From the outset this hybrid nature has produced definitional and curatorial ambiguities. These works were often cataloged as works on paper, artist books, or silk screen prints, with little acknowledgment of their computational conception. As a result they remain underrepresented both in museum collections and broader historical accounts of computer art. Whether to define them as computer-based, computer-generated, or simply as works on paper is not just a matter of semantics; it directly affects how they are cataloged, preserved, and interpreted. To address this challenge, museums require critical and methodological approaches capable of identifying, documenting, and contextualizing works that resist conventional categories. This involves providing detailed information on production processes, materials, and techniques. Such strategies allow for a more profound and nuanced understanding of the works themselves and their place in art history.
Within this framework the Time-Based Media Conservation Department at the Stedelijk Museum in Amsterdam conducted research focusing on acquisitions prior to 1980, a period preceding the widespread availability of personal computers. The study revealed the existence of hybrid works that had not been fully recognized in the museum’s official records, and identified the museum’s earliest computer-based acquisitions. Combining tangible materiality with computational conception, these works occupy an intersection between the physical and the digital. From a conservation perspective they are classified solely by their material support, but from theoretical, conceptual, and historical standpoints they are far more complex and occupy a gray area. These findings not only fill gaps in documentation but also encourage a rethinking of what qualifies as computer-based art in museum contexts, suggesting new approaches for cataloging, preservation, and interpretation.
Computer art first emerged in the United States in the early 1960s, with pioneering experiments at Bell Labs and the first public exhibitions of computer-generated pictures in New York in 1965. Around the same time West Germany also became an early center of this new artistic practice thanks to philosopher and semiotician Max Bense’s theoretical work and his curation of the landmark exhibition of works by Frieder Nake and Georg Nees in Stuttgart, which took place in 1965. By contrast computer art appeared several years later in the Netherlands. Apart from a few examples in 1968 and 1969, the medium became more prominent from 1970 onward.
In this broader historical context, the computer-generated works acquired by the Stedelijk Museum before 1980 offer a valuable insight into how this emerging art form was understood and collected in its early years. The discovery of these artworks has prompted a comprehensive rethinking of the genealogy of the museum’s earliest computer-based acquisitions.
The first computer-generated work to enter the collection was Dutch conceptual artist Pieter Laurens Mol’s OKVU komputerkaarten-akkumulatie-assemblage, created in 1968. Acquired in the same year, this small assemblage is composed of used computer punch cards arranged in a grid and mounted between transparent sheets to form a tapestry-like surface that interacts with light and composition. In this sense the work relies on a modular principle, where each part gains meaning through its relation to the others, like pieces of a larger whole. The title includes an invented acronym: OKVU, which stands for Oriëntatie-Kader Voor Universum—in English, “Framework of Orientation for the Universe”—, reflecting Mol’s early interest in systems, acronyms, and the pursuit of cosmic or conceptual order. The work combines the cold, mechanical aesthetics of early computing with a poetic, almost romantic sense of exploration, embodying the 1960s spirit of experimentation with automation and machines as new artistic tools. Although Mol was not primarily an artist who experimented with computers or electronic devices, his background in conceptual art enabled him to integrate computers into a broader artistic practice. Early computer art was often criticized precisely because it lacked aesthetic sensibility, as it was initially created by scientists, engineers, and mathematicians without artistic training. By contrast Mol could combine artistic intention with technological processes. Between 1968 and 1970 he produced four more computer-generated works, with access to IBM computers at the IBM headquarters in Amsterdam and the support of programmers at the software company Volmac.
Fig. 1. Pieter Laurens Mol, OKVU komputerkaarten-akkumulatie-assemblage, 1968. Punch cards, ink on paper. 33.2 x 37.5 cm. Stedelijk Museum Amsterdam 2026. Photo: Peter Tijhuis
The second computer-generated work acquired by the Stedelijk Museum, Komputerstrukturen IV (1969), was created by Peter Struycken, a Dutch artist with a strong background in concrete art. The Stedelijk acquired the work from Galerie Swart in Amsterdam in 1970.[4] Struycken was one of the first Dutch artists to experiment with computers, transitioning from abstract geometric painting to algorithmic exploration. In 1968 he took courses in electronic music with G. M. Koenig and in the ALGOL programming language at Utrecht University’s Institute for Sonology (Instituut voor Sonologie). This led to his first computer drawings the following year; Stan Tempelaars and Greta Vermeulen, part of the institute, assisted him in writing the programs. The computer enabled Struycken to investigate the interplay between chance and control in the creative process by generating countless visual alternatives for arranging colored squares across the picture plane. Komputerstrukturen IV is a computer-generated painting in lacquer on Perspex, produced by programming the composition on a computer with the OSTRC program—the first computer program that Struycken employed for artistic purposes. Written in ALGOL 60 by Tempelaars, the program generated patterns based on numerical instructions that specified the grid size, repetition, and random variation. Rather than designing an image, Struycken designed a set of instructions, allowing the computer to execute them and produce complex visual structures from simple algorithmic rules. The resulting pattern was then manually transferred onto Perspex and painted with lacquer, translating the digital output into a tangible artwork. Through Komputerstrukturen IV Struycken examined the relationship between order and chance, as well as between human intention and machine autonomy. He demonstrated early on that computers could act as creative partners, capable of generating visual systems that transcend traditional compositional boundaries. This work remains a landmark in algorithmic and generative art, combining computational rigor with painterly sensibility.
