A Longer History of Generative Art

To properly understand the importance of NFTs as a means of transacting digital works of art, it is useful to review the origins of generative art and consider how its most radical feature was also the greatest barrier to its wider acceptance by the art world. The meteoric success of NFTs has provided generative art with much needed visibility in the art world, but its history spans more than six decades and its evolution has been largely discrete and separate.

The paradox here is that the underlying concept of the NFT — as a means of assuring non-fungibility and “uniqueness” of digital assets — defies the historical essence of generative art, which has been about grappling with the basic fact that digital information is by its nature fungible, with no identifiable “original”. 

This is baked into the medium and a brief reflection on Pop Art reinforces it. 

Growing up in the 1980s, my introduction to computer imagery came through the Commodore Amiga which was first released in 1985. As part of the marketing campaign, the company commissioned Andy Warhol to participate in the launch of the Amiga A1000 on 23 July, 1985 at New York’s Lincoln Center. Using a video camera linked to a digitizer, he produced a recognizably Warholian rendition of Debbie Harry to demonstrate the new computer’s multimedia and creative potential. Six months later, interviewers from Amiga World magazine found him most enthusiastic about the computer’s potential as an artistic tool and he had some interesting thoughts about its role in visual art:

Interviewer: How do you see this work being displayed? How would you show something that you create on an Amiga to the general public?‍

Andy Warhol: Well, we could get a printout. I could just print this out if we had the printer.‍

Interviewer: Would you see the prints or distribute the disk itself?‍

Andy Warhol: Well [Jean-Michel] Basquiat goes to the xerox machine and puts xerox all over his paintings. So, if we had a printer right here I could do it this way and just sign it as a print [...]‍

Interviewer: So you don’t see any problem? Something you do on the computer can be recreated pixel for pixel, an exact duplicate?‍

Andy Warhol: Well, in prints they are supposed to be exact duplicates.¹

From the outset, both the interviewers and Warhol grasped the special problem of digitally-generated art: it is inherently duplicable. Anything created on a computer — whether an image, sound, animation, or interactive experience — comprises a sequence of code that can be copied exactly. Unlike analog media where the original is inscribed into a physical material and copies are iterative, digital copies lose nothing between generations.²    

In his seminal book, The Reconfigured Eye (1992), William J. Mitchell considered digital images to be “problematic” since they lack negatives or any other unique instance. “An image file may be copied endlessly, and the copy is distinguishable from the original only by its date since there is no loss of quality.” He also pointed out that digital images synthesized from geometric data have no obvious original due to the range of possible renderings of the same data. “Is the rendering procedure really the original”?³

One solution he considered is the idea of “one-stage” and “two-stage” arts, proposed by art theorist Nelson Goodman. Sketching or taking a Polaroid is one stage, for instance, while the production of music (composition plus performance) is two-stage: an act of composition followed by a performance. 

For Mitchell, digital imagery is two-stage: it is first coded then displayed. He adds that “in a two-stage process, the work is often divided among different individuals,” with different people working on different stages in the production of the same piece of art.

Goodman also made a distinction between autographic and allographic systems. A musical score, which is written in musical notation and thus comprehensible to performers, is allographic, whereas a painting, which is produced as a unique image rather than a symbolic system, is autographic. In Goodman’s opinion, not only can a score be copied exactly, but each instance of it may be regarded as “genuine”. The authenticity rests not on its material form, so long as it is accurate to the composer’s work, but in the more intangible aspects of its performance.⁴  

Goodman proposed that allographic art forms are amenable to replication because they consist of sequences of symbols — like musical scores or text — rather than arrangements of concrete structures executed in a one-off painting by an artist. On these terms, any painting made by another hand than that of the purported artist is a forgery. Mitchell, in turn, emphasized that autographic or analog works cannot be reproduced without noise or degradation, whereas allographic and digital works can be copied exactly. Of course, appropriate software is required to produce and view these additional copies. For instance, the layers of a Photoshop image may only be retained as separate layers — and thus amenable to alterations — in a program supporting the Photoshop file format.⁵

As Timothy Binkley explains, whereas analog media carry image and sound through transcription, digital media work by conversion, turning the original information into “formal relationships in abstract structures.”⁶ Thus the computer creates a visual display from a set of instructions, though the viewer always approaches it as an image with a set of inherent assumptions about its existence on or within a flat surface.

