Ordinary language is all right.
One could divide humanity into two classes:
those who master a metaphor, and those who hold by a formula.
Those with a bent for both are too few, they do not comprise a class.
Fog rain snow all from the same gray sky. The sky is indifferent to what transpires.
In remarks subsequent to §81 the prospect of saying (or not being able to say) that 'wer die Sprache gebraucht, müsse ein solches Spiel spielen' is investigated mostly in terms of the idea of a rule—so, in terms of an idea that stresses a potential commonality between the juxtaposed terms, 'games' and 'calculi with fixed rules'. The idea of a rule lends these remarks some generality. Does it lend them enough to make clear Wittgenstein's reasons for having said, so generally, in §81 that in philosophy we can't say this 'müsse'?
§82 begins with a grammatical question: what do I call 'die Regel, nach der er vorgeht'? So Wittgenstein is considering different sorts of things he would ordinarily be willing to call rules of this sort: rules someone else, a third person, proceeds by, goes on according to, acts with reference to. And his answer to his own question produces three different cases: a hypothesized rule, gleaned from observing how the third person uses words, which gives a satisfactory description of that usage; a rule the third person consults—say, on a table, in a book of rules, however—when using 'signs' (letters, marks, orders scribbled on slips of paper); or a rule, expressed in words, given by the third person in answer to our question about his rule. So, three different things which might be called rules, one of which is indifferent to what the third person says but takes what he does as problematic, in need of some form of explanation; one of which is focused on what the third person does, in an entirely ordinary, unproblematic way (however he's doing it, it counts as 'consulting a rule' as far as we're concerned, and that's enough); and one of which is focused on what the third person says, perhaps (Wittgenstein does not indicate) with reference to his or our understanding of what a rule is or perhaps simply in asking him what he's doing or what he means to do—he may even cite some rule, some procedure, some authority elsewhere which he has learned to consult, to which he has learned to defer, on which he has come to rely—but at least allowing him some say (initial, if not final) as to what rule he proceeds by. —Or, in short: no say (we say); letting his actions speak for themselves; some say.
Wittgenstein's—or someone's, some voice's—subsequent question concerns the first and third of these possibilities—concerns, specifically, their failure: '—Wie aber, wenn die Beobachtung keine Regel klar erkennen läßt, und die Frage keine zu Tage fördert?'. The second possibility has, presumably, been left behind. It would have involved a clear case of going by a rule—an ordinary case. But that does not mean that failures of the other two possibilities to obtain are obviously less ordinary. Focusing on the latter (supposing, perhaps, that for the former, however our observing and hypothesizing works out, it can easily fail to pronounce definitively enough on the third person's use of words to determine 'the' rule by which he proceeds), Wittgenstein invokes, somewhat confusingly (crossing first and third persons), the possibility from §79 of saying that 'N is dead', then finding that more and more of the facts supposed in the explanations we offer as to who we mean by 'N' have proved to be false, and thus of being (per §82) 'prepared to withdraw this explanation and alter it'—an eventuality which seems to leave us at a loss:
'–So how am I to determine the rule according to which he is playing? He does not know it himself. —Or, more correctly, what is left for the expression "the rule according to which he proceeds" to say?'
'He does not know it' is a conclusion, drawn on something like this basis: that the third person has no rule to give us, and when explaining himself, is ready to cast explanations aside, to exchange them for different ones. To such an extent, it is imagined, that he maintains something like 'N is who I mean!' purely on the basis of his as yet unexhausted ability to explain differently who he means. I imagine this terminating in something like the third person's insisting: 'I know who I mean!'—without being able to say more about it. In §79, after all, before the various falsified facts began to be introduced into the situation being imagined, the person who said 'N is dead' meant (so he thought) a human being who he had seen, who looked like this, who did these things, who bore the name 'N' among others. So: 'I saw him! The one I saw! I don't care who he really was!'. Or something to that effect.
(I may be overinterpolating a little here. In §79, the question is, what if some point in this were to turn out to be false? So that the issue is when one would be prepared to declare, or admit, 'N is dead' to be false in light of the discovery that some such point bearing on who was meant by 'N' had turned out to be false. Perhaps I find it hard to imagine the kinds of points mentioned turning out so false. Perhaps I am drawing on the mood of §80's questions about the disappearing, reappearing chair.)
I said that the supposed cases of the third person's use of words failing to count as instances of 'rules by which he proceeds' were not obviously less ordinary than the simple one of his not consulting a rule, as when it is not the case that someone looks at a table, takes a rulebook from a shelf, etc. For the third possibility, of our going by what the third person says in order to determine the rule he's going by, if it turns out that he can't tell us (can't tell us any more than he already has, despite maintaining, say, that he means such-and-such by the words he has used, that this is what he does with them)—that too is ordinary. Sometimes, he can tell us. Sometimes, he can't (any further). Just as, when hypothesizing about him, sometimes, we can say that he is proceeding according to a rule, and what the rule is. Sometimes, we can't (to our satisfaction).
The questions with which §82 breaks off are typical Wittgensteinian questions. The usual understanding of them is as directed at a 'tempted' interlocutor or reader, one who wishes to say more than the ordinary meanings of his words would appear to say, and who has, it is supposed, had his claim to have said more, or to have had more to say, undermined by Wittgenstein's pointing out (via a grammatical investigation, or related ones) what it is that those words ordinarily do (are used to) say. But I am more interested in the role of these questions (and the investigation which precedes and succeeds them) in bearing out what Wittgenstein said in §81: that in philosophy we compare the use of words to games, calculi with fixed rules, 'but cannot say that someone who is using language must be playing such a game'. Has he shown something about this 'cannot'? (About its being something 'we' cannot say?)
