Ludwig Wittgenstein and the case against the two-sentence summary

Ludwig Wittgenstein is the philosopher most people can summarise and fewest have read. The summary runs to two sentences: first he drew the limits of language and declared everything past them unsayable, then he changed his mind and decided that meaning is use.

It is tidy, memorable, and the largest obstacle to understanding him. It turns one life into two careers, and a method into slogans repeatable by anyone who has never opened either book.

Specialists have largely stopped believing it. The Stanford Encyclopedia of Philosophy's entry records that the orthodox two-stage reading is under pressure: "In more recent scholarship, this division has been questioned: some interpreters have claimed a certain unity between all stages of his thought, while others talk of a more nuanced division, adding stages such as the middle Wittgenstein and the post-later Wittgenstein."

What looks like two philosophers may be one experiment run twice#

Why the argument comes out as it does is itself instructive. Emphasise the move from logic to grammar and "the break between the philosophical stages is easily marked". Emphasise the resistance to generalisation of any kind, and "the similarity in Wittgenstein's therapeutic motivation throughout his life is much more clearly seen".

That is not a quarrel about dates but about what the man was doing. On the first reading he is a theorist who built a system, found it wanting, and built another. On the second he is rarer: one radical experiment about what philosophy is, run in stages over thirty years, early and late sharing a target - the philosophical problem itself. Not solving it. Getting rid of it.

A book that ends by cancelling itself#

The Tractatus Logico-Philosophicus, written largely in the trenches and in an Italian prisoner-of-war camp and published in German in 1921, is a short sequence of numbered propositions reading like theorems. It applies the new logic of Frege and Russell to the oldest questions there are: what the world is, what a thought is, how a sentence is about anything at all.

Its answer is the picture theory. A proposition represents a possible state of affairs the way a model represents a collision, by sharing a structure with it. What can be said is what can be pictured so; everything else is not false but nonsense, in the technical sense of saying nothing.

The consequence is the book's real payload. A great many classical problems turn out not to be unsolved but badly formed: they arise because we misunderstand the logic of our own language and take the confusion for a deep question. Clear the confusion and the question is not answered. It disappears.

Then comes the paradox. By its own standard, almost everything the Tractatus says is itself unsayable: you cannot use language to picture, from outside, the relation between language and the world. Wittgenstein does not flinch. Near the end he tells the reader that his propositions are elucidations for whoever has climbed up through them and beyond; the ladder is to be thrown away once used. The book closes with the sentence everyone knows: "Whereof one cannot speak, thereof one must be silent."

Most philosophers defend the system they build. This one tells you to kick the ladder over.

Ten years of not being a philosopher#

Having, as he saw it, finished the subject, he left it. In the 1920s, the Stanford entry records, Wittgenstein was "now divorced from philosophy" and "gave away his part of his family's fortune and pursued several 'professions' (gardener, teacher, architect, etc.) in and around Vienna."

The fortune was among the largest in Austria; he divided his share between his siblings, on condition that nothing be held in trust for him. He trained as a primary schoolteacher and taught in mountain villages - Trattenbach, Puchberg, Otterthal - from 1920 to 1926, compiling a spelling dictionary for his pupils. He spent a summer gardening at a monastery. From 1926 he took over the design of a house in Vienna for his sister Margaret, commissioned from the architect Paul Engelmann, who later regarded him as the author of the finished building. He spent a year on the door handles, and with the house all but complete had a ceiling raised thirty millimetres.

It is tempting to file this as colour: the millionaire, the door handles, the ceiling. Read it instead as evidence. Philosophy was never for him a profession with a system attached; thinking the work done, he did not stay in the field on the strength of a finished book but did something else entirely for ten years. The exactitude he brought to a radiator is the exactitude he brought to a proposition.

The list in section 23 is the whole argument#

He came back in 1929, partly through contact with members of the Vienna Circle who had been reading him closely and misreading him intolerably. What followed was less a retraction of the Tractatus than a reversal of its central assumption: that beneath the surface variety of language lies a single hidden machinery.

The Philosophical Investigations, published posthumously in 1953, replaces that machinery with language-games. The term is deliberately homely. Section 23 lists things we do with words - the Stanford entry calls the list astonishing - taking in reporting an event, forming and testing a hypothesis, making up a story and reading it, play-acting, guessing riddles, making a joke, solving a problem in practical arithmetic, translating from one language into another. It ends:

Asking, thanking, cursing, greeting, praying.

Read it slowly and the point lands. Ordering a colleague about, joking with a friend, swearing at a stubbed toe, telling a child off, praying, doing a long division, explaining the rules of a game, saying that you love someone: all of it is language, and no single account of how words hook onto objects will cover it. His earlier self hunted the essence of the sentence. His later self observes that a hammer, a saw and a glue-pot are all tools, and that nothing is gained by hunting the essence of the tool.

Meaning is use - but read the qualification#

The slogan is real. The sentence behind it is more careful: "For a large class of cases of the employment of the word 'meaning' - though not for all - this word can be explained in this way: the meaning of a word is its use in the language." Note the hedge. He is not offering a theory of meaning to replace the picture theory, but a way of looking that dissolves the demand for one.

The companion idea is family resemblance. Ask what all games have in common. Board games, ball games, card games, the Olympic Games, a child bouncing a ball against a wall. Some involve competition, some not; some turn on luck, some on skill. Don't think, he says, but look. What you find is not a shared essence but "a complicated network of similarities overlapping and criss-crossing", like the build and gait running through a family without any one feature appearing in every member.

