On the Artist and Long-Form Generative Art

In March 2022, artist Shilly Preston found himself in the middle of a dispute. The issue had to do with Preston’s Royal Fidenza, an edition of ten that used AI to mash together Xer0x’s Royalty (2022) and Tyler Hobbs’s Fidenza (2021). To lift his subsequent suspension from objkt.com, where the artist had published and listed his work for sale, Preston was required to recoup from his collectors all of the Royal Fidenzas sold up to that point and to have them burned. Having succeeded in reclaiming all but one in the short window between April 1 and 5, he reminted the project on April 6 under a different title, returning works to all but one of his collectors, Xer0x, who still holds one of the original versions. 

The problem was a legal one. Xer0x’s Royalty was, appropriately, royalty-free, with its free distribution secured through a Creative Commons license (CC0). The artist explicitly encouraged the community to create derivatives from it, in order to show how CC0 works can accrue value. Hobbs’s Fidenza, on the other hand, is not only not CC0-licensed, it’s also further protected by a series of trademarks associated with NFTs that feature graphic content. 

Hobbs subsequently issued an infringement notice against Shilly Preston on objkt.com and the platform took action — the dispute centering on the word “Fidenza” and the visual content corresponding to it. Hobbs has since outlined in detail his thoughts on the issue of fair use, but the consequence of the initial flashpoint has been the birth of the #fidenza4fidenza movement “in support of remix culture and the freedom of expression and interpretation,” fuelled by a wider CC0 initiative in the NFT space led by digital artists such as ROBNESS and Xer0x. 

Rene Magritte’s The Treachery of Images (1929) famously highlighted the fragmented meaning of art by presenting a literal image of a pipe accompanied by a text reading “Ceci n’est pas une pipe” (“This is not a pipe”). The artist challenged how images, objects, words, and concepts all compete with one another in the process of signification. Fidenzas are as treacherous as painted pipes. So what does the word denote? A city in Italy? Hobbs’s code? That code’s output? All of the above? Either way, Preston’s work extends the historical debate to “long-form” generative art (a term coined by Hobbs) and the social and philosophical puzzles that arise from it. 

The questions “what is art?” and “how does art become what it is?” are ontological questions. The case of Shilly Preston’s Royal Fidenza raises similar questions about the ontology of long-form generative art on the blockchain, and about the law itself. It also revives old tensions between the creative process and the ways in which it is defended. Whether Hobbs is the author of Fidenza is a different question to whether he can make a legal claim over the corpus of visual representations that we identify as Fidenza.

Generative art on the blockchain destabilizes traditional notions of authorship, reviving another question forever associated with interactive art: “Who made it?”¹

The modern (and romantic) concepts of artist and authorship have long dictated how information about art is delivered to us and, in turn, how we think about art. Yet new technologies and theories constantly contest the relevance and legibility of museum labels and written descriptions of digital art. Take, for instance, a description of a digital work on a platform like Art Blocks. As in a museum context, the artist’s name is shown adjacent to the work’s title, as is normal in the formatting of image captions. But this seemingly mundane act of labeling also alerts us to a conceptual problem.

The ways in which art historians and spectators have been conditioned to consume information about works of art is evident from the case of the inscription “GISLEBERTUS HOC FECIT” (“Gislebertus made this”) on the Romanesque Cathedral of St. Lazare in Autun, France. Gislebertus was long touted as “one of the greatest sculptors of Western Europe.”² However, in Legends in Limestone (1999), Linda Seidel reminds us that “in the Middle Ages, names meant other things and were employed in other ways.” Present-day viewers might read the inscription on the cathedral as an artist’s signature. Instead, the inscription may have designated the patron. In the Middle Ages, the making of this often meant something else.

