Process Legibility — Construction as Image

Primary sources: critical text on Stained Unravel, a series within the Senescenence exhibition (Interface Gallery + Nguyen Wahed, New York, April 2026); "Magicians Conceal, Artists Reveal" (→ summary).

Magicians conceal, artists reveal

The opposition is ethical before it is aesthetic. The classical doctrine ars celare artem ("art is to conceal art") held that supreme craft should erase its own traces — no toolmarks, no seams, no visible labor. Technique existed to disappear. The modernist tradition systematically dismantled this:

By Conceptualism, the logic of production is not the image's underbelly to be hidden — it is its architecture, its substance. This is the lineage van den Dorpel inherits and radicalizes. The radicalization is structural: in a cellular automaton, "the form cannot hide the logic that produced it any more than a woven cloth can hide the structure of its weave." Concealment is not refused — it is impossible. The image is not an effect. It is a record.

Pentimento — from Italian pentirsi, to repent — names the visible trace of revision: earlier states showing through later ones, the artist's changed mind legible in the work. To leave pentimento is to carry history rather than erase it. The cellular automaton generalizes this into a structural condition: every transition is pentimento.

The central claim

"Harm van den Dorpel devises systems in which the construction of an image is the image. The process does not precede the work — it is legible within it, leaving its traces in every configuration, every transition, every arrested form."

Note 992 formulates this as an equation: The Work = The Work + Its Documentation. That is additive — documentation supplements. The Stained Unravel text goes further: the construction is the image, not a supplement to it. The process does not precede the work and then leave; it remains legible within every form the work takes.

The image at any given moment is described as "a record of the logic that produced it: its restrictions show, its seams are exposed, and the history of each transformation remains readable in the forms left behind." The work is its own archaeology. To read it is to read the decisions — and the rules — that produced it.

This makes a claim about what a generative image is. It is not a thing produced by a process; it is a process at a particular moment of its unfolding, frozen enough to be seen. The distinction matters: if the image were merely the product of a process, the process could be discarded once the image existed. If the image is the process made legible, the process is irreducible — it persists in what you see.

Stained Unravel: the specific system

Stained Unravel — Stitched Gaze
Stitched Gaze (2026) — Stained Unravel; "the grid unravels in all directions simultaneously, like a sweater coming unwoven"

The works in the Stained Unravel series use a cellular automaton whose rules are "some constructed by the artist, some generated by the software itself." The grid "unravels in all directions simultaneously, like a sweater coming unwoven."

A crucial reversal: "Birth, survival, and death are not the rules but their consequences." In Conway's Game of Life, birth/survival/death are the rules. Here they are outcomes of a deeper rule-set — the system generates its own vital logic from more fundamental operations. What we observe (cells appearing, persisting, dying) is the surface of something more structurally remote.

When a ruleset reaches equilibrium — when the system has worked through its current logic — "the rules undergo reconfiguration, restoring motion." The system does not terminate; it transitions. Equilibrium is not the end but a threshold that triggers a new beginning. This is senescence precisely: not catastrophic death but the exhaustion of one form followed by the emergence of another (→ senescenence).

The hermetic vs. the gentle process (Struggle for Pleasure, 2024)

The Struggle for Pleasure press release documents an explicit methodological shift that sharpens this page's central claim:

"Historically, I dedicated myself to an iterative algorithmic process which in the end hermetically generated the final 'pleasurable images' to be released. In the past, that iterative process involved a great deal of loss, discarding all outputs before the final algorithm would be complete."

The old method was hermetic: iteration invisible, intermediate states discarded, only the terminal output presented. The new method — first used in Struggle for Pleasure — is archival:

"Keeping images generated at all points during the iteration process and curating them into a final selection. The collection is not merely a product of the final outputs of the final state of the algorithm but a kind of historical and procedural window on the chronological process itself."

The collection is the iterative history. Every intermediate state that was previously destroyed is now preserved and released. This is the concrete, structural version of "the construction of an image is the image" — not as a claim about a single work but as a curatorial principle: the series is the process made visible.

The word "gentler" — contrasted with the loss-generating hermetic process — signals that this is also an ethical shift. The old algorithm discarded most of what it produced. The new approach values the whole history, not just the endpoint.

The criterion vs. the code

The text addresses directly the artist's use of Claude Code (Anthropic's AI coding agent) as a collaborator, describing "a feedback loop" with the practice. It frames this as philosophically unsettling:

"If a craft built over thirty years can be approximated, accelerated, even extended by a machine, what exactly was the craft? It is a Copernican displacement: not of the earth from the centre of the universe, but of the artist from the centre of his own practice."

But the question, the text argues, "does not resolve so much as reframe." The resolution:

"What the AI cannot replicate is not the code but the criterion — the specific dissatisfaction that sends Van den Dorpel back to revise, the capacity to recognise when a surprise opens something and when it merely fills space. Every micro-decision continues to pass through a single sensibility. The machine accelerates; it does not adjudicate."

This is the most precise distinction yet between what the human brings and what the system does in a generative/AI-assisted practice. The AI can write code — functional, even elegant code. What it cannot do is evaluate: it cannot feel the specific dissatisfaction with a configuration, cannot sense the difference between a surprise that opens something and one that merely fills space.

This distinction has a philosophical name: criterion. A criterion is the standard by which something is judged adequate or inadequate. The text claims that criterion is what thirty years of practice accumulates — not technical skill, which can be approximated, but the capacity for a specific kind of evaluation that is inseparable from a particular sensibility. The machine accelerates the execution of code; it cannot replace the sensibility that sends the work back to revision.

This connects to subconscious computation: the algorithm follows the logic of taste that has been encoded in the system, but it cannot be the taste itself. Taste is the criterion. The system is its externalization.