Fig. 2. Peter Struycken, Komputerstrukturen IV, 1969. Laquer on Perspex, mainframe computer, OSTRC programme. 150 x 150 cm. Stedelijk Museum Amsterdam 2026. Photo: Peter Tijhuis
The next piece, Geluid- en beeldprogramma I (1970) by Struycken, acquired from Galerie Swart in 1972, is an audiovisual work that integrates light and sound in a computer-controlled system. Consisting of a light box with a white, translucent Perspex screen divided into a grid of 256 square compartments, each containing four independent electric lamps, it produces variable combinations of light and dark squares according to a programmable computer sequence. This creates a continuously evolving visual surface, accompanied by a taped sound program featuring sixteen combinations of electronic sounds that are not synchronized with the shifting light patterns. The computer program, designed in collaboration with Stan Tempelaars, chief engineer at the Institute of Sonology in Utrecht, controls both the image and sound sequences, thereby merging algorithmic computation with sensory experience. The piece lasts approximately nine minutes and is an example of an early attempt to combine visual art and electronic sound, exploring automation, controlled randomness, and human-machine interaction.
In 1972 the Stedelijk Museum added five computer-generated artworks to its collection. Two were donated by the American artist Lillian Schwartz; two others—one by the American artist Charles Mattox and one by the German artist Manfred Mohr—were acquired from the art centre in Bussum; and one, created by the Dutch computer scientists Lambert Meertens and Leo Geurts, was purchased from Galerie Swart.
Fig. 3. On the left: Lillian Schwartz, Hippie, 1971. Ink on paper, IBM 7094 mainframe computer, BEFLIX programme, S-C4020 microfilm recorder. 96,6 x 63,8 cm. Stedelijk Museum Amsterdam 2026. Photo: Peter Tijhuis
Lillian Schwartz is a pioneer of computer art and one of the first women to work extensively with computational technologies. From 1969 to 2002 she was a resident at Bell Labs, where she produced films, videos, serigraphs, drawings, and other experimental works. In her early years at Bell Labs, Schwartz employed a working process that translated computer-generated imagery into screen prints. She started with a hand-drawn pencil sketch, then re-created the composition on graph paper, assigning color values to alphanumeric characters as a coded map. These drawings were then converted into punch cards and processed on IBM 7094 mainframe computers using a program, originally developed by Ken Knowlton and Leon Harmon, that rendered the image in gray scale through character-based patterns. The result was then output on a Stromberg-Carlson 4020 microfilm recorder as 35mm film negatives. The microfilm allowed her to verify that the programmed translation aligned with the intended image.[5] From there the negatives were enlarged to produce acetate stencils for screen printing. Thus the computer’s grayscale character renderings were transformed into serigraphs, giving the algorithmic designs a tangible form. This multistage process is exemplified by one of her works in the collection: Hippie (1971), a computer-generated image of a human face. The other piece is Homage to Duchamp—Nude Descending Staircase (1970), which reinterprets Marcel Duchamp’s Nude Descending a Staircase, No. 2 (1912) through computer-generated abstraction and computational processes.[6] Created in collaboration with programmer Robert J. Tatem at Bell Labs, the work was conceived not as a reproduction of Duchamp’s imagery but as an investigation of his underlying concept: the decomposition of movement in a static image. Schwartz used a program originally developed for circuit design, modifying it to draw triangular shapes and to control line thickness. The program’s instructions were stored on magnetic tape and processed by a computer-controlled laser, which etched the design onto black acetate.[7] The resulting image, composed of white linear forms, abstracts the human figure and breaks it down into geometric, almost circuit-like fragments. By repeating and overlapping these triangular patterns, Schwartz explored how movement could be represented algorithmically, thus transforming Duchamp’s concept into a study of dynamic motion generated by electronic computation.
