Chapter 3

Updated 2026-07-02

"Don't say: 'There must be something common, or they would not be called "games"' — but look and see whether there is anything common to them all." — Ludwig Wittgenstein, Philosophical Investigations §66 (1953)

A kid trying to win an argument about whether tag is a "real game" will, without knowing it, walk the exact path a philosopher once walked on purpose. A real game has rules. Tag has rules. A real game has a winner. Tag doesn't — someone's "it," but nobody wins tag, the game just keeps going until everyone's tired or it's dark. Fine, a real game has an objective, then. Don't get tagged. That's not an objective, that's just not losing. Around and around, every rule patched the moment an example breaks it, until the kid gives up and says the thing every kid eventually says about every word: you know what I mean.

Try to define the word game yourself and you'll patch the same holes in the same order.

Chess has winners and losers; so does poker. Catch has no winner, no score, no end. Solitaire has no opponent at all, just a person, a deck, and an hour to kill. Every rule you propose — competition, two or more players, a winner — survives a few examples and dies on the next one.

The assumption doing the failing here is much older than any dictionary. Aristotle set it down some twenty-three centuries ago: a category is a checklist. To be a man is to be a rational animal — meet both conditions and you belong, miss either and you don't, and nothing in between exists. Membership by necessary and sufficient conditions, all-or-none, crisp at the edges: the scheme worked so beautifully for triangles and odd numbers that it became the unexamined default for everything else, and it held the job, essentially unchallenged, for over two thousand years.¹ Every definition you have ever looked up is built on Aristotle's template — which is why defining "game" feels like it ought to work, right up until it doesn't.

This is not a hard question that happens to lack an answer. The philosopher Ludwig Wittgenstein, working through it in the years before his death in 1951, decided it doesn't have the kind of answer being asked for. "Don't say: 'There must be something common, or they would not be called "games"' — but look and see whether there is anything common to them all," he wrote. "For if you look at them you will not see something that is common to all, but similarities, relationships, and a whole series of them at that."²

Look and see. Not reason your way to a definition — look at chess and solitaire and catch and ring-around-the-rosie side by side, the way you'd lay out photographs of relatives, and notice what Wittgenstein noticed: a family resemblance. A nose here, a chin there, nothing shared by all of them, everything shared by some pair of them. The category doesn't hang from one rule the way a coat hangs from a hook. It holds together the way a rope holds together — not because one fiber runs its entire length, but because enough fibers overlap that the whole thing doesn't come apart in your hands.

So where does the rope end? Where exactly does a game stop being a game? Wittgenstein's answer is short: you can't say. And — his next move, almost a shrug — that never troubled you before, back when you were just using the word, before anyone asked you to defend it.³

That parenthetical describes every category in your head, not just "game." You have never once needed a boundary to use a word correctly. You've used "game" your whole life — board games, mind games, the Super Bowl, "the game is up" — without an edge to stand on, and it never once slowed you down. The boundary only becomes a problem when someone builds something that requires one: a tax code that has to decide if poker winnings count as gambling income, a toy-store website that has to put Monopoly in exactly one aisle. The systems will demand it. But the mind got here first, and it got here working differently than anyone assumed.

---

A psychologist, a generation after Wittgenstein, decided to actually measure the fuzziness instead of just arguing for it.

Ask people whether a robin is a bird. Yes, obviously, instantly. Ask them whether a penguin is a bird. Also yes — penguins have feathers, lay eggs, are unmistakably classified as birds in every encyclopedia that exists. But ask both questions to a room full of people with a stopwatch running, the way Eleanor Rosch and her colleagues did through the 1970s, and something strange shows up in the data: people agree to "a robin is a bird" faster than they agree to "a penguin is a bird."⁴ Not a little faster. Reliably, measurably faster, across hundreds of trials and dozens of categories. Apples are faster than figs. Chairs are faster than lamps. Robins are faster than penguins, and penguins are faster than ostriches, who can't fly any better than penguins can but at least look the part standing up.

The category "bird" isn't a flat container where everything inside is equally inside. It has a center and it has edges, and the center answers faster because the center is what the word actually means to you, moment to moment, before you stop and think about the dictionary. Robins are good birds. Penguins are bad birds — bad at being the example of bird, while being, by every formal test, completely real ones. Rosch's subjects didn't just answer fast or slow; asked directly to rate how good an example something was, they agreed with each other almost perfectly, "even for categories about whose boundaries they disagree."⁵ Two people can argue for an hour about whether a tomato counts as a fruit and still agree instantly, without discussion, that an apple is a better fruit than a tomato either way.

This is what a prototype effect looks like from the inside: categories with a best member, fading by degrees to a barely-qualifying one, and everybody, without being taught, ranking them the same way. Membership isn't a light switch — it's a dimmer, and the setting is something you can clock with a stopwatch. Rosch primed people with a category name, "bird" or "fruit" or "vehicle," then flashed a word and timed how fast they could say yes or no. Priming "bird" sped up "robin" and slowed down "penguin," as if the word "bird" had quietly handed the listener a picture of a robin first and only reluctantly admitted the penguin afterward. It is one of the strangest things a stopwatch has ever measured: not whether something is true, but how typically true it is, in milliseconds, agreed on by strangers who have never discussed it.