Leaky abstraction — the concept as work title

The work Leaky Abstraction (2025, Cloud Writings exhibition) performs the central claim of this page as a title. In software engineering, a leaky abstraction (Joel Spolsky, 2002) is an abstraction that exposes the underlying implementation details it was designed to hide: "All non-trivial abstractions, to some degree, are leaky." TCP/IP abstracts unreliable networks into reliable transmission — but when the network drops packets, the abstraction leaks; the underlying unreliability shows through.

As a plotter drawing, Leaky Abstraction literalises this: the algorithmic logic the drawing was generated from is not hidden but visible in the plotter lines — the implementation shows. The drawing is a leaky abstraction: it presents an image (the abstraction) whose generative logic (the implementation) cannot be fully concealed by the surface. Every line carries the evidence of the path that drew it.

The work is accompanied by Leaky Abstraction (rag) — the same underlying algorithm on a rougher paper surface. The abstraction leaks differently depending on the substrate: the materiality of the support is another implementation detail that the surface cannot fully hide. Two material states of the same conceptual leak.

This gives a software engineering name to what the practice has always enacted: the refusal to let the mechanism disappear behind the image. Where AI-generated imagery is the perfection of the non-leaky abstraction (seamless output with no visible implementation), the plotter drawing is structurally, constitutively leaky. The restriction shows; the seams are exposed; the implementation is legible in every mark.

Tested, not executed

"The initial vision is not executed so much as tested against what the process reveals — each unexpected configuration evaluated, kept or discarded, folding back into an intention that is itself continuously reshaped by what emerges. The artist is both author of the rules and subject to them, simultaneously setting the system in motion and being surprised by where it goes."

This advances from Senescenence's "less composed than initiated" to something even more tentative: the initial vision is not even a fixed intention to be initiated. It is a hypothesis — tested against what the process reveals, revised by what the revision reveals, continuously reshaped. The intention is not prior to the work but co-constituted with it.

"Both author of the rules and subject to them" — this doubled position has appeared throughout the vault without quite being named this clearly. Note 856 observed that "the artist did not fully relinquish artistic control"; note 793 described discovery and trust as the structure of following an impulse. Here the two are held simultaneously: author of the rules (control) and subject to them (being ruled). The artist is not above the system but inside it, surprised by what its own rules produce.

This echoes 952 ("Script"): "the destination emerges from our move. It wasn't there before and we weren't there." The move (setting the system in motion) produces a destination (the emerging configuration) that then becomes the new departure point. The process is recursive: what emerges folds back into intention.

The loom and thirty years

"Van den Dorpel has been doing it since the late 1990s — thirty years through which an almost physical intuition for how logic behaves has accumulated."

The craft of coding is described as "as repetitive and meditative as weaving." The Anni Albers reference, running through the exhibition, is grounded here not aesthetically but structurally: "both weaver and programmer build form from the patient application of local rules, and in both practices the method is inseparable from the result. You can read the logic of the loom in the cloth. You can read the logic of the algorithm in the image."

The thirty years matter philosophically. The criterion is not innate — it is accumulated through repetition, through failure and revision, through the specific dissatisfactions that arise from actual engagement with the material. This is what 609 ("Undo") describes as "workmanship of uncertainty" — not carelessness but disciplined engagement with the unpredictable, built up over time into something that functions as intuition.

The Copernican displacement (the AI approximating what took thirty years) is unsettling precisely because it challenges the relation between time-in-practice and capacity. But the text's reframe holds: what was accumulated is not the code but the criterion. The code can be approximated; the criterion cannot, because it is inseparable from a specific history of specific dissatisfactions.

Herbert W. Franke: beauty vs. aging

The text compares the plotter drawings to Herbert W. Franke (1927–2022), who "hand-programmed algorithms rendered by drawing machines and plotters, and in the 1990s produced his own Cellular Automata series after encountering Wolfram's research."

The formal parallel is direct. But the distinction is philosophically sharp:

"Where Franke sought to make the beauty of mathematics visible, Van den Dorpel is interested in its aging: the moment a system reaches its limit, and what comes after."

Two orientations toward mathematical/algorithmic form:

This connects to senescence as theme: the work is interested not in the optimal form but in what happens when a form has run its course. The equilibrium is not the destination — it is the threshold that triggers reconfiguration.

Against seamlessness

"In an era when the generated image has become a commodity produced at scale without visible process or traceable decision, the insistence of Van den Dorpel's systems on showing their working — on leaving the rule legible in the result, the restriction visible in the form — constitutes a position. The work does not optimise toward seamlessness. It ages, visibly, according to laws it does not conceal. And in that transparency lies its claim: that how something is made is not separate from what it means."

AI-generated imagery is the perfection of ars celare artem at industrial scale: "infinite images, zero visible labor, no readable logic, no traceable decision. The output is seamless in the most literal sense: it has no history you can read." Where the modernist tradition from Impressionism through Conceptualism systematically dismantled the doctrine of concealment, AI reinstates it automatically and at scale. The seamlessness that Cézanne refused, that Cubism exposed, that Pollock made impossible — AI produces it as a default condition of the medium.

Against this, the cellular automaton ages visibly according to laws it does not conceal. The reveal is not a dramatic exposure — "it is continuous, structural, built into the logic from the beginning." There was no concealment, so there is no moment of exposure. The work is honest at every step, not honest at the end.

This is a position on the current discourse about AI and art. The question is not whether AI can make beautiful images (it demonstrably can) but whether those images carry their making within them. They do not — seamlessness is precisely the erasure of making. Showing the working, aging visibly, is the counter-practice. The closing formulation — "how something is made is not separate from what it means, and a visible restriction is not a flaw but a form of honesty" — is the strongest statement yet of what was already implicit in 992 and 1301.

See also