The temperament of a bespoke algebra

We chose the ingredients.
The character chose itself.

We often come to know what we are made of by watching what happens when we meet what we did not choose. A child tends to show their temperament long before they can name it, or anyone can name it for them, in the way they meet a loud room, a slammed door, a broken promise. We learn a horse's temperament in the flex of our own fingers, the pressure of our knees, the way we sit and lean and bend; we learn a friend's in the hour the plans fall through at the last minute. Little about a true character tends to announce itself beforehand; what is real in a character often arises spontaneously, and without coercion. It tends to surface through our behavior, and our sense of things tends toward the trustworthy once the behaviors repeat — once the same disposition shows up again and again, we start understanding we are perceiving one thing and not many.

This is a habit of more than living things. An algebra, too, seems to have a character, and once you have the eye for it a serious one often does — and, stranger, its character tends to be the very set of virtues that let it take hold of the problem it was raised to meet. Clifford's algebra keeps a scrupulous memory for orientation: a turn in it always leaves a mark, a full revolution comes back changed in sign and only a second brings it home, and that refusal to forget a rotation is much of the reason it can carry the spin of an electron. Grassmann's exterior algebra abhors repetition — anything joined to itself falls to nothing — and that one severity is what lets it measure oriented area and volume, and vanish, correctly, on everything degenerate. The Batalin–Vilkovisky algebra is exhaustive and a little obsessive, pairing every field with a shadow and bending the whole construction to a single master equation, and that thoroughness is much of what a gauge theory too tangled for gentler methods asks of it. The Virasoro algebra carries its anomaly in the open, the central charge it will not hide, and that refusal to pretend it is classical is what fits it to the symmetries of a string. The operator algebras of quantum field theory are local to the bone, one algebra to each region of spacetime, faithful to causality past most temptation to shortcut it, and locality held that strictly is much of what a quantum field demands. Each of these is standard, inherited, taught in its turn. Each also behaves like a thing with a disposition, and in each the disposition tends to be the shape of the fit between the algebra and the problem it was made to meet.

Our own algebra belongs to that company, and its character we came to know the way you come to know any character at all, from the outside. It made no announcement. The proof assistant set it in front of situations it never asked for — a product to be closed, a boundary to be crossed, a diagonal case that looked as though it would break something — and none of them yielded to anything small. Each demanded its own long construction: whole families of lemmas raised to carry a single sign through a Koszul pairing, months given to one correction family that would not close, hundreds of small results stacked before a single situation would come to rest. Nothing was easy, and nothing was quick; the work ran to eighteen months and better than half a million lines of proof. What stayed constant across all that labor was a disposition — the same edge condition surfacing at the unit again and again, the same refusals falling in the same places, the same insistence on an honest pairing however far the proof had to reach to honor it. Somewhere in that long repetition a pile of lemmas became a temperament, and we found we could say what the object was like the way you say what a person is like, from having watched it, at length and under strain, meet a pressure it did not arrange.

Its dispositions seem legible now, and each reads as a way of conducting itself. It is loyal: nothing in it resolves alone, a residual holding its value until its own partner arrives, balance found always inside the package and never in isolation. It is temperate: it never wrestles the dangerous term to the ground, it is built instead so the dangerous term can never come near enough to matter, the wrong thing kept out by the sheer length of what would have to reach it. It is scrupulous about edges: it will not let a boundary be smoothed into the interior, and it makes the lone term at the unit answer on its own terms. It is honest: a quotient in its hands is spent only on a real physical sameness, because to declare two things equal for the convenience of making a problem disappear would be a lie told in a second language. Fidelity, restraint, care at the edges, refusal to fudge. A short list, and a recognizable one.

