Saying and Showing
The Tractatus draws a line. On one side: propositions that picture possible states of affairs, that can be true or false, that constitute the world as totality of facts. On the other side: logic, ethics, aesthetics, the sense of the world, the mystical. These cannot be said; they show themselves. The distinction between saying (sagen) and showing (zeigen) is among Wittgenstein's most consequential contributions and one of the most productive lenses through which to read the AI language moment. The later philosophy relocated the distinction rather than abandoning it: within ordinary language itself, some dimensions resist propositional capture — the tone of a remark, the timing of a pause, the quality a designer means when she says spacious.
In the AI Story
The Tractatus ends with the observation that logic is not a subject among subjects because it cannot be pictured — the form that makes representation possible cannot itself be represented. Ethics and aesthetics are similarly not subjects about which true or false propositions can be formed; they manifest in how one lives and what one attends to. These are not empty regions. They are the conditions under which the world of facts has weight.
Applied to computing, the distinction is diagnostic. The formal language says: this button at these coordinates, this function returning this type. It does not show: this interface should feel spacious, this interaction should feel responsive, this error message should reassure rather than alarm. Every design review contains a moment where someone says something that cannot be converted into specification. It doesn't feel right. The flow is off. The hierarchy is aggressive. These are pointings at quality, not descriptions of formal properties.
The language interface begins to close this gap — not by making the unsayable sayable but by creating a channel through which showing can enter the computational process. The designer tells Claude I want this to feel spacious but not empty; the text should breathe. The machine has absorbed the statistical association between quality-language and implementation-decisions and can generate approximations.
This creates the iterative process the Ludwig Wittgenstein — On AI volume analyzes in Chapter 9. The human shows. The machine generates. The human evaluates against the felt sense of rightness. The cycle converges through successive approximation. What is gained: access to the dimension that formal specification excluded. What is risked: the confusion of recognition with understanding — the machine's statistical approximation of showing mistaken for the act of showing itself. The Orange Pill Cycle's ascending friction finds its philosophical home here: the old friction was in saying; the new friction is in showing — in knowing what one wants, recognizing when the machine has captured it, distinguishing outputs that feel right because they are right from outputs that feel right because they are smooth.
Origin
Formulated in the Tractatus propositions 4.12–4.1212 and elaborated through the book's conclusion. The later philosophy modifies rather than abandons the distinction: showing migrates from the mystical into the texture of ordinary language use.
Key Ideas
Two regions of meaning. The sayable is propositional; the shown is revealed in structure, form, and practice.
Not empty, not unsayable in a deficient sense. What shows itself is often what matters most — the framework within which facts acquire weight.
Design as showing. Quality, tone, responsiveness, spaciousness — these are shown, not specified.
AI's partial access. The language interface creates a channel for showing to enter computation, but the machine approximates statistically what humans do through participation.
The new friction. Removing the friction of saying reveals the harder friction of showing — the judgment that cannot be transferred to the machine.
Appears in the Orange Pill Cycle
Further reading
- Ludwig Wittgenstein, Tractatus Logico-Philosophicus 4.12–6.522
- Cora Diamond, The Realistic Spirit (1991)
- James Conant, "Must We Show What We Cannot Say?" (1989)
- Peter Winch, The Idea of a Social Science (1958)