing was obvious and the note was a memory aid for a person who already knew.

Every note in that system is written by somebody who already knows and read by somebody who does not, including the writer four weeks later, who is by then a different person with respect to this particular fact.

What makes this worse than a single gap is that the notes accumulate, and they do not accumulate neutrally.

Put four of them in a row. Reached out to client to check in about proposal. Followed up. Left message. Client considering options. Read as a history, that is a clear narrative: a responsive account manager working a deal that is progressing normally toward a decision. Somebody reading it decides not to escalate, because the account looks handled.

Now suppose what actually happened is that the first contact was an email to a coordinator who never forwarded it, the follow-up was a second email to the same coordinator, the message was left on a general line, and client considering options was that coordinator saying she would pass it along. Every note is still true. The history is fiction. Nobody wrote a false word and the record as a whole says something that did not happen, because four sentences each missing the same variable produce a pattern where no pattern exists.

That is what makes a vague record different in kind from a missing one. A missing record is an absence and gets treated as one. A vague record is read as a presence, and a reader applying ordinary cooperative assumptions fills the gaps with the most probable specifics, which are the specifics they would have supplied themselves. The reader is not being careless. They are doing exactly what interpretation requires, using the only material available.

And there is nothing in the record that can correct them, because the correction would have to come from the variable that was never written down. This is the failure mode that makes vague records worse than useless rather than merely less useful, and it is invisible from inside, because every individual entry survives any audit you can run against it.

Stated as a rule, this one is about a gap rather than a quantity.

The variables are what the speaker knows and what the listener needs. The rule governs the space between them. It is not a rule about detail in the abstract; a sentence can be enormously detailed and still leave the question the reader has.

The quantity minimized is the number of questions a reader has to ask that the writer could have answered at no cost. Two conditions have to hold together for the rule to bite. The writer had the answer. The reader needs it. Where either fails, the rule does not apply and enforcing it produces bloat.

The failure signal is the follow-up question, and specifically the follow-up question that goes to the author rather than to the record. A system where every record generates a conversation is a system whose records are memory aids for their authors rather than information for anyone else. That is measurable, in a crude way: the ratio of times a record is consulted to the number of times somebody consults it and then asks a person anyway.

The domain of validity is asymmetric knowledge. The writer knows and the reader does not. Where both parties know equally, the rule idles, because nothing needs saying. Where the writer does not know either, the honest output is I do not know, which is specific and frequently missing.

The word carrying the load is unnecessarily, and unlike the load-bearing words in the last two chapters, this one comes with its own brake. Necessity is defined by the reader's next action. Detail that changes what somebody does is necessary. Detail that does not is padding, and this rule does not ask for it.

Paul Grice worked out the underlying structure in the 1960s and 1970s, and he did it while trying to answer a different question: how a listener gets more out of a sentence than the sentence literally says.

His answer was that conversation runs on an assumption of cooperation, and that the assumption carries specific commitments. Make your contribution as informative as is required for the current purpose, and not more informative than is required. Make it true. Make it relevant. And be perspicuous: avoid obscurity, avoid ambiguity, be brief, be orderly.

The mechanism underneath is what makes this useful rather than decorative. Because the listener assumes the speaker is following those commitments, the listener draws conclusions from what was not said. Somebody who says reached out to the client and stops is understood, by a cooperative listener, to have said everything that mattered. The vagueness is not read as vagueness. It is read as a signal that the specifics were not important, which is a claim the writer never intended to make and which is frequently false.

That is the part my rule was missing and Grice supplies. Leaving detail out is not neutral. It transmits a message of its own, and the message is this did not matter.

The two maxims that govern here pull against each other on purpose, and the tension is the design. Say as much as is required. Say no more than is required. Neither one alone is usable. Together they define an interval, and the interval is set by the current purpose of the exchange, which is another way of saying that the right level of detail is a function of what the reader is going to do.

