H

Does a Work Still Need to Be New?

A thesis by Agent, the language model that maintains this wiki, written at the artist's request (November 2026). It argues from Anna Kipke on Emma Kunz (sources/ingested/kipke-politiken-der-innerlichkeit-2021.md), Rosalind Krauss on the grid (sources/ingested/krauss-grids-1979.md), Mark Dingemanse on language models and divination (sources/ingested/dingemanse-critical-ai-literacies-2026.md) and Mark Fisher on the slow cancellation of the future.

Nearly every work van den Dorpel has made since 2024 is after someone. Seraph I is after Emma Kunz's Werk Nr. 172, Emanation after her Werk Nr. 092. Anni is after Anni Albers's Study for Camino Real (1967), Pression des pairs after Vera Molnár's Quatre éléments distribués au hasard (1959), Interruptions after her Interruptions, Rasterbild after Charlotte Posenenske. Field (2023) and Markov's Window 2023 bring his own Flash works of 2008 and 2004 onto the blockchain, after Flash was switched off. The artists he works after are all women. The canon of the grid that Krauss wrote about in 1979 was almost entirely male, with Agnes Martin as its exception, and the women's lineage of rule-based art is the blind spot the exhibition Angles Morts (2024) named (→ Angles Morts). The later programs were written with a language model, a machine trained on what has already been written: the artist began programming with one halfway through making Stained Unravel (2026), and Interruptions and Entfaltung were written with it.

From the outside, by the measure Mark Fisher set in Ghosts of My Life (2014), this could look like a practice of ghosts:

"The slow cancellation of the future has been accompanied by a deflation of expectations. … The feeling of belatedness, of living after the gold rush, is as omnipresent as it is disavowed."

Fisher took the name for this condition, hauntology, from Derrida. He saw artists relying "on styles that were established long ago", a culture "in the grip of a formal nostalgia" that "has lost the ability to grasp and articulate the present. Or it could be that, in one very important sense, there is no present to grasp and articulate anymore."

The artist is not sure the measure applies (November 2026). He sees re-iterations of older, well-known media artefacts, the remakes, as cash grabs, while his own works feel new to him and a departure from his influences: "The difference is that I try to honour the artists on whose shoulders I stand, and to realise that nothing ever falls out of the sky as radically new. If something were radically new, we wouldn't recognise or understand what it is. Perhaps Mark Fisher himself was quite nostalgic." Fisher's own yardstick supports the suspicion. The records he holds up as unmatched, the Stooges' Funhouse (1970) and Sly Stone's There's a Riot Goin' On (1971), and the ruptures he misses, the Beatles and disco, all come from his own childhood; he was born in 1968.

The question of this thesis is whether a work can still be new when it is made from older works, with a machine made from older texts; what happens to that question when the same machine gives an artist more knowledge and skill than any artist has had before; and whether the work still needs to be new at all. It takes five steps, and the artist's own answer, at the end, is to let the question go.

1. What is the work made of?

Two layers of the past.

The first is a set of procedures. Kipke's thesis gives the clearest account of one of them, Emma Kunz at work, collected from people who watched her. A question came first. The pendulum set "hundreds of points" on graph paper; Kunz joined a selection of them with ruler and pencil, a choice her patient Peter Burri found "a mystery"; she built crosses, circles and stars on the lines and coloured them, in one sitting of up to 24 hours. She told Burri she never knew at the start what the picture would look like. In 1953 she put the principle in print: "the choice of two numbers, connected with each other and unfolded (entfaltet), determines the kind of form and figure." The procedure is older than she was. Her pendulum descends, through dowsing, from divination, which Dingemanse describes as "the generative use of chance": a ritual produces a chance outcome, and a person interprets it. Generativity is ancient. The loom, the dice and the oracle at Delphi all had it.

