What Is a Detail

The Immutable Past · What Is

What Is a Detail

A detail is a small part of a bigger thing, and small parts matter less — except that the proverb says they matter most. Derived from the compass at the base of the foundation, the word turns out to be reporting on an axis that has no size in it, which is why the two proverbs were not rivals, why the question “how much detail is enough” has no answer in the form it is asked, and why looking harder will not supply one.

  • AuthorJohn Rector
  • Published
  • FormEssay · 18 min
  • Primitive doing the workS3 · read on the imaginary axis alone
  • RegisterDoctrine · three marked constructions
  • FoundationSpine v1.30

The count the ordinary account does not produce

2

Questions folded into the single word detail. Ordinary usage has it answering one, which is why two proverbs about details can each be obviously true and cannot both be.

The compass at the base of the foundation carries two axes, not one scale. One asks whether rank is present and permitted to stand. The other asks whether anything stands out. The second has no size in it — not small, not large, not more or less. A great many English sentences about details answer both questions with one word, and then two people argue about the answer without noticing they were sent to different questions.

Contents
  1. Two proverbs that cannot both be right, and are
  2. Three cases where size cannot do the work
  3. The compass — two axes, two questions
  4. What a detail is, then — the first marked construction
  5. Not a fact about noticing — the second
  6. No exchange rate — the third, and the problem underneath it
  7. What it costs, stated without softening
  8. If the details are what you cannot stop turning over
  9. On an ordinary Tuesday

Somebody hands your work back with a note on it: this is just a detail. You know what they mean and you do not need it explained. A detail is one of the small parts. The thing has big parts and small parts, the small ones can be dealt with later or not at all, and being small is what earns them that treatment.

That account is ordinary, it is old, and it is not stupid. State it properly before taking it apart. Detail came into English through French détail, from tailler, to cut — a cutting-up, the division of a whole into its pieces. The picture inside the word is a thing sliced, and a slice is smaller than what it came from, so a detail is a lesser part by construction. Retail, entail and tally sit beside it in the same family of cutting and counting, and they all carry the same implication of portions taken off something larger.

And then, living in the same language without embarrassment: the devil is in the details. Or the older and more flattering version, God is in the details, which gets attributed to Flaubert, to Aby Warburg and to Mies van der Rohe depending on who is doing the attributing. The proverb says the small parts are where the thing is decided. The definition says the small parts are the lesser ones. Both are in ordinary use, both are believed by the same people, often in the same week, and the collision is rarely felt because the two are seldom said in the same breath.

They cannot both be claims about how much details matter. Either the small parts are the negligible ones or they are the decisive ones. Put them side by side and a minor detail becomes a contradiction or a tautology, depending on which proverb you reached for first. Neither reading is what anybody means.

Three ordinary cases finish off the size account, and they are better up front than built toward.

Case 01 · the misspelled name

A twelve-page agreement, signed by both sides, with one letter wrong in the counterparty’s legal name. By any measure of size the page offers — characters, words, ink, minutes to fix — it is the smallest thing on the document. It is also the thing that decides whether the instrument does what it was made for. Call it a detail and nobody objects. Call it minor and somebody corrects you. The word applied; the size reading came apart from the word the moment the word was tested.

Case 02 · the movement nobody sees

Two watches. Same dial, same hands, same case, same time kept. One has a movement finished by hand on surfaces sealed inside it, which a watchmaker might look at twice in a century and which the owner will not see. Everybody calls the difference detail; the word turns up without effort. Now ask whether that difference is large or small. The question does not have a hard answer, it has no answer — there is nothing it is being measured against. The word applied regardless, which means the word was not waiting on the measurement.

Case 03 · the argument that will not end

Two people look at the same draft, the same build, the same plan. One says the thing under discussion is a detail. The other says it is the whole point. They do not disagree about any fact: both describe the item identically and neither has overlooked anything. A disagreement about size ends with a measurement. This one does not end, and it does not end because measuring is not what it is short of. Something else is under contest and the single word is concealing which.

Three cases, one shape. Where size is measurable, the word comes loose from it. Where size has no measure at all, the word applies anyway. And where two people fight over the word, measuring does not settle the fight. Whatever detail is reporting, size is not it.

