PhronesisNeither word named a thing
Computer is the easy case, and it is easy because the migration is documented rather than argued. From the early 1600s it meant one who computes, and for three hundred years it was an occupation: the Harvard Computers who classified stellar spectra, the women at Langley whose arithmetic put vehicles in the air, the rooms of people at Bletchley. The word did not stretch to cover machines. The machines took a job that already had a name, and the people who had held it kept the title until they lost the work.
Referee is the sharper case, and its etymology gives the whole game away. It is refer + -ee: from Latin referre, to carry back, with the suffix that marks the one to whom something is done. An employee is one who is employed; a referee is one to whom a matter is referred. First used in the 1500s for disputes carried to a third party, an umpire by the 1660s, a court-appointed decider by the 1680s, sport by 1820, and today also the stranger who reads your paper and says whether it may be published.
“Computer” at least names an activity. “Referee” names only a position — it does not say what the referee does, only where they stand.
That makes the second word the more fluid of the two, which is the opposite of how it feels. A computer must at least compute. A referee need only be the place a question goes. Whatever occupies that position holds the office: a person, a panel, a rulebook, a video system, a coin.
Where unlimited fluidity goes vacuous
Here is the cost, and it is not a worry — it is a result. Putnam (1988) argued that every ordinary open physical system implements every finite automaton: take any object over any seven minutes and you can map its states onto the sequence the automaton calls for. Searle (1992) reached the same place from the other side: whether a system implements a computation depends on how an observer interprets it, so any sufficiently complex object can be described as implementing any computation.
If those hold without qualification, then the wall is a computer, and so is the weather, and so is a bucket of water. The claim “S computes” stops distinguishing S from anything whatever. A word that applies to everything has the informational content of a word that applies to nothing. So the fluidity is real and it is not free, and anyone who enjoys the first half of this note without reading the second has bought a slogan.
The bound is the practice
The second word supplies what the first one hides. There is no referee without a game. The role is not conferred by anything intrinsic to the occupant — not by being human, not by wearing black, not by any property you could find by examining them alone. It is conferred by a practice with rules, and by the fact that questions under those rules get carried to that position. Remove the game and the person in the middle of the field is not a referee who has stopped working; they are not a referee at all.
Run that back onto the first word and the triviality dissolves into something workable. A physical process is a computation relative to an interpretation — someone or something has to take the output as an answer to a question that was asked. The wall implements the automaton only under a mapping nobody uses for anything. Role words are fluid in what can fill them and strict about what must exist around them. The substrate is free; the surround is not.
Which is this site’s own claim about readings, applied one level up. Nothing reads its own state except against a second term — and nothing holds a role except against a practice. There is no proprioception in a void and no refereeing in one either.
Why it matters: a role can be vacant
This is where the fluidity earns its keep rather than just being clever. If “referee” named a kind of person, then asking whether some domain has one would be a question about staffing. Because it names an occupied position, the question becomes sharper and worse: is anyone standing there at all?
Money is a sport with no referee. That note argues making money shares sport’s structure — irreducible, played in real time, scored — while lacking its two saving graces, symmetric rules and contained stakes. Read through this one, the diagnosis gets more precise. It is not that money has no officials; it has a great many. It is that the practice does not carry its disputes to a position with authority over the rules themselves. The office exists in the shape of the game and stands empty, which is a different and more alarming condition than having no office.
The same move runs the other way on the machine. The Smart Loom holds that every machine that takes a card and acts is one loom or another — Jacquard’s cards became Babbage’s program, and the substrate never mattered. If “computer” is fluid in exactly that way, the interesting question was never what is it made of. It is who holds the card: which is a question about the surrounding practice, not the device, and the reason a factory loom and a loom in your kitchen are the same machine and not the same thing at all.
What this is not
It is not a claim that every term is fluid this way. Gold is fixed by atomic number and does not care what job it is doing; a bar of it in a vault and a filling in a tooth are the same kind. The old distinction between substance terms and role terms is not being dissolved here, only observed — and the interesting terms in any argument are usually role terms wearing the costume of substance terms.
It is also not a claim that the fluidity is a discovery of this note. Functionalism and multiple realizability are the standing name for it in philosophy of mind, and the triviality results above are the standing objection. What is offered is the pairing: putting the two words beside each other makes the bound visible, because the one whose etymology is a relation cannot hide that it needs a practice, and the one whose etymology is an activity can. Reading “computer” through “referee” is the whole contribution, and it is a small one.
Rests on: Putnam, Representation and Reality (1988), on unlimited pancomputationalism; Searle, The Rediscovery of the Mind (1992), on observer-relative computation; the triviality problem as set out in the Stanford Encyclopedia entry Computation in Physical Systems; and the ordinary etymology of referee (refer + -ee, disputes 1500s, umpire 1660s, law 1680s, sport 1820, peer review current). Offered, not proven: that a role’s bound is always the practice that confers it — stated here for two words and not shown for the class.
Kin to The Silent Second Term (readings need one too), Money, the Sport with No Referee (the vacant office), and The Smart Loom (who holds the card).