A temperament, in a person, is the thing we usually call contingent — it could have come out otherwise, another childhood would have grown another character. These dispositions arrived under compulsion. The kernel had no room to be gracious; the unit truly is an edge, the support lengths truly are what they are, and once the construction was in the machine its behavior was settled past appeal. So the honest question is whether this is a character at all, or a necessity wearing one. The answer lives in the seam between the two, and the seam is the whole of our interest in it. We chose the body — the generators, the correction, the quotient, the reading that makes the surviving words physical — and that choosing was free; everything the body then had to become was compelled. A character is something like this shape: free in its origin, fixed in its consequence. No one chooses a temperament, and a temperament is still the fixed point of a particular biology held to consistency across a life. This one is the fixed point of a particular construction held to consistency by a proof.

That is why the words bespoke and temperament want to be said in one breath. When we describe people we move between two registers without noticing: the register of shared traits, the warmth and caution and appetite everyone carries in some measure, and the register of the single unrepeatable person, this one and no other, whose exact arrangement of those traits has never occurred before and will not occur again. The virtues themselves — the loyalty, the temperance, the care at the edges, the honesty — belong to the first register: they are what any algebra built to certify would have to be, and they would appear, in some form, across the whole family. What is this one's own is the second register, the exact hand it carries them in — which residual waits for which partner, how long a carrier must run to fall out of reach, where precisely the boundary lands and how it is paid. And the question the work leaves open, the one we hold without pretending to have closed it, is which register finally governs: whether that hand, too, is bespoke to one instrument and one boundary, or whether it is the first showing of a species, a way of certifying to be recognized again the next time a measured boundary is asked to yield a physical observable. We do not know. The knowing waits on the structure turning up somewhere its author never went.

Our own algebra keeps the pattern. There is a fittingness in its list of virtues that we did not design and cannot take credit for. Set down what a witness has to be — faithful to what it saw, restrained in what it claims, exact about where a thing ends, unwilling to call two things the same for its own convenience — and the list is the same list. The temperament the kernel carved into being seems to be the temperament a certifier must have. An object whose whole office is to decide when structure has earned the name of a physical observable would be worthless if it were loose with its pairings, or careless at its edges, or willing to quotient away whatever it found awkward. It is none of those. And it made those virtues good at the one place they were most tempting to spend: a looser construction, meeting the lone term left standing at the boundary, would have quietly treated it as interior or hidden it in an equivalence and called the work done; this one carried the term into the open and held it there until the surrounding structure determined it made no admissible contribution, the quotient left unspent and the edge kept honest. Its character is fit to its work, and the fit fell out of holding the construction to consistency, which is the same thing as holding it to honesty.

One resonance remains, and we set it down carefully, because it is either an accident of how we are speaking or the deepest thing in the whole enterprise. The definition this algebra exists to certify is a definition of health, and health, in our hands, is a bounded thing that keeps being itself — without collapsing, exploding, or losing its internal relation. Turn now and look at the algebra's own temperament. It keeps its consistency, which is a way of keeping being itself. It accounts for its boundary and keeps the edge from dissolving. It holds every coefficient back from running away. It lets nothing resolve outside its pairing, so its internal relations are never lost. The certifier, asked only to measure a certain shape, turns out to have that shape. A thing built to recognize health kept its own health to do it. We notice this without leaning our weight on it. It may be a trick of the mirror we are holding. It may also be the reason the whole thing coheres.

None of this changes what the algebra is or what it can do. It is the same construction it was before anyone noticed it had a disposition. A maker learns something real about a made thing on the day he stops reading his intentions into it and starts watching what it does under a load he did not design; and what we learned, watching, is that we had built something with a character of its own — loyal, temperate, scrupulous, honest — and that the character was much of the point, because the character is what lets it stand as a witness. We found those virtues already in it, the way we come to know anyone at all: by watching how it behaves when it meets what it did not choose.

Three companion notes: What is genuinely new situates this object against the fields nearest to it; A boundary-observable certification algebra describes what kind of object it is and how it was built; and The making of an observable asks whether its character is bespoke to one instrument or the shared discipline of observation.