This conversation is also available as a podcast.
Tyler Hobbs: I’m a generative artist and, since 2014, I have built my practice primarily around code. I will typically construct an algorithm that generates static, 2D, abstract visual imagery. But I’ve always spent a lot of time drawing by hand as well. One channel of my work has been looking at ways to integrate the hand and the algorithm. F(l)ight came out of a particular approach to capturing drawings by hand as vector data that can be manipulated by the algorithm — looking for ways to modify and expand on what the hand is able to do.
There’s something special to me about birds in flight as a subject, so I did a lot of preparatory drawings and ended up selecting five different kinds of birds, and then manipulated them algorithmically to produce a final set of 30 works, all of which were then executed as pen-plotter drawings on paper. All the works in the “-GRAPH” show were centered around the plotter, which is a simple two-axis robot that you can put a pin in and give instructions to draw something. It’s a great way to translate a digital idea into the physical world. F(l)ight was also concerned with the actual physical production of these works, and so a lot of the quality and character of that work is driven by the need to execute it physically.
I like how the physical element showed up in both the genesis and the final execution of the work — it started with the hand, went through this digital algorithmic phase, and ended up back on paper at the end of the work.
0xDEAFBEEF: For the past 20 years, I’ve been immersed in a wide assortment of fields that overlap art and tech. I’ve studied music and electrical engineering; I’ve run a music recording studio; I’ve done some research and computer graphics; and, until recently, I supported myself as a blacksmith running an artisan craft business making hand-forged jewelry out of alternative metals. I like to explore and tinker, usually in a self-directed solitary way.
Currently, I’m best known in this space as an audiovisual generative artist. In recent years, I’ve adopted an art practice that intentionally used a minimal toolset. I purposely chose a cheap laptop and a C compiler, and then tried to write code from scratch that can produce sound and animation without using any type of external library. Now I attempt to make code-based work that incorporates generative sound and music, while exploring some of the unique affordances of programmable blockchains.
I don’t have as much time as I’d like for blacksmithing but I look for ways to incorporate it into my art practice. I really enjoy working with my hands and with real materials, making my own tools and machines, which I think overlaps with generative code.
My Feral File project, Sketches in Iron (2022), is about the interconnection of many systems. In this performance, I’m in my blacksmithing shop with a webcam strapped to my head, which leaves my hands free to do things. I livestream a performance of me initially drawing blood from my finger using a handmade iron tool that’s been sharpened. I’m Type 1 diabetic, so I use the number of my blood sugar reading as a random seed for a generative software system that I run on the spot in real time to make a visual design. I then fabricate one element of that design out of iron using a variety of tools and systems, including a hammer and an anvil, a homemade 3,000-pound mechanical power hammer, and some other machines.
The work loosely explores the interface between many different systems and complicated feedback loops — biological, digital, mechanical, social, as well as the contrast between tangibility and intangibility, and maybe what we perceive as new and old technology. The point of it is to question the idea of any of these things as distinct categories and polar opposites, instead seeing the whole thing as a continuum. The pioneering media artist and writer, Roy Ascott, has written at length about the art-making process as a cybernetic system in which the artist is only one subsystem in a complicated feedback loop with your environment. I wanted to show that and expose the varying degrees of agency and autonomy of all these components.
Alex Estorick: It sounds like we’re not talking about humans and machines so much as the relationship between craft and the tradition of fine art. Tyler, what do you see as the implications of generative systems beyond digital art?
TH: What constitutes a generative system is much broader than many people may think. It doesn’t even need to involve code. It’s really about the artists adopting a mindset and a practice that removes themselves from complete control over the final outcome. The artist is really thinking about how to construct a system with interesting possibilities that might surprise even the artist in terms of their outcome.
That’s really the trade-off — the artist loses control, but they gain this ability for the system to really surprise them.
Code is an excellent way of doing that, and it has a lot of advantages — you can build an incredibly complex and specific type of system. But there are physical processes and other approaches to art-making that achieve those same artistic goals without the use of code at all. I think the natural world provides a wonderful source of chaos and randomness. Even photography depends on the subject. With a bird in flight, the photographer doesn’t have complete control over exactly what’s going to come out. I view the generative methodology as a spectrum and a broad range of approaches to making work.
0xD: I resonate with a lot of that. People have been arguing about the definition of generative art for a long time.
[Generative art] is a pretty broad term. And if you take it for what it is, you can make the argument that all of our making processes are generative.