Manfred Mohr’s work in the collection is P-52 “quark lines” (1970). The artist moved from Germany to France for academic reasons and began working with computers in 1968 after discovering that the Paris meteorological institute had acquired a plotter. Recognizing the device’s potential for artistic use, he gained access to it and began experimenting independently. He taught himself programming and at first did not disclose his methods. His work was heavily influenced by Max Bense, as well as the musician Pierre Barbaud. P-52 is one of the earliest examples of an algorithm being used to generate an image. Each piece created during this period of Mohr’s practice—between 1969 and 1972—was produced using a distinct algorithm. For P-52 Mohr used a FORTRAN program and a plotter. Initially the plotter connected random points, then it progressively formed a straight line, and finally it reversed the procedure to connect new random points. This cycle was repeated eight times, generating complex visual structures from simple algorithmic rules. In 1971 Mohr presented some of his works at the first solo exhibition of computer art in a museum, held at the Musée d’Art Moderne in Paris. There he demonstrated a Benson 1284 flatbed plotter to the public; however, due to the size of the computer and the operating temperatures required, he executed these demonstrations in “offline” mode, loading a magnetic tape containing prewritten program instructions into the plotter. P-52 was among the works shown.
The work by Charles Mattox, acquired in the same year, is Untitled (1970). Mattox was an American artist active in murals, film, kinetic sculpture, and computer-generated art. In 1968 he joined the University of New Mexico, where he began to explore the intersection of art and science. Using the university’s IBM 360 mainframe and IBM 1403 line printer, he collaborated with his colleague Richard Williams to develop the ART1, a program capable of generating impact prints on continuous feed paper. He later worked with David Caulkins to create DART1, a program compatible with a CalComp drum plotter, which used perforated computer paper to guide the pen across the surface, enabling the precise generation of visual compositions while reducing the need for advanced programming knowledge. Mattox’s plotter drawings from the early 1970s were produced with DART1 on the CalComp plotter. Untitled belongs to this group of works, although few other details about its creation or presentation have been documented. The piece is composed of fine geometric lines forming a wavelike, flowing shape, combining mathematical precision with a sense of organic movement.
Fig. 4. On the right: Charles Mattox, Untitled, 1970. Ink on paper, Mainframe computer IBM 360, DART1 programme, CalComp Plotter, 36 x 40 cm. Stedelijk Museum Amsterdam 2026. Photo: Peter Tijhuis
The last artwork that was acquired in 1972 is Kristalstructuren (1972), a series of four computer-generated silk screens by Lambert Meertens and Leo Geurts, produced while they were working at the Mathematisch Centrum—now Centrum Wiskunde & Informatica (Center for mathematics and informatics)—in Amsterdam. Inspired by Peter Struycken’s Komputerstrukturen series, Meertens and Geurts generated these prints using ALGOL 60 programs on an Electrologica X8 computer. The project began as an attempt to infer the underlying algorithm of Struycken’s work; however, rather than reconstructing it, they developed their own experimental procedures, capable of generating a broader range of visual possibilities. Each composition explores the interplay between order and chaos: The program begins with random black-and-white grids, onto which order is progressively imposed, simulating a process of crystallization, “where a regular structure grows out of chaos.”[8]
Fig. 5. The first to the right of the corner. Lambert Meertens and Leo Geurts, Kristalstructuren, 1972. Ink on paper, Electrologica X8 mainframe computer, ALGOL 60 programme. 65 x 65 cm each. Stedelijk Museum Amsterdam 2026. Photo: Peter Tijhuis
The last computer-generated piece acquired by the Stedelijk Museum before 1980 is a series by Peter Struycken entitled LINARC 16.07.73, which was purchased from Galerie Swart in Amsterdam in 1973. This set of eight computer-generated drawings was produced using the LNARC program written by Stan Tempelaars. Each drawing in the series is constructed from geometric modules arranged according to computational rules, combining repetition and variation with subtle irregularities to create complex, visually compelling structures. The program generated sequences of shapes and patterns by calculating positions, orientations, and relationships algorithmically, allowing Struycken to explore an extensive range of visual permutations while maintaining a coherent underlying structure. Struycken employed LNARC to create numerous other works between 1972 and 1974, including another series of sixteen drawings produced in 1973, called LINARC 24-10-73, which is also preserved in the Stedelijk collection.