Its afterlife outside philosophy has been long: anthropologists use it for kinship and ritual categories that resist necessary and sufficient conditions, social scientists when a concept does real work but will not sit still for a definition, and it is now standard in concept analysis and in the treatment of prototypes. It is probably his most borrowed idea, and most often borrowed by people who would never call themselves Wittgensteinians.

Why you cannot keep a diary of a private sensation#

The private language argument rewards slow setting out, because it is routinely mistaken for a denial that inner life exists.

Suppose I want a word for a sensation nobody else can reach. I write "S" in a diary on every day I have it. No public definition is possible: I cannot point to the sensation, and nobody can check my usage.

A fortnight later I have the sensation again - or think I do - and write "S". What makes that entry correct? All I can consult is my memory of what I meant last time, and my memory is the thing in question. There is no independent standard to test it against, so the distinction between following my rule correctly and merely believing that I am has quietly collapsed. Where it collapses, as Wittgenstein puts it, whatever seems right to me is right - which only means that here we cannot talk about right at all.

The conclusion is not that pain is unreal or that nobody has an inner life, but that a word functions as a word only where there are public criteria for its correct use: sensation language grips through behaviour, circumstance and shared practice, not private ostension.

It is inseparable from the treatment of rule-following, which the entry calls the climax of the book: "This was our paradox: no course of action could be determined by a rule, because every course of action can be made out to accord with the rule." A finite run of examples never fixes a unique continuation. What fixes it is a practice, a training, a shared form of life.

One influential reading of that passage - introduced by Robert Fogelin in 1976 and made famous by Saul Kripke in 1982 - takes Wittgenstein to be advancing a sceptical paradox about meaning and then offering a sceptical solution. It generated a literature of its own. It is also an interpretation rather than a transcript, and many readers, among them the late Anthony Kenny, have held that the sceptical framing imports the very problem Wittgenstein wanted removed.

The half of the work almost nobody reads#

Here is the fact that should reorganise the picture. The Stanford Encyclopedia's entry on his philosophy of mathematics states that "more than half of Wittgenstein's writings from 1929 through 1944 are devoted to mathematics, a fact that Wittgenstein himself emphasized in 1944 by writing that his 'chief contribution has been in the philosophy of mathematics.'"

More than half. And by his own estimate, that was where the important work lay.

Very little of it reaches the general reader. The Remarks on the Foundations of Mathematics, published selectively from manuscripts in 1956, is difficult, and defends positions mathematicians have found unpalatable: a finitism that locates the mathematical infinite in recursive rules rather than completed totalities; a rejection of the idea that mathematical terms refer to objects waiting to be found; the insistence that the mathematician invents rather than discovers. His remarks on Gödel are widely judged a misfire.

The neglect is telling. A discipline that carries a two-sentence summary of a thinker drops whichever half does not fit, and the mathematics does not fit. Readers curious about the state of those questions might start with this year's FilMat congress.

What can only be shown#

The Tractatus is not a positivist tract, whatever the Vienna Circle hoped. Its author held that what can be shown cannot be said, and that the things which cannot be put into words "make themselves manifest. They are what is mystical."

Ethics, on that account, belongs to the shown rather than the said - a claim about the logic of value language, not a licence for vagueness. He is reported to have told a prospective publisher that his book had two parts, the one on the page and everything he had not written, and that the second was the important one, because the ethical gets its limit drawn from the inside.

He took religion seriously without being a believer in any straightforward sense, holding that religious language does something other than assert badly evidenced facts. None of this makes him a mystic. He was unusually hostile to the consoling blur; a limit was drawn to stop people talking nonsense on either side of it.

The question the AI debate keeps skipping#

A model says "I understand". It says "I'm sorry". It produces a page on distributive justice that would pass in a seminar. The argument that follows is always the same: does it really think, or only simulate thought?

Wittgenstein's first move would be to refuse the question in that form and ask a smaller one: under what circumstances do we say of anything that it thinks? We say the dog wants to go out, that a chess engine is trying to control the centre, that someone deep in a book is thinking - and we are neither confused nor making metaphysical commitments. The demand for a hidden inner fact that makes one attribution true and another false may be precisely the demand the Investigations was written to dissolve.

That reframing is not an answer, and it is not a licence. Handled carelessly it becomes cheap deflationism: the machine says it understands, we use the word that way, so it understands. Wittgenstein grants no such permission. Language-games are embedded in what he called forms of life - bodies, needs, training, a shared world of practice - and it is genuinely contested whether a system with no life to have a form of participates in one or merely echoes ours. The public-criteria point cuts both ways: inner facts do not settle the matter, but the criteria are ours.

Both camps have claimed him, which should slow everyone down. What is defensible is narrower: much of the debate runs on a vocabulary nobody has examined, and examining it changes which questions look urgent. That is visible in the arguments over talking to language models and in the afterlife of Searle's Chinese Room.

The legacy is a method, not a position#

Wittgenstein left no doctrine to be examined on. He left a way of proceeding: when a question will not come out, suspect the question. Look at how its words are used, in the ordinary business of asking and thanking and cursing and praying, before deciding what deep fact they track.

He described his aim in philosophy as showing the fly the way out of the fly-bottle: a modest image for an immodest ambition. The fly is not short of information; it has been round the glass many times. What it needs is to look properly at something it has been staring through all along. That is not a summary of a philosophy. It is closer to an instruction, and it is why he outlives the two sentences we use to file him away.