The Moving-Pictures Stage
When the Lumière brothers first screened films at the Grand Café in 1895, audiences saw photographs that moved — workers leaving a factory, a train arriving at a station. The films were evaluated by the borrowed criteria of photography: clarity, composition, fidelity. No one imagined narrative cinema, the close-up, the montage sequence, or the continuity editing system that would later make film a medium with its own grammar. The capabilities were invisible because they were being perceived through categories that could not accommodate them. The moving-pictures stage is the structural phase that recurs with every new medium — the phonograph used as dictation device, radio used as wireless telegraph, AI used as a faster word processor. The medium's distinctive operations are invisible during this phase and become visible only through the development of protocols specific to the new medium.
The Substrate Lock — Contrarian ^ Opus
There is a parallel reading that begins from the material conditions required for AI's operation. Where the moving-pictures analogy suggests an organic evolution toward new forms, the computational substrate tells a different story. Early cinema required only light, chemistry, and mechanical movement — resources that could be democratized. AI requires server farms, rare earth minerals, and energy infrastructures controlled by a handful of corporations. The moving-pictures stage isn't just a transitional phase; it's a permanent condition enforced by the concentration of computational power. The "new protocols" that supposedly emerge through practice are actually predetermined by the architectures of the models themselves and the business models of their providers.
The lived experience of cultural workers during this transition reveals another constraint the moving-pictures framework obscures. When photographers lost work to daguerreotypists, they could at least learn the new craft with modest capital. When writers and artists lose work to AI systems, they face a fundamentally different barrier: they cannot own or meaningfully modify the means of production. The "accumulated decisions of practitioners" that Gitelman identifies as driving media evolution presume a relatively open field of experimentation. But AI's moving-pictures stage unfolds within walled gardens where the terms of experimentation are set by platform owners. The diagnostic moments that flash through — those glimpses of AI's distinctive capabilities — aren't seeds of new protocols but glimpses of a future that's already been designed elsewhere. The gap between what AI can technically do and what protocols allow isn't a space of potential; it's a carefully managed zone where the appearance of evolution masks the reality of capture.
In the AI Story
The stage recurs with such consistency that Gitelman treats it as a structural feature of media transitions. The technology is used to perform tasks that existing media already perform, and the evaluation criteria are borrowed from the displaced medium. The borrowed criteria are coherent — they describe what the new medium does well relative to the old one — and also profoundly inadequate, because what the new medium can become is not deducible from what it currently is.
AI-assisted cultural production is at the moving-pictures stage. The Orange Pill itself is in most respects a moving-pictures artifact — a book that uses AI to produce a faster, smoother version of what books have always been. The structure is conventional. The argument is sequential. The voice belongs to a named individual. But the text contains moments — the laparoscopic surgery connection, the cross-chapter links — that point beyond the moving-pictures stage toward capabilities specific to the new medium.
The exit from the stage is slow. It requires not just technical refinement but the wholesale development of new protocols — new production practices, new distribution channels, new evaluative frameworks, new categories for crediting participants and assessing value. The exit is never directed by any single participant; it proceeds through the accumulated decisions of practitioners, institutions, and audiences operating without established norms.
The political significance of the stage lies in the gap between what the technology can technically do and what the existing protocols allow it to be understood to do. The gap is the space in which fluent fabrication thrives — where AI-generated outputs inherit the epistemic guarantees of print-culture formats without the processes those formats conventionally imply.
Origin
The concept derives from Gitelman's historical work on early cinema, early phonograph culture, and early Internet development — three cases where the structural pattern is clearly documented in the archival record.
Key Ideas
Faster or cheaper. The new medium is first deployed to produce faster, cheaper, more convenient versions of what existing media already produce.
Borrowed evaluation. The criteria for assessing the new medium's products are borrowed from the displaced medium, producing a coherent but profoundly inadequate framework.
Invisible capabilities. The medium's distinctive operations are invisible during this phase because they are being perceived through categories that cannot accommodate them.
Exit through practice. The stage ends through practical experimentation, not through theory — through formal innovations that accumulate into conventions specific to the new medium.
Diagnostic moments. Flashes of the new medium's distinctive operations occur within the borrowed framework; these moments are the seeds from which new protocols grow.
Appears in the Orange Pill Cycle
Evolutionary Capture Dynamics — Arbitrator ^ Opus
The tension between organic media evolution and infrastructural determinism depends entirely on which temporal frame we examine. At the phenomenological level of user experience, Edo's moving-pictures framework is 95% correct — we are indeed using AI like faster typewriters, and new capabilities do emerge through practice. But at the political-economic level, the contrarian view dominates 80% — the substrate requirements and ownership structures fundamentally constrain what "new protocols" can emerge. The question "How does a new medium develop its grammar?" yields one answer; "Who controls the development?" yields another.
The synthetic frame that accommodates both views is evolutionary capture — a process where genuine formal innovation occurs within increasingly narrow channels determined by infrastructure ownership. Early cinema's grammar did evolve through distributed experimentation, but within the constraints of the studio system that eventually emerged. AI's moving-pictures stage similarly allows real creative discovery, but only along vectors compatible with platform architectures and business models. The "diagnostic moments" Edo identifies are genuine — they do point toward new capabilities — but they point toward a future whose basic parameters are already set by computational infrastructure.
What both views correctly identify is that we're in a hinge moment where the appearance of openness hasn't yet collapsed into the reality of closure. The moving-pictures stage is both a genuine phase of creative experimentation (Edo's view) and a managed transition toward predetermined ends (contrarian view). The right weighting shifts depending on whether we're asking about immediate creative possibilities (where experimentation is real) or long-term structural outcomes (where capture is nearly complete). The borrowed clothes of print culture aren't just hiding AI's capabilities — they're also hiding the mechanisms by which those capabilities will be channeled into profitable, controllable forms.
Further reading
- Tom Gunning, "The Cinema of Attractions: Early Film, Its Spectator and the Avant-Garde" (1986).
- Lisa Gitelman, Scripts, Grooves, and Writing Machines: Representing Technology in the Edison Era (Stanford University Press, 1999).
- André Bazin, What Is Cinema? (University of California Press, 1967).