This research demonstrates that, with the exception of Struycken’s Geluid- en beeldprogramma I, the earliest computer-based works in the Stedelijk Museum’s collection are, in fact, computer generated. Although these pieces are materialized through traditional media such as paper, Perspex, or lacquer, they owe their existence to computational processes that have remained largely invisible until now. Their rediscovery not only reveals the extent to which early algorithmic experimentation informed artistic production in the late 1960s and early 1970s but also exposes the conceptual and curatorial blind spots that have long obscured their hybrid nature. Recontextualizing these works as computer generated rather than simply works on paper entails more than a terminological adjustment, it implies a methodological and epistemological shift in how museums approach documentation and classification. Their dual identity must be recognized through practices that articulate both components with equal precision. Detailed records of techniques, materials, programming languages, devices, and hardware are essential for reconstructing their full ontology and enabling informed conservation strategies. This approach enhances institutional transparency and provides future researchers and curators with a richer interpretative framework through which to situate these pieces in the broader histories of art and technology.
Equally importantly, this process enables the reappraisal of lesser-known artists whose pioneering computational experiments have often been overlooked. Presenting their work through this renewed lens, which foregrounds the algorithmic and procedural dimensions together with the physical output, reveals an alternative history of early digital art shaped by individuals operating outside mainstream technological centers. Acknowledging their contributions expands the canon of computer art, demonstrating that computational creativity emerged in diverse contexts and through a variety of hybrid practices that defied conventional artistic taxonomies. Ultimately this study invites to reconsider the genealogy of computer-based acquisitions in institutional collections. By identifying the computational foundations of these early works, and by proposing an expanded descriptive taxonomy that makes the technological basis behind their traditional materiality visible, museums can contribute to a more nuanced, historically grounded understanding of computer art.
The forthcoming exhibition of several of these uncovered works, included in the new display of the 1950–80 collection, Everyday, Someday and Other Stories. . . in Gallery 1.26 from March 2026, will provide an opportunity to apply these interpretive frameworks, highlighting their hybrid ontology and reintroducing underrecognized artists to reaffirm their importance in the early history of digital and computational art.
This article was developed during a research internship at the Time-Based Media Conservation Department of the Stedelijk Museum Amsterdam, as part of a broader initiative focused on the study of computer-generated and computer-based artworks in the museum’s collection. I would like to express my sincere gratitude to the museum for this opportunity and to the entire conservation team. Special thanks to Flaminia Fortunato (Supervisor Time-Based Media Art Collection) for her guidance and support, to Leontine Coelewij (Contemporary Art Curator) for her assistance and collaboration in the research on several works discussed here, and to the artists Peter Struycken and Pieter Laurens Mol for their invaluable insights.
Alessio Tozzi is a PhD student in Visual and Media Studies at IULM University in Milan, with a focus on computer art and the preservation of this type of work. With his research project he aims to develop methods and protocols for archiving and preserving time-based media art, in particular computer-based art, in Italian institutions. He recently completed an internship in the Time-Based Media Conservation Department at the Stedelijk Museum in Amsterdam. Tozzi holds a bachelor’s degree in Cultural Heritage Studies from the University of Pisa and a master’s degree in Art, Valorization Strategies and Markets from IULM University in Milan.
[1] See Herbert W. Franke, Computer Graphics, Computer Art (Phaidon: London, 1971), x.
[2] Magnetic tapes are large reels of tape, enclosed behind glass cases, that work just like audiocassettes.
[3] A plotter is an output device that produces images by moving a pen over a flat surface according to precise coordinates.
[4] The original work was acquired in 1970, whereas the piece currently held in the collection is a copy received in 1976 and produced by the artist himself after the original suffered irreparable damage.
[5] Lillian Schwartz, The Computer Artist’s Handbook: Concepts, Techniques, and Applications (New York: W. W. Norton, 1992), 34.
[6] This type of work, inspired by the great masters of the past, is a practice undertaken by some of the most important computer artists, who started exploring the potential of computer art in depth, often constructing their experiments around the tension between humans and machines and, therefore, attempting to simulate human creativity through algorithms. See Michael Noll’s Ninety Parallel Sinusoids with Linearly Increasing Period (1964), drawing on Bridget Riley’s painting Current (1964), as well as Noll’s Computer Composition with Lines (1965), based on Piet Mondrian’s Composition in Line (1917). See also Frieder Nake’s Hommage à Paul Klee nr. 2 (1965), inspired by the works of the artist Paul Klee, as well as Vera Molnar’s series À la recherche de Paul Klee (1970), a visual and computational reflection evoking Klee’s painting Variations (Progressive Motif) (1927).
[7] Schwartz used the same computer-controlled laser technique to etch images onto a range of materials, including copper, aluminum, brass, and even gold- and silver-plated circuit boards.
[8] Leo Geurts and Lambert Meertens, “Crystallization,” Computers and Automation 19, no. 8 (August 1970): 22.
