(Rather than describing a language as it is used—a game—in §48 Wittgenstein has described a language that can be used—an instrument.)

'Just be done': or, better, the language does it. For 'no people', it could just as well be said that the language described in §48 has no use, but it could have one.

§48's color-square language-game contrasts with the other major games in the Investigations in giving little to no hint of the people (one must imagine) playing it. As it is an abstraction, the players are an afterthought, or, say, some things about gameplay are still to be determined. Or may be. (Wittgenstein gestures at some relevant circumstances in §49b). Of course, the language 'serves to represent combinations of colored squares on a surface', as language (2) was 'meant to serve for communication between a builder A and an assistant B'. And that can, as it were, just be done, without taking the players into account. —But §49b does invoke the idea of a move in a game. Akin, one might want to say, to some notion of validity or invalidity.

Odd that there aren't more denunciations of the eisegetes among Wittgensteinians.

—The patterning there is akin to that of the introduction of the color-square game in §48, where an application of 'the method of §2' to the account just quoted from the Theaetetus follows a quick, separate grammatical investigation in §47 (cf. §156b–g). But it's not obvious (?) whether there is a corresponding source for an 'account' in the case of the 'reading' interpolation—not that I know whether there need be. It seems significant that Wittgenstein chooses to punctuate his text's return to 'the method of §2' with another substantial external quotation of an idea he can just as well self-cite—as he does parenthetically at §46c. —But is §157 (the reading-machines) a 'return to "the method of §2"'?

One way in which Wittgenstein's interpolated remarks on 'reading' starting at §156 are special is that they juxtapose his two styles of representing language-games, designatively (like §23, referring to analytically separable parts of ordinary language) and stipulatively (like §2 or §143). The use of human (or 'other') beings as 'reading machines' considered in §157, with variant cases to follow, is undoubtedly a 'stipulated' language-game even though it's not referred to as such. Yet in §156b we have:

'Der Gebrauch dieses Worts unter den Umständen unsres gewöhnlichen Lebens ist uns natürlich ungemein wohl bekannt. Die Rolle aber, die das Wort in unserm Leben spielt, und damit das Sprachspiel, in dem wir es verwenden, wäre schwer auch nur in groben Zügen darzustellen.'

He goes on to compare a normally fluent reader (who can, it is noted, read in a variety of ways, at a glance, word-by-word, with comprehension, paying no attention, etc.) with a beginner who reads words 'indem er sie mühsam buchstabiert'. And with the contrast, an onset of philosophizing.

—So observations about a 'real language-game' are met with a description of an invented one, the reading-machines of §157. But whereas earlier games could often be described as ways of using language, using words (like tools), in order to consider the phenomenon, the language-game, at issue in this case, Wittgenstein must describe a way of utilizing, using, people who use (?) words. As if in the case of reading, we come to use words by learning to use ourselves in a peculiar way which can only occasionally come to view from where we now stand.

In his second concession in §28, Wittgenstein reverses his first: the listener could take a name for a number, rather than a number for a name. In effect, this asserts that names are to possess no special attraction for the kinds of misinterpretations under consideration in the remark. The third concession underlines the significance of this parity: for what's meant as a name could just as well be taken to be a word for a color, a racial designation, or a cardinal direction. In other words, it's not just that some special kind of word (a name) might attract certain misunderstandings of ostensive definitions, or even that any word apparently properly attachable to some ostendable thing (a name, a number, a color, etc.) might do so: even words whose significance attaches to the manner of pointing involved in defining or using them might be misunderstood because of pointing which indicates, by accident, some person rather than the 'nothing' 'that way' of a direction. It's not (proper) names as such, it's not things as such: it's the overlap between language-games.

Or Bach at night too.

Bach in the morning, Kraftwerk at night.