The compass — two axes, two questions

The foundation this series writes from opens with a compass, numbered S3, and it is the least quoted thing in the document because it reads like bookkeeping. It is two axes. The whole of its content is that they are two and not one.

The real axis runs between hierarchy and fairness, and what it measures is ordering: is rank present, and is it permitted to stand? The imaginary axis runs between symmetry and significance, and what it measures is distinguishability: does anything stand out?

Those are separate questions — separate not in the loose sense of being about different subjects, but in the strict sense that an answer to one leaves the other open. Settle that something is distinguishable and you know nothing yet about where it ranks. Settle a ranking and you know nothing yet about whether anything in it stands out. The compass is a plane rather than a scale, and a plane is what you need when two readings do not reduce to one.

Figure 01

Four poles, two questions — and which one this piece is standing on

  • Real axis · positive pole Hierarchy Rank present, and permitted to stand. The order is the answer and the order is allowed. Question: is ordering present?
  • Real axis · negative pole Fairness The same question answered the other way. Rank interrogated and refused rather than absent — which is why the two poles are one axis. Question: is ordering present?
  • Imaginary axis · positive pole Symmetry Nothing distinguished. Not a low ranking — no ranking has been asked for. A symmetry is not significant, and that is a statement about distinguishability alone. Question: does anything stand out?
  • Imaginary axis · negative pole Significance A distinction present. Significance requires broken symmetry, and it says nothing about where the broken thing sits in any order. Question: does anything stand out?
The compass as the foundation states it, with the two poles this piece stands on marked. The pole names and the two measuring questions are quoted doctrine. Which axis an English word reports on is not in the foundation and is the construction marked below — that part of the figure is argument, not record.

Two readings of this compass already exist in this series, and both are conjunctions: each takes a state on one axis and reads it through the other. One looks at a broken symmetry and finds a rank inside it. The other looks at a rank and watches it refuse to break. Useful, both of them, and neither is an axis on its own.

What has not been done is the imaginary axis read alone, with no ordering in it anywhere. The reason it has not been done is that few ordinary English words manage to carry distinguishability without carrying rank as well. Special, important, remarkable, outstanding, notable — each one has a ladder built into it. Say a thing is remarkable and you have already put it above the unremarkable things.

Detail is the word that does not. It says a distinction is present in the thing. It says nothing whatever about where that distinction sits in any order, which is precisely why the adjective has to be supplied separately, by hand, on each use: a minor detail, a crucial detail, a damning detail. The adjective is doing the ordering work. The noun was not doing it and had no way to.

Marked construction 01 · new here, not in the foundation

“Detail” is a word on the imaginary axis. It reports distinguishability — that a distinction is present in the thing — and reports nothing about ordering. The size-and-importance vocabulary travelling with it is doing separate work on the other axis, and the two readings are independent.

Provenance. S3 supplies the two axes and what each of them measures. It does not classify English words, and nothing in the foundation says which axis a given word reports on. The assignment is mine. The evidence for it is the three cases above: a word that keeps working where its supposed measure has no value, and comes apart from that measure wherever the measure is available, was not reporting that measure.

Take the assignment seriously for a moment and the two proverbs stop contradicting each other.

Just a detail is a real-axis sentence. It reports a rank: against the other things on the list today, this one is low. The word detail is not what makes the sentence dismissive. The word just is.

The devil is in the details is an imaginary-axis sentence. It reports that the distinctions present in a thing are where the thing is decided — which, on the foundation’s account of what an artifact is, is closer to a definition than to a warning. It makes no claim about rank, and no claim that the distinctions are small.

Orthogonal claims cannot contradict each other. The two proverbs were not in competition; they were answering different questions, and one noun has been carrying both for six centuries.

The word detail was not the dismissive one. The word just was.

Not a fact about noticing

There is a reflex at this point and it should be met head-on, because it is the reflex that would wreck the reading. The reflex says: a detail is what somebody notices. Distinguishability is in the eye. One person catches the misspelling and another does not, so whether it is a detail depends on who is looking.