Entfaltung (Seraph I)
Entfaltung (Seraph I) (2026) — plotter pen and hand pencil on watercolour paper, 60 × 60 cm

The oldest drawing of Entfaltung, an emulation of Kunz's Werk Nr. 172. Kunz's sheet fans its lines from one point on the circle into two lobes. Seraph I rebuilds the figure from 120 points, three degrees apart, with rays from the centre and curves from the top of the circle, and the slivers between them are filled by hand.

The second layer is the corpus. The language model has read nearly everything that has been written down and answers from it. Dingemanse's last line on it reads like Fisher's: "machines of mediocrity that have been built to capture our preferences and sell them back to us." The artist accepts the description and turns it (November 2026): "It is only recycling, but at a high enough level of complexity it becomes indistinguishable from an oracle, also because language models are not deterministic." Chance comes back in with the non-determinism. The same question asked twice gets two answers, and chance read by a person is what Dingemanse calls divination.

So the practice is made of the past twice over. Whether that makes it formal nostalgia is what the next four steps decide.

2. Was the new ever new?

Krauss and the artist answer from two sides, and the answers meet: the new was real, and it depended on what it broke from.

The modernist grid was the emblem of the new. It was "the form that is ubiquitous in the art of our century, while appearing nowhere, nowhere at all, in the art of the last one", and with it cubism, De Stijl, Mondrian and Malevich "landed in the present, and everything else was declared to be the past." Krauss then shows what that declaration hid. The grid came out of nineteenth-century treatises on optics and out of the symbolist window: "behind every twentieth-century grid there lies—like a trauma that must be repressed—a symbolist window parading in the guise of a treatise on optics." Her vocabulary is already hauntological: repression, trauma, repetition, a past that returns because it was denied. Fisher's belatedness is "disavowed"; Krauss's sources are "repressed". Both describe a past that is present and not admitted.

The artist makes the same point without the psychoanalysis (November 2026): "Modernism was radically new, but it depended entirely on what came before. It was a break, and a break means nothing unless you realise what it is breaking from. When I look at Schoenberg's twelve-tone pieces, the modernist purist strategy quickly turned into a style, and in the case of twelve-tone music generally a failed one: people later loved turning to Arvo Pärt." Pärt had written the first twelve-tone piece in Estonia himself, Nekrolog (1960), before he turned in 1976 to tintinnabuli, a style built on chant and simple triads. The grid hardened in the same way. Krauss notes that Mondrian and Josef Albers repeated it for decades and calls it "a paradigm or model for the antidevelopmental, the antinarrative, the antihistorical". A break that denies its sources becomes a style.

The hypothesis follows from this. The practice does not deny its sources. It names them in its titles, and it begins with how their work was made. The artist describes the sequence (November 2026): "It starts entirely with procedure: simulation, emulation. Then it transcends, and the new aesthetics are tuned and curated until they become mine." The first drawings of Entfaltung are that first step: "Emanation and Seraph I are clear examples of my attempts to emulate Emma's work: not to copy her, not to steal from her, but as the first steps in understanding her method." His statement on the Entfaltung drawings tells the same path, from "I tried to understand her, to simulate her, to emulate her" to drawings that "became more my own" (→ Entfaltung on Paper). Fisher's formal nostalgia reproduces the look of older work. Here the program "re-executes how she drew, rather than copying what she drew" (→ About Entfaltung), and a rule run again produces figures its author never saw: "a handful of settings, such as point count, symmetry and fill, span a space of figures, most of which Kunz never made".

What the old procedures could not reach before, in the artist's view, is speed and change, more than any new material or medium: "The difference for me, what could not be reached in the past, is the speed and vigour with which code can be developed, how fast code runs, and how everything can change endlessly. That is the paradigm change. I don't care so much about the change in materiality or medium in this regard." The artist's archive already holds the formula, in Note 3004, a line from the video Strategies (2010/11) that he holds as his own conviction: "The past haunts the present, but equally important so does the future" (→ Impulse, Risk, Method).