Art history’s canonical textbook, Gardner’s Art Through the Ages, first published in 1926, lists several core questions for art historians, one of which is “who made it?” The issue of authorship has been central to other canonical art histories, notably that of Ernst Gombrich, for whom “there is no such thing as Art. There are only artists.”³ This was a response to Heinrich Wolfflin’s 19th-century doctrine of “art history without names,” whereby artistic styles and movements superseded individuals.⁴ Gombrich chose instead to carve out a theory of art that took account of the intent and perception of both creator and spectator. Shortly after, in his famous opening to Painting and Experience in Fifteenth Century Italy (1972), another art historian Michael Baxandall proclaimed:

A fifteenth-century painting is a deposit of a social relationship.⁵

Baxandall’s stress on the participation of patrons in the genesis of Renaissance works of art produced a model of art history in which style and social history were intermingled. The contractual relationship between artists and patrons dates back long before the Artist’s Reserved Rights Transfer and Sale Agreement (1971), not to mention the smart contracts of the blockchain. Works of art in the Renaissance were negotiated through legal contracts that often stipulated what the artist should paint, as well as the materials they would use, and how they should deploy them. 

A 1488 document between Domenico Ghirlandaio and Fra Bernardo, the Prior of the Ospedale degli Innocenti in Florence, denotes the terms for a painting of The Adoration of the Magi, whereby the artist was “to colour [sic] and paint the said panel all with his own hand in the manner shown in a drawing on paper […] in every particular according to what I, Fra Bernardo, think best [...] and he must colour the panel at his own expense […] and the blue must be ultramarine of the value about four florins the ounce…”⁶ Long-form generative art is also socially mediated, with collectors having agency over visual outputs in new and distinct ways.

Traditional notions of authorship came under attack in the 1960s from philosophers like Jacques Derrida and Roland Barthes, with the latter proclaiming in 1967 that “the birth of the reader must be at the cost of the death of the author.”⁷ Under these conditions, the reader becomes an active participant in the enactment of a written text. The effect of deconstructionism and post-structuralism was thus to unshackle the work of art from all dogmatic meanings intended by their creator. Barthes proposed that we reconceive the author as a scripteur, one who constructs a range of parameters within which the reader can operate and imbue meaning.

Appropriation subsequently became a central strategy of artists in the 1970s and ’80s, perhaps most notably the “Pictures Generation.” Works produced by these artists sought to lay bare “culturally coded, and consciously crafted artifacts — as pictures.”⁸ Richard Prince thus rephotographed Marlboro cigarette advertisements; Cindy Sherman reimagined herself as B-movie stars; and in her series, “After Walker Evans” (1981) Sherrie Levine rephotographed a catalogue of Depression-era photography in what the Met has labeled “a landmark of postmodernism, both praised and attacked as a feminist hijacking of patriarchal authority, a critique of the commodification of art, and an elegy on the death of modernism.” Many of these artists faced allegations of copyright infringement, not unlike Shilly Preston and his Royal Fidenzas. While generative art, and long-form generative art in particular, forms part of a complex history of appropriation and authorial intent, it also raises further questions about the process of creation.

Technical process is always central to debates over copyright, as it is to the situation surrounding Tyler Hobbs. Indeed, Hobbs himself has taken pains to document his technical process as a way of offering his audience, and fellow artists, some context to his practice. In a recent essay, collector and theorist Peter Baumann elaborated Hobbs’s notion of “long-form” generative art into an entirely new paradigm of “Art3,” which he views as emerging in parallel to the blockchain. 

For Baumann, new forms of artist-collector collaboration are presently redefining once again the question of authorship in art: 

In Art3 the users — artists, collectors, and platforms — own the art together.

‍This is born out by the process, whereby an artist uploads an algorithm, a collector mints an NFT, and, on receipt of a (random) number, the number is plugged into an algorithm. The algorithm subsequently produces a visual representation, as well as an interesting parallel between generative art and Barthes’s dead author. Indeed, the notion of the scripteur becomes even more charged when we think of generative artists as writers of [computer] “scripts.” For Hobbs, long-form generative art encourages artists “to create a special class of artistic algorithm” that cedes a great deal of control both to a technological system and to its collectors.

‍

Generative art, by all accounts, involves a cession of control to an autonomous system. In one of the most cited definitions, art theorist Philip Galanter declares that:

Generative art refers to any art practice in which the artist cedes control to a system with functional autonomy that contributes to, or results in, a completed work of art. Systems may include natural language instructions, biological or chemical processes, computer programs, machines, self‐organizing materials, mathematical operations, and other procedural inventions.⁹

Galanter’s definition seeks to encompass a wide field of generative art that doesn’t limit itself to the purely digital. However, in From Fingers to Digits (2019), Margaret Boden and Ernest Edmonds establish a more complex definition, proposing a taxonomy of 14 generative art categories, including Interactive Art and, separately, Computer-Interactive Art. According to their definition, for an artwork to qualify as “interactive,” “the form or content of the artwork is significantly affected by the behavior of the audience.” For it to be Computer-Interactive, “the form or content of some CG [computer-generated] artwork is significantly affected by the behavior of the audience.” The latter, they say, represents a distinct form of generative art because “the artist handed over control of the final form of the artwork to the computer interacting with some other human being.”