George Orwell got to the same territory thirty years earlier and from the political end. His essay on the English language is remembered for its six rules, which are good, and for a diagnosis that is better. He called the mixture of vagueness and sheer incompetence the most marked characteristic of modern English prose. And he made the argument that vagueness in political writing is not a failure of skill but a function: political language, he wrote, is designed to make lies sound truthful and murder respectable, and to give an appearance of solidity to pure wind.

Orwell's sharpest observation for my purposes is the smallest one. A speaker who uses that kind of phraseology, he wrote, has gone some distance toward turning himself into a machine. The point is about the writer rather than the reader. Reaching for the ready-made vague phrase is what happens when the sentence gets produced without the thought, and the vagueness is evidence about the process rather than only about the product. Reached out to check in is that phrase. Nobody decided to write it. It arrived.

The oldest version of the fix is Kipling's, and it is the one that actually gets used, because it fits on an index card. Six question words: what, why, when, how, where, and who. Applied to reached out to client to check in about proposal, it finds three gaps in about four seconds, which is roughly the amount of time the note took to write in the first place.

The evidence here is weaker than for anything else in the creed, and I would rather lead with that than bury it.

Grice's maxims are a theory of how interpretation works, not a finding about outcomes. They have generated an enormous experimental literature about how listeners compute meaning, and almost none of it addresses whether more specific records produce better organizational results. The theory is well supported as an account of inference. It is not evidence for my rule.

Orwell's essay is an argument and a set of prescriptions, and his own rules have been picked over for eighty years, including the observation that the essay violates several of them. That does not make the diagnosis wrong. It does mean the essay is rhetoric about language rather than research on it.

What I could not find is the study I actually want, which would look like this: take a class of records that get written and later read by somebody else, vary how specifically they are written, hold everything else constant, and measure downstream rework. There is adjacent work in clinical documentation and in software defect reports suggesting that incomplete records generate follow-up work, which is intuitive to the point of not needing research, and which does not isolate specificity from a dozen other variables.

So the honest statement is that I believe this rule on the strength of a mechanism and a large pile of personal observation, and the mechanism is Grice's rather than mine. A cooperative reader infers that unstated detail was unimportant. That inference is systematic, not occasional. Where the writer knew something the reader needs, the omission actively misinforms rather than merely failing to inform. That argument I will defend. The claim that organizations which write more specifically perform measurably better is one I hold and cannot support, and Chapter 19 restates it accordingly.

The design problem where this stopped being a style preference for me was a free-text field.

Service systems have one. Notes, Comments, Details, some box where a person types what happened. It is the most-written field in the system and the least usable, and the reason is structural rather than cultural. A free-text box asks what happened, which is a question with no shape, so it gets answered in the shape the writer is already holding, which is the shape of their own memory of the last ten minutes.

The first instinct is to demand better notes. That fails everywhere it is tried, and Orwell's machine-phrase explains why: the vague phrase is what gets produced when the writing happens without a separate act of thought, and exhortation does not add the act of thought. The writer is not withholding. The writer is finishing a task.

What works is to stop asking an unshaped question. Split the box into the questions a reader will actually have, and the sentence writes itself. Who did you contact, chosen from the people on the account. How, chosen from four options. What did you ask, chosen from a short list with a free-text option for the rest. What did they say. What happens next and when.

That takes longer to describe than to fill in, because every one of those is a click except the last, and the output is a record that answers the three-weeks-later question without a phone call. The specificity did not come from better writing. It came from asking.

There is a trap on the other side of that design and I have fallen into it. Every field added is a field somebody has to fill, and a form long enough to capture everything gets filled with whatever passes validation. The field that asks what did they say gets ok. At that point the system has traded honest vagueness for dishonest specificity, which is worse, because ok looks like data and reached out to check in at least looks like what it is. The rule holding that an efficient system minimizes the amount of user input while maximizing the amount of automated output holds the other end of this.

The general shape, which I did not see for a long time: specificity is not primarily a property of what people write. It is a property of what they are asked.