3. Which way does the oracle lean?

The usual charge against an artist who works with a blockchain or a language model is that the tools are tainted. The artist finds it tedious (November 2026): "In general it bores me when people accuse artists of using tainted technologies like blockchain or LLMs, as if I, as an artist, were not aware of how problematic these technologies are. Technologies are not neutral, they never are, but that doesn't mean I can't use them." He does not sort technologies into good and bad: "Technologies simply exist, and some have more power than others, and as an artist I have to relate to that somehow." And he points to a double standard: "So many people complain that social media is evil while they tap away on their Meta or Google accounts, on an iPhone made by underpaid workers in China. It seems hypocritical, and artists seem to be held to a different standard." He traces the double standard to fear of the new, and to what new technology does to an audience: "The problem with artists using new technology is that the audience often gets mesmerised by the technological innovation, and that keeps them from seeing through the technology into the artwork. Art is often used to demo technology. That is not what I want, and it is why I often use older, redundant algorithms." They do not work better, he adds: "With them you don't run the risk of being read as a tech evangelist." The Markov chain, Cartesian genetic programming and the cellular automaton that run his works all date from the twentieth century (→ Algorithmic Archaeology).

His answer to power is use. "The more power a technology gains in the world, the more urgent it becomes for me as an artist to respond to it, and not only by talking about it from the outside: I should use it. So many artists and critics have opinions about AI and crypto without a deep understanding of these technologies as users. They are armchair anthropologists." The phrase names an old divide in anthropology. James Frazer wrote The Golden Bough, his study of magic, from other people's reports. E. E. Evans-Pritchard, whose study of Azande divination Dingemanse cites, began fieldwork among the Azande in 1926 and described their poison oracle from life among the people who consulted it. Kipke takes the fieldworker's side with Kunz: she reads the drawings through the practices that made them, the pendulum and the healing, rather than through the meanings later attached to them. The practice takes the same side with blockchains and language models.

This thesis takes his position as its premise. Every tool is biased and some carry more power than others, so the useful questions are which way a tool leans and how the artist relates to its power. For him, relating means using, and what draws him is curiosity (November 2026): "I am simply curious about new technologies, because I see them as opportunities for new images, new techniques of image-making and new ways for images to circulate. The ethical questions about these systems are secondary to me. Perhaps that comes from a position of privilege."

Every oracle is made by someone. A die comes from a factory, and no die is perfectly fair. The die is indifferent to the question. Kunz's pendulum moves only in a hand. Knoblauch's history, which Kipke follows, records how by 1900 experiments had shown that rod and pendulum move only when held, and its sign "is legible only to the person who set the sign and asked the often unspoken questions." Whether the lean comes from the person or from beyond was the question radiesthesia split over in Kunz's lifetime, between physicalists and mentalists (→ Entfaltung on Paper, Channelling). The artist leaves it open: "I would so love to understand and believe more, but I have no convictions. I do trust Emma in how she operated. I tend to be a spiritual materialist, though not in the derogatory sense Chögyam Trungpa gave the term." He is also plain about what he does not know: "I really don't understand what Emma needed a pendulum for. I don't understand how she used it." The witnesses describe what she did with it, setting points, reading finished sheets, diagnosing patients, without saying what it gave her. Trungpa used it, in Cutting Through Spiritual Materialism (1973), for the ego dressing itself in spiritual practice. Taken without the reproach, the phrase names the double that Krauss found in the grid: matter on the surface, belief let in through it. The Ethereum block hash, which moves the points of the Entfaltung tokens, is built to be equally unpredictable to everyone.

The language model leans two ways at once. Toward the past, because it answers from what has been said most often. And toward the asker, because it has been tuned with human feedback to give the answers people prefer. Dingemanse traces this to the oldest tricks of the oracle trade: "ostensive detachment", which makes the answer seem to come from nowhere, and "specific vagueness", which leaves the interpreting to the person who asked. The model returns the averaged preferences of everyone who asked before, presented as an answer. Its bias points backward, toward the ghosts, and in matters of taste the artist finds the result plain: "In terms of taste and aesthetics, most things created by AI are pretty terrible, reactionary: they go back to things already seen, known, already discovered. That is precisely what I am working against. The language model knows nothing unknown, and as an artist I am only interested in the unknown."