For Warhol, the issue for artists working in the digital medium was that its inherent replicability challenged the art market’s operating model of authenticity and scarcity. The challenges of authentication are obvious to anyone who followed the long-running story of the world’s most expensive painting, the Salvator Mundi, and its contested history, rediscovery, and restoration. Warhol was sanguine about the implications for his own digital work:

For Interviewer: Do you think it will have any effect on the value of an “Andy Warhol original”?‍

Andy Warhol: No, it would just be a sketch. Call it a sketch.‍

Interviewer: Do you ever see that as becoming an artwork in its own right?‍

Andy Warhol: Oh yeah.⁷

The mid-1980s marked a step change in computer art. As the Warhol interview implies, it was at this point that relatively inexpensive desktop computers that could handle color and higher resolutions were made widely available to artists and designers. Research by Xerox during the 1970s into human-computer interfaces was famously appropriated by Steve Jobs for the Apple Macintosh, whose convergence of display and printing technology made the computer a much more visual tool than it had been previously. Using a mouse in a graphical environment also came much more naturally than typing command-line instructions. 

At the time, I remained unaware of the earlier generation of computer artists who had honed their skills at a time when computers were room-filling behemoths without real-time interfaces that could only be programmed with punch-cards. Only later, when I read Frank Popper’s Art of the Electronic Age (1993), did I appreciate the longer history of computer art and it became the focus of my PhD study. Long before Warhol unpacked his Amiga, a range of pioneering artists were experimenting with digital systems.

One of the key figures in the first generation of digital artists was a philosopher of aesthetics at Stuttgart University, Max Bense, who in the 1950s defined art as an exchange of information that sets up a feedback between the artist and the viewer. 

Bense proposes a “generative aesthetics” which would explain how aesthetic states are generated.⁸

This led a number of his students to create early works of generative art from 1960 onwards, including Frieder Nake, who observed similarities between generative and conceptual art as Sol LeWitt understood it:

‍The algorithm is the concept in its strictest form of description. Conceptual art usually is “free from the dependence on the skill of the artist as a craftsman”. Some years before Sol LeWitt wrote this in 1967, algorithmic art had already eliminated the skilled craftsman. We see: algorithmic art is the final form of art in times of industrial production. […] “The idea becomes a machine that makes the art.” Exactly this, dear Sol LeWitt, had happened some years before you wrote it. Insofar, algorithmic art is the mother of conceptual art.⁹

Inherent to generative art is the primacy of the concept, or generative code, over the material form of a work. Where Sol LeWitt issued instructions detailing how to execute his Wall Drawings rather than producing the outputs himself, Nake and his contemporaries developed algorithms that allowed computers to generate images via pen plotters and early vector-based image displays. These images typically consist of multiple variations on an initial image, or developments of shapes plotted in sequence, expressing the mutability of the visual form. As Nake remarked:

The greatest idea before its time, in computer art, is its generative approach. The algorithmic artist does nothing that is not generative. At a second look, nothing is so great about this. Once an artist decides to use a computer for his production, he is bound to design a program (an algorithm). Without an algorithm for art, [there is] no algorithmic art.¹⁰

In this way, a device that was primarily aimed at data processing rather than producing visual outputs was turned to artistic ends. But while the number of computers in developed countries increased significantly between 1955 and 1965, they were not accessible to the general public and were mostly to be found in scientific research institutes and larger corporations. The popular fascination with the computer as an “electronic brain” coupled with fear of its Cold War consequences provoked a negative reaction to the earliest works of computer art. 

When Howard Wise, a far-sighted New York gallery owner, held the first US-based exhibition of computer art by A. Michael Noll and Béla Julesz in 1965, none of the works sold. At the time, both were employed as researchers by Bell Labs, whose PR department advised that the pair make no mention of their employer and instead copyright the works in their own names to prevent widespread reproductions. But while Julesz had generated his images as “stimuli for his investigations of binocular vision,” Noll had intended his images as works of art.¹¹

Jasia Reichardt’s groundbreaking 1968 show, “Cybernetic Serendipity,” at London’s Institute of Contemporary Arts was subjected to similar attacks for eliding the distinctions between “art” and “scientific imagery,” giving equal billing to scientists, engineers, and artists. Yet it proved a popular success in terms of visitor numbers.

In my own research, it became clear to me that the first generation of computer artists saw the vast potential of digital imagery, including interactive works of art that responded to the viewer, but were simply held back by the cumbersome equipment of that era.