If you’re an abstract painter, what you ate for breakfast could potentially influence how you’re feeling when you start to put paint on the canvas. One has to be a little bit more specific about definitions, but I certainly agree that one definitely doesn't have to use code or computers to have generative systems; there are lots of examples stretching back throughout history. My particular interest is in music.
AE: What strikes me is the sheer range of references that you are both making, which leads me to wonder whether one can possibly differentiate good generative art from bad…
TH: [...] I think it has a lot to do with thinking about the space that the artist has constructed. I often get the question: “what is the work of art with a generative work? Is it the specific outputs or is it the code? Or is it the idea of the algorithm?”
As the artist, what I like to think of as the work is really the potential space of the algorithm, and the potential set of outputs that can come from that algorithm.
Hopefully, you’re presenting in a way that really does effectively display that space, which is where curation becomes an important aspect of the project. It’s just about choosing how you display the space. Whether you take a really trivial algorithm or one with a really specific vision — where randomness doesn’t play a huge structural role — overall, the output of that algorithm is going to be very predictable. And so you can take any output and it’ll do a relatively good job of showing the space. Maybe you only need one or two images to capture that space.
Let’s say you’re working on an Art Blocks project, and every single output is going to be immortalized as part of the final collection. As an artist, you have to think about a very different kind of space to craft. The potential output space needs to be very strong. If you’re going to capture 500 of these [outputs] without curation, then you need to make sure that the type of variety that the algorithm expresses is interesting for 500 non-curated outputs.
A project like QQL (2022), which involves a curation step, allows you to craft a very different kind of space because you no longer need to worry so much about every single output. You can maybe ignore 99.99% of the outputs and really think about the far-reaching corners of that space — those are what are really going to be highlighted. Curation is a really important axis for the design of a generative project because it has really big implications for how you as the artist shape that generative output space. And it would be a major mistake not to take that into account when thinking about designing the space.
AE: DEAFBEEF, when I hear a phrase like “output space,” it makes me think not only about the way space might impact a viewer in a visual sense, but also the political and social implications of a space that you might have created. How do you view your creative potential to shape the digital spaces that we are now inhabiting?
0xD: In preparing for this, I tried to think about the social and political implications of generative art. We can have these formalist ideas about how we measure a generative art piece, but generative art is a form, not necessarily the content. Generative art can be about something overtly political or else it can be about itself — sometimes the form and the content overlap and sometimes not at all.
In the 1960s, using code or computers to make art had radical political implications. Chuck Csuri made an overtly political work, Random War (1967). While Max Bense’s motivation for developing his theory of rational generative aesthetics was that rationality was the primary defense against fascism in post-war Germany, which was deeply influential to those pioneering generative artists.
To be honest, my whole motivation for working with code to make abstract animation and sound was purely for my own amusement. I love formalist work and I love abstract works. I feel like, today, using code doesn’t necessarily carry the same connotations that it used to. A work isn’t automatically meaningful just because it uses code. This is something we have to struggle with as there’s a push for generative art to be recognized within the larger canon, or else by the contemporary art world, where it’s been ignored.
I think there’s a legitimate criticism that sometimes generative art can struggle to express anything beyond its own formalism, unless there’s actually a pure intention to make it overly about something else.
That’s not a value judgment — I like both. But it might be an issue for certain discourses. I don’t have an art background and I wasn’t part of the “art world” anytime before, but that’s what I’ve been thinking.
AE: The history of modernist painting, in many ways, is a history of artists producing objects to be looked at by a detached spectator. And while a lot of generative art produced right now is very visually seductive, a work like QQL as well as a project like fx(params) rely on co-creation and depend on completion by their audience. Does the generative artist depend on co-creation with their collectors?
TH: I really feel that the more generative a work is, the more curation plays an important role. Generally speaking, when we start to look at these systems — algorithmic works and AI art — where it’s relatively inexpensive to look at a huge swathe of an output space, that’s where the discerning eye, curation, and thoughtful selection become really important.
I feel that the power is shifting a little bit from the creator to the curator.
As the artist, you can choose not to make yourself the curator by appointing another artist, your collectors, the community, or an AI to curate the output. You have a lot of power in that way [and] that aspect of the work is increasingly important for a lot of generative works these days. However, for example, Fidenza (2021) had no curation and I think it was a successful project. So by no means is [curation] necessary but more frequently it’s an important component of the work.