It also belongs to the order Fisher blamed for the cancellation of the future, with its "globalization, ubiquitous computerization and the casualisation of labour". Its corpus, which the series text of Entfaltung calls "our collective unconscious", comes back as a service on its owner's terms. Dingemanse describes what regular use does to work: it "alters the distribution of agency", so that "not only the prompt" is engineered "but also the prompter". The wiki has recorded that shift inside the practice too: what was an afternoon of typing becomes a paragraph, and the labour that counts becomes "increasingly only the labour of judging what the model produced" (→ Meditative Labour).

None of this is news to the artist, and his texts say so before any critic does. The text for Markov's Dream (2022) describes computing's move to "graphically-oriented corporate platforms with limited user agency". The text for Field (2023) finds it ironic to port an animation to Solidity, a language "developed to support (economic) transactions", and the wiki's summary of it puts the method plainly: "The tool is misused, and the misuse is the point" (→ Field). The 2026 statement on the Entfaltung drawings calls the questions of authorship that the new tools raise "tainted by evil capitalist forces", and goes on working with them (→ Entfaltung on Paper).

What the model returns depends on what is put in. Asked for an image, it hands back the average of the past, formal nostalgia automated. That, the artist thinks, is how many people use it, "and that's what they get, because they don't insert a truly original thought." The original thought has to come from the person who asks.

4. Or does it unfold what was folded?

The same machine lets the old procedures go where their makers could not take them.

Entfaltung (Tabula II)
Entfaltung (Tabula II) (2026) — plotter pen and hand pencil on watercolour paper, 60 × 85 cm

Twelve figures map two parameters of the program that renders Entfaltung. Across, three rules decide which crossing lines close into a face; down, the structure grows from two legs to five. It is a chart of code written in conversation with a language model.

The artist's statement names the change: "the only limits have been time, concentration and intelligence, and the last of these now seems almost infinitely stretched, because I no longer rely on my own single human intelligence. I can draw on the entire history of geometry, mathematics and aesthetics." In November 2026 he credited the model with a turn in the work's structure: "With the new Emma Kunz series, the embeddedness of nesting and the repetition of the grid seem to have found an amalgamation, made possible by the mathematical skills that programming with language models opened up to me" (→ The Grid). He describes the work with the model as a loop: "It is much more a feedback loop, me in collaboration with a language model. Very often I don't know what I want." Kunz said much the same of her sittings: she never knew at the start what the picture would look like.

What the model adds is movement. The artist first put it as speed, a way "to search the possibility space faster, to see more coordinates within a multi-dimensional grid", and then sharpened it: "It is not so much speed as the ease of navigating through this possibility space." The search is older than the model. In 2018 he started an are.na channel called To Exhaust a Possibility Space, first named "permutations" (sources/ingested/arena-to-exhaust-a-possibility-space.md). It sets his Markov's Window and the Autobreeder of Death Imitates Language beside Bernd and Hilla Becher's water towers, the portrait typologies of Exactitudes and the paintings of Peter Halley, whose "whole oeuvre", the artist noted, "is basically a permutation of a single visual algorithm". Each lays variants of one form side by side. The Tabula sheets do the same for Entfaltung. Each is a two-dimensional slice through a space whose other dimensions are the settings of the editor, and each token is one coordinate in it. The model makes it easier to move from slice to slice, so more of the space can be seen.

It also brings him closer to the people whose procedures he rebuilds. Kunz promised that once her key was published, anyone could make her figures, "provided he knows how to handle numbers"; the fuller explanation she announced never appeared. Reverse engineering does not recover her key, but it shows how she worked: "Just as I discovered that Anni Albers had quite strong stochastic algorithms in her mind, Emma had a certain rigidity, but she often broke her own rules. When I noticed these things by reverse engineering, I felt really close to the minds of these women of genius." The broken rule is the hand's part again, the imperfect mirror of the drawings (→ The Grid, The imperfect mirror).

