The Statement

Neither of those filters sorts for beauty. Fire, war, damp, and reuse decide the first one. The second is decided by people with reasons — status, faith, politics, money, attention. A thing that survives and is needed gets the label. A thing that survives and is needed by nobody sits in storage, and after a while it stops being in storage too.

In every period there was a party whose job was to decide what counted. A temple, a court, a guild, an academy, a dealer with a critic beside him, a movement with a manifesto, a government, a collector with a checkbook, and now a ranking system.

What changes is who they are and where they stand in the chain. Early on, the same person paid for the work and decided it was good, before it existed. Later the deciding moved to an institution that certified after the fact. Later still, the money arrived years ahead of the demand, and the whole business became speculation on the appointment itself. In the last two decades the appointment has begun to arrive before the object.

The mechanism doesn't change. Only its staff.

"Art," as a single category holding painting, sculpture, music, poetry, and dance together, was assembled in the middle of the eighteenth century and given a public building at the end of it. Before that, the grouping did not exist in that form.

Everything older was moved into the category afterward. Altarpieces were made to work, not to be looked at. Grave goods were made to be buried. Trophies were made to be carried through a city. Calling them art is a decision that was taken later, by people with their own reasons, and it is a decision that can be dated.

A place in the canon = talent × the power of the machine that appoints.

Machine power = resources × story × channel.

Resources are whatever the period ran on: money, but also power, ideology, institutional prestige, storage, restoration, staff. Story is the reason a name is worth repeating: the martyr, the recluse, the pioneer, the first one on a new technology. Channel is whatever puts the name in front of people at scale — a treatise, a Salon, an illustrated magazine, a touring show, a recommendation engine.

The second line exists for one reason. The first line alone can't be tested. Left by itself, machine power gets measured by the outcome it is supposed to explain: this was canonized, so the machine must have been strong. Breaking it into three inputs takes it out of that circle, because resources, story, and channel are all visible before the outcome, in documents that exist whether or not the appointment worked.

It is multiplication. A zero on any one of them and the name is gone, whatever the work was.

There is a second finding, and it has nothing to do with quality.

A canon lasts about as long as the machine that appointed it. One system in the ancient world ran for roughly three thousand years, and its canon held for three thousand years. A state that was the sole appointer in its country for seventy years had a canon that faded when the state did. A price-driven canon of the last decade was assessed as largely worthless inside three years.

Strength is not what separates them. Some of the strongest machines were the shortest-lived. The difference is whether the appointment can be handed on. A single appointer takes the canon with him. A network passes it between nodes, and any one node can fail without the canon failing.

So the useful question about anything appointed recently is not how loudly it was appointed. It is how many independent parties are holding it now.

Nothing here is an accusation, and none of it requires anyone to have lied.

Culture cannot keep everything. Storage costs money, restoration costs money, wall space runs out, and attention runs out faster. Something has to be dropped, and dropping is a decision somebody makes. Selection was never optional. The only real questions are who does it and whether you can see them doing it.

What we describe is a relay, not a plot. One party buys cheap. A second writes the essay that makes the purchase look inevitable. A third inherits the estate and spends thirty years building the legend. They did not coordinate, most of them never met, and several of them lost money. These machines have run over their own operators more than once.

We find the thing impressive. Six centuries, rebuilt from scratch every few generations, still working. That is worth understanding before it is worth judging.

We do not claim that art isn't real. The first factor in the formula is talent, and it is not a zero. Work can be extraordinary and appointed at the same time, and one does not cancel the other.

We do not claim that famous artists didn't deserve it. Nothing here reduces to "they weren't good." Any sentence in that shape is a misreading. The claim is narrower: talent is spread fairly evenly and machine power is not, which is why two comparable artists end up in different places.

We do not claim to know who was erased completely. Anyone the machine never touched left no documents, so they are invisible to us by construction. We think the shape of the evidence still says something, because every new appointing machine raises its own wave of rediscovered geniuses, and that is not what you would expect if the canon were simply tracking quality. But the missing denominator is real, and we would rather name it ourselves.

And we cannot prove sufficiency. We can show that a machine ran, name who ran it, and date it. Nobody gets to rerun the century without it.