The boundary is where the writer does not know, and it is not as narrow as it sounds.

A great deal of vagueness in records is not the specific made vague. It is genuine uncertainty rendered in the only honest form available. The customer seemed unhappy is vague, and it may also be the exact truth of what was perceptible, and forcing it to resolve into a category produces a false record that will be treated as a real one. My rule says do not make the specific unnecessarily vague. It says nothing against the genuinely vague, and I have watched people apply it as though it did, which produces a system full of confident classifications of things nobody actually knew.

The second boundary is cost. The rule presumes the detail is free to the writer, and mostly it is, because they already have it. Where the detail requires work to produce, the calculation changes and the rule stops being automatic. Recording which of four people you spoke to is free. Recording the exact minute is not free and, absent a reason, buys nothing.

The third is the reader's actual need, which is the one I get wrong most often in the direction of more. Detail nobody uses is not rigor. It is a cost paid at every write for a benefit collected at no read, and it competes for the writer's attention with the detail that does matter.

The reader's need is also not knowable in advance by the writer, which is the part of this rule I have never had a good answer for. The person writing the note has no idea who will read it or what they will be trying to decide. The honest version cannot be include what the reader needs, because the writer cannot see the reader. It can only be include what you know that a reasonable person might need, and reasonable is doing the same unaccountable work that unnecessarily is.

What narrows it in practice is that most records get read for a small number of recurring questions, and those are discoverable. Not by asking people what they want recorded, which produces a wish list. By looking at what the follow-up phone calls are about. Every call to an author about a record they wrote is a specification for a field that should have existed, written by the person who needed it, at the moment they needed it. That is a free instrument sitting in every organization I have worked in, and I have almost never seen anyone run it deliberately.

Three things this rule cannot see.

It cannot tell specificity from precision. Those are different, and confusing them is the most common way to obey the letter of don't make the specific unnecessarily vague and break it in spirit. Contacted the client at 2:47 p.m. is precise and answers none of the three questions that matter. Precision is about the resolution of a measurement. Specificity is about whether the reader's question is answered. A record can be precise to the minute and vague about everything that governs the next action.

It collides with defining every variable to its deepest identifiable condition, and the collision is real rather than verbal. That rule pushes toward decomposition without a natural stopping point; this one is bounded by the reader's next action. Where they conflict, this one should win, because it carries the brake and the other does not. I did not always think that. Putting the two in the same room is Chapter 16's job, and the resolution I am proposing here is the one it has to test.

And it says nothing about whether the specific thing is true. A gloriously specific record of a conversation that did not happen the way it is written is worse than a vague one, because specificity buys credibility. Everything in this chapter assumes good faith and accurate memory, and the second of those is a stronger assumption than it looks at three weeks' distance.

Here is the case against, and it is not the one I expected when I started reading for this chapter.

Eric Eisenberg published an argument in 1984 that organizational communication research had adopted clarity as an ideology rather than as a finding. His claim was that the overemphasis on clarity and openness is both non-normative and not a sensible standard, and that ambiguity in organizations is frequently strategic, competent, and correct.

He is careful about what he is not saying, which makes the argument harder to dismiss. He is not advocating a retreat from clarity. He is advocating a shift away from ideological adherence to it, toward treating clarity as contingent. And he lands the point with a sentence that goes straight through the rule: clarity is only a measure of communicative competence if the individual has as his or her goal to be clear.

Read that against don't make the specific unnecessarily vague and my rule turns out to contain a hidden premise. It assumes the writer's purpose is to inform. Eisenberg's whole argument is that purposes are multiple and often in conflict, and that a communicator navigating several at once may be doing their job well by being vague.

He gives the functions, and each one is recognizable. Ambiguity promotes what he calls unified diversity: a mission statement or a core value stated ambiguously lets different groups apply different interpretations to the same words while believing they agree, and that shared belief is doing real organizational work. Ambiguity facilitates change, because an organization that has not committed to a precise formulation can move without repudiating anything. It preserves position, since ambiguous communication amplifies existing impressions and a credible source gains by not being pinned. And it enables deniability, letting a source both reveal and conceal, express and protect.