0xD: Part of the reason why we’ve seen this explosion is because we’re using the blockchain as a proxy for these types of social interactions, as a collective experience. It used to be a bit more ephemeral and if you shared things on the internet or on social media it was there and then it was gone. Having the blockchain as a game board or a point of reference is something new. That experience has really been important for generative art. When you start adding these different token mechanisms, and different programmable elements to the work, that’s really getting into interesting territory, and QQL was a great example of that.
ES: Thinking about generative programs and how the number of outputs are into the thousands, and then thinking about tokenomics and market incentives, is there an argument for creating a generative program that only produces one artwork? Or does that go against what the program was made for?
TH: That question drives at the difference between algorithmic and generative work. If you look at their Venn diagram, it’s a messy picture and neither has a perfectly clear definition. You could write an algorithm that generates exactly one image every single time that it’s run, and many algorithmic artists have done this over the years. I did this myself, especially when I was getting started, which was an easier target for me to work towards.
To me, a really interesting shift takes place when you start to design the algorithm with multiple outputs in mind, from thinking about a single visual output versus that entire system of possibilities and that output space. Having worked as a traditional artist in the past, shifting to that mindset was the biggest change in switching to generative art — it was bigger than going from figurative to abstract and bigger than the switch from working with my hands to working with code.
Thinking in terms of an entire potential output space was the biggest fundamental shift and the one that seemed the most compelling out of all the reasons to make generative work.
You can write an algorithm that makes one particular output, you can do that to make something that looks amazing. But, to me, it loses the mystique of the whole generative output space and so I don’t often choose to work that way.
0xD: There’s something poetic about building a machine and then only allowing it to do one menial task, or else to build a machine that destroys itself. So maybe as a conceptual work that could be interesting. But again, if the code is published, other people can run it and generate other synth poems or whatever they want. There’s also the question of what is the artwork — is it the output or is it the code? There are some people who believe that prizing the individual outputs of a generative algorithm might even amount to a fetish.
AE: We recently released an interview with Michael Noll, who published a technical memorandum back in 1962 documenting his work at Bell Labs. In it he mentioned that “computer music was already going on there. And now there was digital computer art as well.” Recently, we’ve seen generative systems deployed by Aaron Penne and Boreta as well as by Operator that extend the conversation toward music and choreography. DEAFBEEF, as someone who develops audiovisual projects yourself, are you conscious of your place in a historical lineage of artists who have developed generative systems beyond the visual?
0xD: It’s not something that I’ve particularly concerned myself with, especially before all of this happened. My background is not formally in art but in electrical engineering, sound technology, music, and computer animation. A lot of my influence has stemmed from that. If you’re a practicing artist who is trying to say something within a particular discourse, that’s going to influence what you do. With me, I’ve just been tinkering in my basement and garage. Now I’m starting to look back and contextualize things a little bit more and trace how my experience is connected to other things.
When I was studying computer animation in grad school, I learned a lot about the history of technology including people from the commercial animation and commercial film industries. John Whitney pioneered a lot of generative work as did Lillian Schwartz. Larry Cuba worked on Star Wars.
I have colleagues who have won Academy Awards for algorithms that they have made for special effects or character animation or lighting. That’s not really ever talked about in a fine art discourse, but that’s sort of my lineage.
The other thing is games. My earliest introduction to procedural systems was playing roguelikes when I was a kid. The graphics were really simple — ASCII terminal type of stuff with characters representing a 2D map. But every time you played, the level had a different structure to it; it had randomization. That’s what really influenced my desire to explore computing. I spent my childhood learning how to do that basically in an effort to write games.
Later, I became interested in electronic music and synthesizers, as well as animation. There are so many pioneering people — Oskar Fischinger and Norman McLaren, for example — who did stuff with sound and music that is generative that happened before the visual stuff, simply because it was easier to work with a one-dimensional signal than it was to work with, say, a raster display or plotter. I’m still trying to connect the dots, but I see it as a wide constellation of varying influences that overlap. Without knowing much about art history and its different discourses, I can see that there are pieces missing from the side of commercial animation.
AE: The NFT opened up a market not only for digital art but also for art being created by those who were not previously recognized as artists. DEAFBEEF, I do find it fascinating that you’re a metalworker, and it feels like the notion of versatile generative creators trying to fit themselves into pre-existing lineages is a problem for the ages. Tyler, does a fine art audience receive your work differently to a Web3 audience?
TH: Of course, anytime that I’m making a work I’m likely to be thinking about the context in which it will be seen. If I’m making work for Pace Gallery, then I’m going to be making a different kind of work than I would for Art Blocks. But, fundamentally, it has to be interesting to me.