The foundation holds otherwise, and the derivation is short.

A condition, in the locked sense, is a prerequisite in order for something to happen or exist. That is P6, and it is a definition rather than an argument. P7 adds the part that does the work here: an idea is a named condition, and naming does not create the condition — naming makes the relationship conscious. The prerequisite held before anyone named it and holds if nobody ever does.

Distinguishability, on the compass, is a property a condition has. So take the misspelled name again. Is the distinction between the right spelling and the wrong one a prerequisite for the instrument doing what it was made for? That is settled by what the instrument has to be, and not by who read it. The distinction was load-bearing at the moment of signing, in a room where nobody had seen it. It was named later. It did not become a distinction later.

So the first thing under the word is a property of the condition, and it does not wait on an audience.

There is plainly a second thing people mean, though, and it deserves separating rather than dismissing, because ordinary speech runs both of them under the one noun.

When somebody says I noticed a detail, they are reporting a reading. The foundation’s account of a reading is exact: what a host has is the quotient, R = A/E, an Actual over an Expectation that belongs to that host and to nobody else (P4), and the human experiences the quotient rather than either term (R8). Two hosts with different denominators produce different readings of one situation. That is why one editor catches the name and the other does not — and it is also why neither of them is thereby right about what conditions the contract.

Marked construction 02 · new here, not in the foundation

Ordinary usage runs two senses under the one noun: a distinction that conditions the artifact, and a distinction a host has read. Only the first sits on the compass. The second is a quotient, and it reports on the reader as much as on the thing.

Provenance. The foundation supplies both notions independently — conditions that hold whether or not they are named, and readings that belong to a host. It does not say that this English word straddles them. The split is mine, and it earns its place by predicting which arguments about details are winnable: the ones where the disputed item can be shown to be a prerequisite, and no others.

That prediction is worth stating as a test, because it can fail. If the split holds, an argument about whether something is a detail should end as soon as one party demonstrates that removing the distinction stops the thing from being the thing — and should grind on indefinitely where no such demonstration is available, however carefully both sides look. If arguments of the second kind routinely resolve by one party simply looking harder, the split is wrong, and the reading of the word goes with it.

No exchange rate

The most common request anybody makes of this word is the one the compass has no answer for.

Which details matter enough? Where do I draw the line? How much detail is enough?

Each of those asks for a conversion — for the distinguishability reading to be expressed as a quantity on the ordering axis, so that one number comes out and the line draws itself. The compass carries two orthogonal axes. Orthogonal means there is no projection of one onto the other, and where there is no projection the conversion being asked for does not exist to be found. The question has no answer in the form it is asked, and the person asking it is not being stonewalled.

What is available is two questions, asked separately and held at the same time.

Figure 02

One item, two questions, two answers that do not combine

The item Imaginary axis — does it condition the thing? Real axis — where does it rank this week? What the single word would have hidden
One letter wrong in a signed party name Yes. Remove the distinction and the instrument stops being the instrument. Low, on effort and on visibility. Nothing else on the list is cheaper to fix. “It’s a detail” would have been heard as the ranking, and the prerequisite reading would have gone unstated.
Hand-finishing on a sealed watch movement Yes, for the thing that object is being made to be. No, for a thing being made to keep time. Undecidable without the price, the buyer and the house. A position, and a revisable one. That the two answers depend on two different things, and that neither of them is about how big the finishing is.
A wording change nobody outside the room will notice Unknown until the removal test is run, which takes a minute and is rarely run. Low, on attention and on audience. That the low ranking was doing all the talking while the first question was left unasked.
Three worked items, with each question answered on its own. The items are the author’s illustrations and not reported cases; the point they carry is the structure of the two columns, not the verdicts in them. Note that the second row answers the first question twice, because “the thing” was two different things — which is a separate way the single word goes wrong.

Both questions have answers. Neither answer is the other one. There is no third number that fuses them, and no procedure that produces one. The manager who wants a policy on details and the maker who wants to be told when to stop are asking for the same nonexistent quantity from opposite ends of the same room.