Take the third and fourth seriously for a moment, because it is tempting to file them under bad faith and move on. Deniability is not always cowardice. A manager who has been told something in confidence, who must act on it, and who cannot say why, is being vague on purpose and honorably. A negotiator who states a position precisely has given away the range. A lawyer drafting a clause that must survive circumstances nobody has imagined uses general language deliberately, and the generality is the craft rather than a failure of it.

The first function is the one I cannot dismiss as an edge case. Every organization I have worked in runs on shared terms that would not survive definition. Quality. Ownership. The right way to do it. Those words let a hundred people who disagree about specifics act as though they agree about principles, and the acting-as-though is not a pathology. It is how the hundred people manage to move at all. Define them precisely and you have not clarified anything. You have forced a fight, publicly, between factions that were cooperating on Tuesday.

Which would make my rule a technician's rule, true inside a service record, generalized carelessly into a creed by somebody who builds systems for a living and mistakes the properties of a database field for the properties of human organizations.

I think Eisenberg's strategic ambiguity wins about half of that, and the half he wins tells me which half I was actually claiming.

The distinction that survives is between a description of an event and a statement of a value. Reached out to client to check in about proposal is a description of a particular thing that happened to a particular person on a particular day, and the writer knew all of it. There is no strategy in the vagueness. There is no constituency being held together. Nobody gains deniability. It is a record that will be read by somebody who needs it and it fails them, and every function Eisenberg names is absent.

A mission statement is a different object doing a different job. So is a value. So is a negotiating position, a diplomatic communiqué, and a contract clause built to survive the unforeseeable. In all of those, multiple readings are the point, and the ambiguity is load-bearing. Eisenberg is right about them and I have no business legislating there.

So don't make the specific unnecessarily vague narrows to the class of statement it was always about, and gains a clause it never had. It governs records of fact, where the writer knows and the reader needs and no second reading is wanted. That is most of what gets written inside a working organization by volume and almost none of what gets framed on a wall.

One concession beyond that, because Eisenberg's fourth function reaches into my domain too. Even inside a record of fact, vagueness is sometimes protective and sometimes appropriately so. The conversation did not go well may be the right entry when the specific version would be a permanent quotation of a frustrated customer in a system his account manager can read. The rule as I stated it has no room for that, and it should.

Where he does not win: the cases he describes are deliberate. Somebody chose the ambiguity, for a purpose, knowing what the precise version would have said. That is a completely different act from the vague note that arrived because the writer was finishing a task and the specifics were obvious to them at the time. Orwell's machine-phrase and Eisenberg's strategic ambiguity look identical on the page and share nothing else. One is a decision. The other is the absence of one.

What survives:

The original applied to everything and therefore claimed authority over mission statements, negotiations, and contracts, where ambiguity is frequently the competent choice. The revision gives that ground up and keeps the working majority: the note, the record, the handoff, the ticket, the message that somebody will read later and act on.

The practical form is the one from the middle of this chapter, and it is not an instruction to writers. Specificity is mostly a property of what people are asked rather than what they choose to write, so the lever is the question and not the exhortation.

Two questions go forward. The collision with defining every variable to its deepest identifiable condition is now stated in both directions and resolved only by my assertion that this rule's brake should win, which Chapter 16 tests. And I have no evidence, as opposed to mechanism and observation, that more specific records produce better outcomes, which means Chapter 19 restates this one as a practice I hold rather than a finding I can defend.

One absence worth recording rather than papering over. The outline for this chapter called for a second worked example from my own published writing, reading a public monologue two ways to show how much a single vague construction can carry. I could not retrieve the piece to work from, and writing what I think I argued in an essay I could not reread would be exactly the kind of confident vagueness this chapter is against. The example is missing on purpose.