Interactions come in all forms and shapes. But decisions not to interact equally shape the scope of visual representations, which we might term the “visual corpus.” Projects that aren’t fully minted (for instance, selling 308/750) don’t meet the range of possibilities that would have otherwise been made visually available, but nonetheless impact the creation of the visual corpus. Such was the case with generative artist Aaron Penne’s Within/Without (2022), whose minting closed after the secondary market price fell below the minting price. Penne’s action imbued his collection with more scarcity than originally planned, but at the expense of the visual scope of the corpus.

Penne’s project further probes the possibilities of appropriation and remixing. Within/Without produces three coexistent visual representations, each bearing an instructional relationship to one another. The basic glyph structures take their cue from Larva Labs’s Autoglyphs (2019) but are overlaid with instructions intended to create visually distinct images. Penne’s work guides the viewer through the making process, exposing the “primitive forms” on which his code operates. The subsequent outputs are animated and complex visual representations, complicating the relationship between the initial code and the final output in a way that defies easy (legal) classification. This ambiguity is made explicit in Penne’s work.

In digital artist Kevin McCoy’s definition of Quantum (2014) — itself a work of “short-form” generative art — he states that: “This artwork is the data that returns a value of d41b8540cbacdf1467cdc5d17316dcb672c8b43235fa16cde98e79825b68709a from a SHA256 hash function.” McCoy’s answer is more complicated than it may seem, since what he deems to be the work of art may not necessarily be that work’s visual representation. With long-form generative art, the coherence of the work is even less clear. For Aaron Penne, the “artwork” is both the code and the output at the same time, though they are not necessarily one and the same.

Penne’s hybrid constructs sit well within philosopher Bruno Latour’s socio-technical systems. Latour’s idea of “technical mediation” stipulates that agency is distributed among various actors. In his example, a person and a gun — in themselves two separate things — combine to form a new thing.¹⁰ Latour highlights how relationships between humans and technologies create complex, hybrid constructs that are often inseparable. 

Complex and hybrid structures run throughout long-form generative art. On the blockchain, smart contracts negotiate and intertwine sets of interactions that in turn produce a new unique hybrid construct. Unlike with copies of discrete artworks, both digital and physical, long-form generative artworks do not have a prototypical image. This is exemplified by the endlessly mutating cover-photos of each generative art project on Art Blocks that serve to identify different mints. 

Like other generative art projects, Fidenza is an aggregate of many singular occurrences (mints) produced by the interaction of collectors with its code, and mediated through a smart contract. Out of these processes emerges a new kind of thing that we can define as a distributed (and decentralized) assemblage. 

In his essay on long-form generative art, Hobbs notes that “nobody, including the collector, the platform, or the artist, knows precisely what will be generated when the script is run, so the full range of outputs is a surprise to everyone.” Under these circumstances, the corpus of work, even if scripted to produce a very narrow range of outputs, is ultimately left to the community to enact. The work on the blockchain is neither the representation, nor the code alone, it is the assemblage that they create. Moreover, it is a corpus produced by a community within a set level of tolerance by the writer of the code. This isn’t to question the degree of agency that the code writer had in the creation of the corpus — on the contrary, their role was greater than that of the community by several folds. However, the corpus would not exist without the intervention of that community, as small as it might be, and as such, embroils itself in a larger socio-technical assemblage that challenges traditional notions of authorship.

While I don’t wish to call into question the validity of Hobbs’s trademarks with the US Patent and Trademark Office, I would like for us to look differently at how these works are labeled and delivered to us, and to think critically about the idea of authorship in relation to certain classes of generative art. The next time we look at the label of a long-form generative artwork, let’s question “who made it?” and envision a wide network of actors and agents responsible for its creation. A lot is at stake in this debate, the least of which is Shilly Preston’s suspension from objkt.com.