Marked construction 03 · new here, and it inherits an open problem

There is no exchange rate between the two axes. A question of the form “which distinctions matter enough” asks for a projection between orthogonal axes and therefore has no answer in that form.

Provenance. The claim follows from orthogonality, and the foundation does not hold orthogonality as proven. It sits in the open problems in exactly these terms: all four quadrants of the compass populate with real cases, which is evidence and not argument. So this construction is as secure as that and no more — well supported, load-bearing in practice, and underived. It is not being closed here, and the next section says what closing it would take.

The open problem this piece runs on, left open

What is unsettled
Whether the two axes are orthogonal. The foundation asserts it and supports it by observing that all four quadrants populate with real cases. That is evidence. It is not a derivation, and the foundation says so about itself.
What would close it
One of two things, and both are specific enough to be held to. Either a derivation from the complex ground the compass sits on — a demonstration that an ordering question and a distinguishability question cannot be posed as one question without loss. Or, from the other side, one clear forcing case: a situation in which settling the distinguishability reading settles the ordering reading by itself. A single forcing case collapses the plane into a scale and takes this piece down with it.
What this piece does
Neither. It uses the orthogonality, names it as unproven, and states the two ways it could be finished. That is the honest shape available today, and a reader who wants to break the argument now knows where to push.

What it costs, stated without softening

Four costs. The third is the one that bites.

One · looking harder is not the method

If distinguishability belongs to the condition, then concentration is not how you find out what counts. Attention yields a reading, and a reading is a quotient with your own denominator inside it. The way to find out whether a distinction conditions an artifact is to ask what the artifact has to be — which is slower, less flattering, and not improved by staring. Two people of equal diligence will read differently, and the diligence is not what decides who is right.

Two · there is no threshold, so there is nothing to cross

The foundation holds that each condition has an exemplar: a complete and exact form (P10). Complete belongs to the exemplar and not to any list a person could write down — the circumference of conditions is nameable without limit (S1), so enumeration does not terminate. Which means the sentence I have attended to all the details is not modest and not immodest. It is not a sentence about anything. There is no line in the structure to be on the right side of.

Paired with it: each artifact is a miss (R10). Not the poor ones. Each one. The distinctions you did not reach are not a backlog that a better week clears; they are what makes the thing an artifact rather than an exemplar, and there is no version of the work in which they are absent. That is why the evaluative question is fidelity rather than success (R11) — how near the miss came, in which direction, and what gets corrected in the next one. Correction still running is the difference between approximating and settling, and settling is the single failure this vocabulary names as such.

Three · the load-bearing distinction with nobody to speak for it

Because the axes do not convert, a distinction can be a prerequisite for the thing existing at all and still rank at the bottom of any list a reasonable person would build. Those two facts do not constrain each other. This is not a defect in badly-run organisations; it is the ordinary output of running two independent questions and letting one of them do the scheduling.

We will catch it later is a real-axis sentence wearing an imaginary-axis costume. It answers the ranking question, which was a fair question with a fair answer, and leaves the prerequisite question untouched. The delay is not an answer to the question it did not address.

A prerequisite that ranks low has no advocate in the room, and the room will keep producing decisions that look defensible line by line and fail as a whole.

Four · naming it bends you toward it

Smaller, and worth saying. Naming a distinction orients you toward it and prejudices you toward it in one motion — that is R14, each idea biased toward itself, including this one. The detail you named this morning will look more load-bearing to you tomorrow than it does now, for reasons that have nothing to do with the thing. Orientation and prejudice arrive together and there is no arrangement in which you get the first without the second. The practical residue: your own list of what conditions the work is worth re-running from the removal test rather than from memory, because memory has been rewarded for the items already on it.

If the details are what you cannot stop turning over

There is a way of living inside this word that is not a philosophical problem, and nothing above touches it. Some people are held by details — checking, re-checking, the same distinction returning at three in the morning with the same force and no new information in it. That is not a mistaken theory of what a detail is, and a better theory does not loosen it.

What is above can take one thing off, and only one: the idea that there is a correct amount of detail which a sound person would have located by now, and that failing to locate it is a defect in you. There is no such amount in the structure. Not for you, not for anyone, not on a good week. That much is real and worth having.

