Pax Grammatica

On the layer of a peace that outlives the power holding it, and what this one’s is made of

#

The Western Roman Empire ended in 476, or in 480, or gradually and at no particular moment, depending on which historian is arguing. Whichever date you take, Cassiodorus was born after it. He grew up in Calabria in a family that had served the Roman state for four generations, entered public life in his twenties, and spent thirty years drafting the official correspondence of a kingdom in Italy ruled by a Gothic king named Theodoric.

The letters are astonishing and they are the point. Cassiodorus wrote in elaborate, ornamented, self-consciously classical Latin, in the voice of Theodoric and then of Theodoric’s successors, on matters of drainage, grain shipment, appointments, the price of things, a complaint about somebody diverting a watercourse. He collected four hundred and sixty-eight of them and published them as the Variae in 537. The chancellery style is intact. The forms are intact. The rhetorical conventions of a state that no longer existed are being executed at a very high standard on behalf of a barbarian court.

Then Belisarius took Ravenna, and Cassiodorus went east, and eventually came home and founded a monastery on the family estate at Squillace. He named it Vivarium, after the fishponds. He wrote instructions for the monks about how to copy manuscripts and why it mattered, and he made copying a form of devotion rather than a chore, and a great deal of what survives of Latin antiquity survives because of decisions made in that scriptorium.

In his ninety-third year he wrote a book about spelling.

The monks were making mistakes. Their Latin was not what his had been, they had not been trained the way he had been trained, and a copyist who misspells does not merely produce an error, he propagates it into every copy made from his copy. So the last work of the last man who could write like a Roman official was a compilation from the old grammarians about how to spell.

Two layers, and we always name the wrong one

#

Every peace in the pax tradition has been named for its substrate. Romana for the legions and the roads, Britannica for the navy and the gold standard, Americana for the carriers and the dollar, Silica for the fabs. Ask who holds the peace and the answer names a form of power that has to be paid for continuously.

Underneath every one of them there was a second layer that nobody named, which is whatever you had to speak, write, or file in order for the order to process you.

Rome’s medium was Latin. Not the Latin anyone grew up speaking, but the administrative register: the forms, the formulae, the conventions Cassiodorus was executing for a Gothic king. Britain’s was English plus double-entry bookkeeping plus a set of contract forms and a body of commercial law. Persia’s was Aramaic, standardized as a chancellery language across an empire that did not speak it, and which went on functioning as the administrative medium of the Near East under successor regimes that had nothing to do with Persia. The Chinese script did the same across Korea, Japan and Vietnam, for states that were not Chinese and did not want to be.

In every case the substrate and the medium are separable, and in every case we name the peace after the substrate, which turns out to be the layer that goes first.

Why the medium is the durable half

#

The two layers fail differently, and the difference is not about importance.

A substrate runs on expenditure. Legions have to be paid every year, forever, and the year they are not paid they stop being legions. A navy is a recurring cost. A currency’s credibility is a thing maintained by continuous institutional effort. Every substrate is a bill that arrives again next quarter, and the peace it holds lasts exactly as long as somebody can keep paying.

A medium runs on equilibrium. Once everyone files in a particular form, each person files in that form because everyone else does, and no annual appropriation sustains it. It costs nothing to keep. It costs enormously to leave, and it cannot be left by one person or one office or even one kingdom, because the whole value of a medium is that other people are using it.

A substrate has to be funded and a medium only has to be shared, which is why the money runs out long before the habit does.

So the legions were gone from the west by the fifth century and Latin ran European administration, law, medicine, theology and natural philosophy for the next thousand years. Newton published the Principia in Latin in 1687, twelve hundred years after the empire that made it official had stopped existing. Nobody was enforcing that. There was nothing left to enforce it. It persisted because switching would have required everyone who mattered to switch on the same morning.

Media do die, and the exception proves the mechanism rather than denying it. Latin finally lost European administration, and it took printing, the vernacular Bible, the Reformation and the nation-state arriving close enough together to move everybody at once. No single actor did it and no decision was taken. It required a competing medium with its own sponsors, its own economics and its own reason for existing, appearing across enough of the map at the same time that the cost of leaving fell below the cost of staying.

A medium is not overthrown. It is outnumbered, and only by another medium.