'[W]ir nämlich in der Philosophie den Gebrauch den Wörter oft mit Spielen, Kalkülen nach festen regeln, vergleichen, aber nicht sagen können, wer die Sprache gebraucht, müsse ein solches Spiel spielen…'
The claim about our relation to the use of words in philosophy is more or less unrestricted: it concerns our relation to (our own) language, especially that aspect of language which has been emphasized throughout the first eighty sections of the Investigations, its use.
The claim says that we (often) (1) compare the use of words with 'games, calculi with fixed rules', but (2) cannot say that someone who uses language must be playing such a game.
How is it that we 'cannot' say this? I gather that Wittgenstein means that we cannot establish it to our satisfaction, can't be sure of ourselves when saying that this is what we're doing: that what we're doing, when using words (i.e., according to the Wittgenstein of the Investigations, when talking), is subject to some sort of 'must'.
And if we could say this—what would we be saying? The claim is that in our comparisons between our uses of words and certain games—which I take to mean, roughly, the specific sorts or forms of the use of language of which Wittgenstein has been giving and imagining examples all along—the games, or, let's say, our descriptions of them, enjoy some kind of priority over our uses of words. That priority is registered in a form of description of what we're doing: what we must be doing is playing those games—as opposed to, I guess, just doing whatever it is we're doing, in using the relevant words in the relevant ways, and however it is that we describe that, or however it is to be described (perhaps even, at the limit, using terms just like those in which the games have been described, excepting the 'must'?).
At least, that's one reading. There's another, because Wittgenstein's phrasing is exactly ambiguous, thanks to the paratactical conjoining of 'games' with 'calculi with fixed rules'. Why not just take this as an elaboration of what he means here by 'games'? —Because it is clearly not what he means by 'games' earlier in the book; because he immediately recurs to the former term, and subsequently the the latter, and in ensuing remarks backs off to such an extent that even imputing to us (when we use language, or do anything at all) the playing of definite games according to any rules at all is called into question—in short, because he seems to want to talk about both ideas, distinctly but in relation to one another.
According to the second reading, Wittgenstein is saying much the same thing, but saying it about a distinct object of comparison. And that other thing to which we compare our uses of language, 'calculi with fixed rules', would appear to offer a way to understand the source of the 'must' we are unable to do more than insist upon: at the end of this section Wittgenstein refers to his former self as having mistakenly thought that 'if anyone utters a sentence and means or understands it, he is thereby operating a calculus according to definite rules', rules with a kind of ideal definiteness such that our own languages would seem to 'only approximate to such calculi'.
Why the ambiguity? Why does Wittgenstein want to talk about both of these ideas in such proximity, practically an entanglement, with each other? Because all along he has been using the analogy to games to prepare the way for a criticism of some attraction to calculi with fixed rules? Because he is thinking of these calculi as kinds of games, which we do play for certain purposes? (But then why not say we cannot say we play those, rather than that we cannot say we play particular games at all?)
—I think because he wants to use the idea of a (language-)game in criticism of that old—Tractatus—idea of using language being a matter of operating a calculus, while also making sure that this criticism is not wrongly taken to imply that thinking of language in terms of games, rather than in terms of (logical) calculi, solves the problems that the latter was meant to. In the manner that the latter was meant to. And mainly, the problem of skepticism, as taken up in the Tractatus.
Perhaps the structure of the Investigations indicates this double aim. §§81–88 are, to me, oddly placed. Although there is a language-game introduced in §86—one which recapitulates much of the development through the book of its model, the builders' game from §2, simply by virtue of being played with a table that serves as a kind of rule for obeying builder A's orders—it seems like nothing, or not enough, gets done in these sections. And it seems like there's a new intensity of interest in the idea of a rule, but before that interest is really pursued (fully from §142 on), we get the remarks on logic's sublimity in §§89–108, and the metaphilosophical remarks in §§109–133. This new interest is interrupted. However: the reference back to Wittgenstein's Tractatus mistakes in §81 is complemented by the direct consideration, in §134, of a tractarian idea, that 'es verhält sich so und so' expresses the general form of a proposition. And soon after, with reference to the same case, when Wittgenstein notes that 'we call something a proposition if in our language we apply the calculus of truth functions to it', he gives the clearest example of what he might have in mind when talking about 'calculi with fixed rules' (in a book where he almost never refers to 'calculi' anyway): truth-functional propositional logic, on which the anti-skeptical hopes of the Tractatus were made to hang. As if the reason that the 'criticism of the Tractatus', or its conception, that one finds in §§89–133 can seem so glancing, is just that there's so little to say directly about it once you have some distance from the thought that our everyday language harbors (must harbor) some ultimately truth-functional analysis of its meaningful (possibly true) statements about things if our making contact with the world in language is to be guaranteed. —Once using the propositional calculus looks like one thing we do, but you retain the idea that something like skepticism is to be addressed, (sort of) answered, on the level of 'us, our language, and the world', with all the things we do that that includes.
Besides this structural oddity, there's just the fact that doubt is not really broached, as an explicit topic of Wittgenstein's remarks, until §§81–88. As if the double aim legible here has to do with the differences between his old and new ways of addressing skepticism, but it was not until this stage in the Investigations that he was in a position to voice skepticism, as he was not, in a different way, ready to rule it out of bounds until quite late in the Tractatus—6.51, nearly the end.
So what is his new way of addressing skepticism? And why is it caught up with this double sense in which there is a 'must' that we cannot, to our satisfaction, say?