It will not do more than that. Understanding a mechanism does not stop the mechanism, and there is no schedule on which anybody is supposed to be over it. If the turning-over is running your evenings, the thing that helps is a person trained to help with it, and a metaphysics is not that and should not be offered as though it were.

On an ordinary Tuesday

  1. Split the sentence in two. When somebody says that’s just a detail, ask both questions out loud and separately. Does it condition the thing? Where does it rank today? Most of these arguments end right there, because the two people were answering different questions and agreeing about the facts.
  2. Notice which word did the dismissing. Just is a ranking. A ranking needs ranking reasons — the deadline, the budget, the risk, the person waiting — and those are arguable in the open. Tucked inside the noun, they are not.
  3. Run the removal test. Take the distinction out and see whether the thing is still the thing. It takes about a minute, it answers the first question rather than the second, and it does not depend on how carefully anybody was looking.
  4. Stop asking for the threshold. Nobody is withholding it. Ask instead what you will rank and spend on this week — a real question with a real answer — and let the answer be a position you took rather than a fact you discovered.
  5. Make the next one instead of retrofitting this one. Records are add-only (R2). The distinction you missed is not editable out of the finished thing; the move available is a new artifact carrying it. In this vocabulary amends are artifacts rather than edits, and so is a second draft.
  6. Write the distinctions down before the ranking starts. Once ranking is underway, the ones that condition the thing and rank low have nothing speaking for them, and the list is the only advocate they are going to get.

A detail is not a small thing. It is a distinction that is there — and whether it is there, and where it ranks, are two questions that have been sharing one word for six hundred years. Say them separately and most of the argument disappears. What is left is the real disagreement, which is the one worth having.

Sources

  • The etymology is the received one and is the only external fact in the piece. English detail comes through French détail, a noun from tailler, to cut, with the sense of a cutting-up or a division into parts. It is cited to show how old the slice-of-a-whole picture is, and for nothing else; no step of the argument depends on it.
  • The proverb attribution is deliberately hedged. “God is in the details” has been attributed to Flaubert, to Aby Warburg and to Mies van der Rohe, and the attribution is not settled in any source I would cite. The piece uses the saying and not the attribution, and nothing rests on who said it first.
  • The three opening cases and the three table rows are mine and are offered as observation. The misspelled name, the sealed movement and the argument that will not end are drawn from ordinary experience rather than from survey findings. They are used as counterexamples, which is the one job an observation can do without being a statistic: each only has to be recognisable to defeat the claim that detail is a size word.
  • Deliberately left out — three things. First, any magnitude on the imaginary axis. The compass names the axis and the foundation carries no expression for how distinguishable a condition is, so this piece says a distinction is present and does not say how much. Second, whether fairness is genuinely the antipode of hierarchy, which is the other axis’s own unresolved question; it is named here and not approached, since this piece works the axis next door. Third, whether a machine has surprise, which the foundation records as unresolved and which nothing above touches.

Read next

Citation stamp

Foundation
The Spine v1.30 · 19 September 2026
Primitives
S1, S3, P4, P6, P7, P10
Results
R2, R8, R10, R11, R14
Registers
DOCTRINE for the primitives and results above, with S3 read on the imaginary axis alone and with no ordering reading anywhere in the derivation. CONSEQUENCE for the three marked constructions, each flagged in a box where it occurs. REPORTED for the etymology; the proverb attribution is named as unsettled and carries no step. The three opening cases and the three rows of Figure 02 are the author’s illustrations and are not reported findings.
Open questions touched, left open
Orthogonality of the two axes — used by the third construction, named as unproven where it is used, with both closure conditions stated in the body. Not closed. Whether fairness is the antipode of hierarchy — named in the Sources block and not approached. Whether a machine has surprise — named and not approached.
Ledger
Not invoked. Nothing here says that unseen care is repaid, that overlooked work evens out in the end, or that anybody is owed anything for the distinctions nobody noticed. The care section states the absence of a threshold and stops, which is the whole of what is on offer.

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.

Discover more from John Rector

Subscribe now to keep reading and get access to the full archive.

Continue reading