When the Computer Arts Society (CAS) was founded in 1968 shortly after “Cybernetic Serendipity,” it was a special interest group within the British Computer Society (now the Chartered Institute for IT). But CAS and its accompanying magazine, PAGE, soon became a nexus of global debate about artificial intelligence, state data collection, and the relative importance of artists and computers. Of course, these remain live issues today.

In 1970, Herbert W. Franke, a German artist, scientist, and science-fiction author (often mentioned alongside such luminaries as Stanisław Lem) wrote a seminal book, Computer Graphics — Computer Art. This is an invaluable source, one that was informed by Bense’s approach to art as a type of information exchange. Franke’s felt that the computer could bring about a democratization of art through its generative and duplicable nature. If the creation of aesthetic objects could be systematized, then the expense and rarity of art, in the sense that the art market valued it, would surely be diminished:

The demystification of art is one of the most far-reaching effects of the use of computers in the arts. No sooner is it recognized that the creation of art can be formalized, programmed and subjected to mathematical treatment, than all those secrets that used to enshroud art vanish.¹²

Herbert W. Franke foresaw a time when networked home computers would be able to download and generate new works of digital art to be sent far and wide, providing the greatest number of people with access to generative images. This would then change the relation of art to its audience, rendering it widely available instead of a precious commodity. Indeed, Franke’s translator, Gustav Metzger, was one of the central advocates of direct action, whose own “auto-destructive” works sought to comment on the destructive nature of industry. In 1974, he also called for an art strike to undermine the production of art.

Works from the first phase of computer art were often printed out for exhibitions, and occasionally in more traditional galleries. Some artists sold them on the market as prints, sometimes playing down their origin as digital images. But with the gradual adoption of computers across the creative industries from the 1980s there was a greater acceptance of works of digital creativity, which nonetheless continued to be exhibited and sold in physical form (other than by artists who made animations and interactive structures not amenable to printing).

The rise of the Internet was the next disruptive event, which led to the rise of net art along with the first online galleries and browser-based exhibitions. 

As with the first wave of generative art, many net artists were firmly against the role of the traditional art market. 

By the early 2000s, most traditional art galleries had developed a Web presence, with art sales slowly moving online. But as ever, the tendency was for computer-based art to follow the same channels as mainstream contemporary art, crystallizing as objects that could be sold and resold, including sometimes as computer systems with screens. 

Museums and galleries have since become more accommodating of electronics, especially since the rise of video and TV art in the late 1970s onwards. But back when Howard Wise exhibited “TV as a Creative Medium” in 1969, a screen in a gallery was sacrilege. The rise of plasma and then LCD (liquid-crystal display) screens and video projectors hastened the acceptance of screen-based art. Nevertheless, today’s ubiquitous flat screens are not suitable for exhibiting certain historic works of video art like those of Nam June Paik, which require the physical manipulation of a CRT (cathode-ray tube) TV set that is increasingly rare.

Ultimately, computer art ended up becoming non-fungible by default, pulled by the gravity of the mainstream art world where it had always sought to play a role despite its technological stigma. 

What changed with the advent of NFTs was the emphasis on the digital, which wrong-footed traditional guardians of artistic value. Many of the artists who continue to benefit from the convergence of blockchain technology and financial reward do not originate from the familiar community of computer artists but rather constitute a growing, and global, community of digital outsiders whose moment has arrived. 

Nevertheless, as should now be clear, born-digital art did not begin with the NFT, indeed Beeple’s Everydays: the First 5000 Days (2021) sits within a history of collages that use high resolution digital files to produce super-images from multiple sources. Likewise, the most expensive NFT of all time, Pak’s Merge (2020-21), utilizes its own token-based sales ecosystem to generate more “mass,” aggregating itself via a gamified reward system whereby the biggest collectors compete to own more. It is hard to conceive of an analog equivalent.

Right now, the more complex digital art takes full advantage of the NFT as a generative medium that can both be seen and purchased in its own environment.

Although the inherently commercial nature of the NFT might seem anathema to the democratizing impulses of early computer artists, in fact it fulfills Herbert W. Franke’s “demystification of art.” It is also notable that Franke remained active in this field right until the end of his long life, and was able to make his own NFTs before he passed away. This would seem to indicate his approval of the NFT as a vehicle for computer art, and suggests that his intuition regarding the digital medium back in 1970 was very much on the right track.