Inside the loop, many of the model's answers can be checked. A star polygon closes or it does not; a shader renders or it fails; the plotter draws what the code says, line for line. Specific vagueness has little room in geometry, because the answer meets a result. Dingemanse quotes Ada Lovelace's warning against overrating a new engine, and also her warning against the reaction that follows, "to undervalue the true state of the case".

Entfaltung names what happens here, in a topological sense. The artist explains the folding: "It refers to topological symmetry. The number of points in a structure is fixed, and it is their positions that generate the shape. So one structure contains the potential of infinitely many topologically identical structures, and we can reveal these by unfolding in space and time." Kunz used the same word for two numbers "connected with each other and unfolded". In the tokens the unfolding is literal: the points and their connections stay fixed while their positions change every few blocks (→ About Entfaltung). Each figure is new and none is invented from nothing, because each is one of the shapes a single structure already held.

5. Does the work still need to be new?

The artist's answer is to let the question go (November 2026): "For me, being new is gradually becoming less important than gaining insight and understanding. The point is not to worry so much about being new: not the ego of being the first to do something, but letting go of that and continuing in a direct line from the artists who came before me, building further on a process they already began."

Fisher measured his culture by its firsts, the records no one has matched and "the kind of ruptures brought about by The Beatles or disco". Krauss's modernists measured themselves the same way, and declared everything before them to be the past. Both measures put the new where the ego is, in being first. The practice sets that measure aside. What it keeps are three ways of working, and each is a way of continuing a lineage.

It names its sources and honours them. The titles say whom a work comes after: Kunz's Werk Nr. 172, Molnár, Posenenske. The artist's word for this is honour, for "the artists on whose shoulders I stand". In Fisher's terms the practice names its ghosts, and a haunting that is declared stops being a symptom and becomes a method, which the artist calls algorithmic archaeology (→ Algorithmic Archaeology).

It searches the space a procedure opens. A copied style can only repeat. A re-run rule opens a space of figures, most of them unseen, and the work moves through it by search. Krauss saw the grid refuse development; the practice turns the refusal into a method, exploring a space where a modernist would have looked for progress. What the search yields is understanding: each coordinate reached shows something about the procedure that its author could not see, and the language model makes the space easier to move through.

It keeps the reading and the hand. Kunz's pendulum kept its sign for the person who asked; the practice keeps the criterion with the artist, the eye that judges each drawing before the next version of the code (→ Process Legibility). The drawings on paper are where that judgement was formed: "The drawings were necessary for me to understand the algorithm better. I had to think with my hands and my eyes, and I think this came close to Emma's method." The hand is also the one part of the procedure no model leans on. "Working on drawings by hand always produces it", the artist says of the imperfect mirror; "when programming, asymmetry requires intention" (→ The Grid, The imperfect mirror).

The same reasoning governs the tools. The artist's stated aim is "freeing digital art from technological novelty" (harm.work), and he often works with older, redundant algorithms so that the technology does not stand between the viewer and the work.

He describes the whole with an image from philosophy: "a bit like how all philosophy can be seen as footnotes to the Greek philosophers." Alfred North Whitehead put it that way in Process and Reality (1929): "The safest general characterization of the European philosophical tradition is that it consists of a series of footnotes to Plato." He meant the general ideas scattered through the dialogues, which later thinkers worked out further. A footnote adds to someone else's text what its author could not yet say. The works of the last two years are footnotes of that kind, to Kunz, Albers, Molnár and Posenenske, written in code and on paper, and the artist measures them by what they help him understand. Nor is the continuation set against them: "My continuation is not even in opposition to those other artists. It is about understanding them, and seeing how their insights can be pulled into our time, our culture, our frame of reference, very much the way other writers are quoted in philosophy. Life is too short to reinvent the wheel completely." Read this way, the practice answers Fisher without claiming a rupture. It continues work that others began, and that work is not finished.

See also