A dictionary lists "fruit" as the seed-bearing structure of a flowering plant, full stop — apple and fig sit on the same line, equally fruit, equally finished. Nobody's head works that way. Everybody's head has a faster apple and a slower fig.

---

A bachelor is an unmarried adult male. By that definition, the Pope is a bachelor. So is a three-year-old boy. So, technically, is a man who has shared a bed with the same partner for fifteen years and never filled out a marriage license. Ask anyone whether these are good examples of "bachelor" and you'll get an uncomfortable laugh, because something about all three feels wrong, even though nothing about the definition is violated.

What's wrong isn't the definition — it's that "bachelor" was never just unmarried-adult-male — it comes with a whole unstated scene attached, a normal-shaped world where adults reach a certain age and are expected, by custom, to marry, and where "bachelor" describes the specific, slightly raffish status of someone who's eligible and hasn't yet. Pull a Pope or a toddler or a fifteen-year, unmarried partnership into that scene and the word doesn't apply cleanly; it just sort of clatters. The linguist George Lakoff called this kind of background scene an idealized cognitive model — not a rule about the word, but the imagined-normal situation the word is standing inside of.⁶ Most words are. You don't notice the model until someone hands you a case it wasn't built for, and a quiet decision about who counts was baked into the word the whole time, waiting. Every form with a box for "marital status" inherited that same buried scene, and didn't know it.

---

Some categories don't even bother pretending to have a single center. They're built more like a spider's web than a circle, radiating outward from one case in chained, motivated steps.

Take "mother." A birth mother carried you. A genetic mother contributed the egg — usually the same person, except when it isn't: surrogacy and donor eggs can split birth and genetics into two different women. A nurturing mother is whoever raised you, which may be neither. A marital mother is your father's wife. A genealogical mother sits at the top of your family tree. In the overwhelming majority of cases all five models stack on one woman and you never notice they were five models at all. But unstack them — a stepmother, an adoptive mother, a surrogate — and nothing breaks. Each case simply foregrounds one of the five strands and drops the rest, and not one of those uses feels like a mistake or a stretch. "Necessity is the mother of invention" foregrounds a different strand again — origin, cause — and that doesn't feel broken either. There is no single definition sitting under all of these. There's a center, with motivated extensions radiating off it the way relatives radiate off a family tree, related without being identical, and every fluent speaker holds the whole structure without ever being taught it as a structure.⁷

Even a word as small as a preposition does this. Over. The plane flew over the city — above. The cloth is spread over the table — covering. She climbed over the wall — across and beyond. The meeting is over — finished. He has power over them — control. Twenty-some senses, easily, all chained out from one image of above-and-across, none of them random, none of them listed as separate entries in the dictionary in your head the way they would have to be in a printed one.⁸ You use four or five of these every day and have never once confused them, never needed to ask which "over" your friend meant. It's not twenty different words that happen to be spelled the same. It's one constellation, and you read the whole shape of it instantly, every time.

---

The clearest version of this — the one that shows the logic isn't decoration, it's structural — comes from a language with almost nothing in common with English.

Dyirbal, spoken by an Aboriginal community in northeastern Australia, sorts every noun in the language into one of four classes, each one marked by a small word the way English marks number with "-s." The class that linguists, working from the fieldwork of R. M. W. Dixon, label balan contains, among many other things: women. Also fire. Also water. Also dangerous things — stinging insects, certain fighting implements, anything that can hurt you.⁹ Women, fire, water, and danger, filed together, which is the kind of fact that gets quoted as proof that "primitive" categories are just arbitrary noise, a grab bag with no logic at all.

It isn't arbitrary — it's chained, the same way "over" is chained, just along different links. In Dyirbal myth, the sun is the wife of the moon — a mythic association that pulls sun-words into the feminine class, and fire is closely linked to the sun, so fire follows. Fire is also, separately and directly, dangerous, which pulls in the other dangerous things on a second, independent thread. Water gets pulled in along its own associative path. Nothing here is a rule you could state as a definition — "balan = female, or related to female by myth, or dangerous, or related to danger" is not a sentence anyone could use to sort a new noun correctly without already knowing the web. But it is also not noise. Every link is motivated by something — a myth, a shared danger, a domain of experience — even when the chain of links runs four steps long and arrives somewhere a stranger to the culture would never predict.

This is the same shape — the center, the motivated extensions — scaled up to a whole grammar: not a rule with a hard edge, and not chaos either. A center, and motivated extensions radiating outward, link by link, each one a small, sane decision that made sense at the time it was made. Penguins are still birds, the Pope is technically a bachelor, balan still holds women and fire and water together — the edges never get any cleaner. But follow any one of these categories back along its threads and you don't find a coin flip. You find a reason, every time, even when the reason is a myth nobody remembers choosing.

The fuzziness was never the absence of logic — just a kind the dictionary, or a database, was never built to hold.

---