Cassiodorus is the seam made visible. He is a Roman official with no Rome, executing the medium at full quality on behalf of the people who replaced the substrate, and then spending his last decades making sure the copying would continue after him. He was not preserving an empire. He was maintaining a format.

What this one’s medium is

#

The medium of the current order is not English, though English is doing well out of it. It is not a language at all, and that may be the first time.

What the systems require is a shape. A thing must be in a form the apparatus can parse, index, embed, rank, and act on. That is a real requirement with real content, and it applies to a clinical observation, a grant application, a résumé, a legal argument, a photograph, a complaint, a life. Put it in the shape and it enters the machinery. Leave it in the shape a person would naturally give it and it sits outside, not refused, simply unprocessed.

The requirement is easiest to see where somebody is failing it. A clinician’s observation counts when it becomes a code, a structured field, a value in a schema that a downstream system can aggregate, and the part of the observation that will not go into a field does not become part of the record in any sense that matters. A person’s working life counts when it becomes a résumé that parses, which is a document written for a reader that is not a person and increasingly is not read by one. A piece of research counts when it becomes an abstract in the form the indexers expect. In each case the underlying thing is unchanged and its standing has entirely changed, and the change was accomplished by a formatting requirement that nobody experiences as a policy.

What falls out is always the same class of thing: what somebody knows that has not yet been given a shape. The judgment held in a doorway, the reason a technique works, the pattern across four hundred patients that has never been put in a sentence. Not suppressed. Not disputed. Just not in the format.

A medium does not decide what is true. It decides what is enterable, and over a long enough period those stop being distinguishable.

This is why the translation story is less comforting than it sounds. Everyone keeps their tongue and the machine bridges, so the surface stays as plural as it ever was. But the surface was never the medium. Cassiodorus’s Goths kept Gothic; what they adopted was the form of an official letter. A peace can leave every language in place and still standardize completely at the layer where things have to be shaped in order to count.

A medium is not what you speak. It is what you have to become in order to be processed, and this one is being installed in a period where nobody is arguing about language at all.

We wonder whether the shape is even describable from inside a period that is still learning it, or whether a medium can only be seen by somebody like Cassiodorus, standing at the end of one, holding the last correct version of a form nobody else can produce.

The layer that will still be here

#

If the pattern holds, and it has held for every case in the tradition, the substrate of this peace will go and the medium will stay.

The compute concentration will move or dissolve. The specific companies will not be the ones holding it in forty years; nothing in the history of any substrate suggests otherwise. The geography will shift, the chokepoints will move from chips to power to something nobody is watching, and eventually the whole enforcement layer that the phrase pax silica names will be a thing historians date.

The requirement to be legible in a particular shape will outlast all of it. It will outlast because it is an equilibrium rather than an expenditure, because it will be embedded in every record system and every form and every process that anyone builds on top of it, and because by then two generations will have been formed inside it and will not experience it as a requirement at all. They will experience it as what writing is, the way a medieval notary experienced the formulae as what a document is.

None of that is where the arguing is. The arguments are all about the substrate: who owns the compute, which country, which company, what the export controls should be. Those arguments matter and they are about the layer that is going to be replaced.

The monks at Vivarium were not classicists. They were men who had been given a job and a set of rules for doing it correctly, and their spelling was drifting because the training that had produced Cassiodorus was no longer available to anybody. He wrote them a book because that was the only intervention still open to him.

It worked, more or less, for a very long time. The book was still being read and copied in the eleventh century, and in the fourteenth. Alcuin knew it. Whatever the old man in Calabria thought he was doing about a local problem with his copyists, what he actually did was help hold a format steady across the six centuries when nothing else was holding anything, and the format outlasted the empire that produced it by longer than the empire had lasted.

He would not have called that a peace. He was worried about spelling.

This essay is the fourth of the five pax essays, which run from Pax Mnemonica to Pax Interlingua, and it supplies the axis the earlier ones did not have: every hegemonic peace has a substrate and a medium, and the medium is the durable layer. The mechanism by which translation at scale produces conformity beneath a plural surface belongs to The Selector and the Composer and is deliberately left there. It stands in conversation with The Monoculture, The Curation Economy, and The Standard with Ears.