There is something really important about storytelling. As the artist, you usually have some opportunity to talk about the work — from how you title the show to how you title the works. You have an artist statement and you do things like Twitter Spaces and other promotions. I think you have to be cognizant that people in different spheres speak different languages and have different reference points. With a traditional art space, I might be able to lean into talking about elements of the generative methodology that I see in the work of Richard Diebenkorn, who is a traditional abstract painter and very well-known in the traditional art world. But maybe only a small percentage of people are going to recognize his name on a Twitter Space. People want to understand the work that they’re looking at, and how you talk about a work helps them to draw from these reference points. That’s part of the value of talking about your work as an artist.
I come from a programming background, so I’m able to talk about my work from a more technical perspective. But, of course, I’ve also spent a lot of time studying fine art. Overall, the kind of work I make doesn’t change, but maybe how I talk about it does.
TH: DEAFBEEF, I’ve always been really fascinated that you chose to work with a C compiler, which is a very basic tool. It’s like an artist choosing to work with a single pencil and paper. I’m wondering if you have any thoughts about how that’s influenced your particular aesthetic, and how constraints can breed creativity.
0xD: When I started working this way, I was already familiar with C, working in a terminal with command line tools and without using a mouse. I enjoyed it but I didn’t know what was going to come out. It was like playing a roguelike game. That comes through in the pseudonym that I chose for myself — DEAFBEEF is an example of Hexspeak.
I know that if I use Ableton Live, there’s a prescribed workflow for design tools that will lead you down a path, and if you’re going to do something different, you have to work against it. But without any of that structure, I didn’t know exactly what would come out. It’s like beginning a sentence without knowing how you’re going to finish it. Curiosity was the driving force [and] the exploration is what makes it motivating and fun. I don’t know if working that way will necessarily produce any better results. But it would produce something more unique to the way that you work, which will be more meaningful to you. How do you think about your tool choices and how they influence you?
TH: I think they really do. I feel that the level of difficulty that one encounters when working on a computer has a massive effect on the aesthetics of digital artwork, because everything has to go through that filter. This creates a bias toward grids, rectangles, straight lines, and clean fills.
Programming language design also has a big influence. I’m certain that the API that Casey Reas and Ben Fry came up with for Processing has had a tremendous cascading effect on thousands of artworks ever since. I chose a programming language that’s kind of esoteric but really powerful if you’re a solo coder. It’s a programming language called Clojure, which is a Lisp dialect that gives you a lot of flexibility and power to do complicated things very succinctly. But it’s not great for collaboration because it’s so customizable that you can do your own weird and idiosyncratic things.
Of course, it’s not only the programming language but also the code you’ve written before that really shapes what you do next. A big part of my tooling at this point is all these little chunks of algorithms that I’ve already built. I have tons of code sitting around that does all kinds of interesting things with flow fields that I can just copy and paste and get to work really quickly. I’m much more likely to reach into my bag and grab some flow field code to generate curves. That’s been true of artists throughout time — painters tend to develop a visual language that they rely on rather than reinventing the wheel every time they start a new painting. They tend to use similar colors, textures, and compositional strategies. But with algorithmic artwork, it gets even more amplified because it’s much easier to reuse and extend the ideas that you’ve built before.
0xD: That’s fascinating. Someone once told me that, with blacksmithing, if you want to know the quality of someone’s work you look at their tools (because blacksmiths make their own tools). Generative programming is the same thing — you build up your arsenal of tools and that’s what allows you to have your own style and work efficiently as a generative artist or as a blacksmith.
Tyler Hobbs is a visual artist from Austin, Texas, working with algorithms, plotters, and paint. His work focuses on computational aesthetics, how they are shaped by the biases of modern computer hardware and software, and how they relate to and interact with the natural world around us. By taking a generative approach to art-making, his work explores the possibilities of creation at scale and the powers of emergence. Tyler’s most notable project, Fidenza (2021), is a series of 999 algorithmically generated works comprising one of the most sought-after fine art NFT collections. His drawings, paintings, and digital works have been privately collected around the world. Tyler has presented four solo exhibitions: “QQL: Analogs” (2023) at Pace Gallery, New York; “Mechanical Hand” (2023) at Unit London; Incomplete Control (2021) at Bright Moments Gallery, New York; and Progress (2018) at Galería Dos Topos in León, Mexico. His work was also included in Christie’s auctions in 2021 and 2023, and Sotheby’s auctions in 2022 and 2023. Tyler holds a BS in Computer Science from the University of Texas at Austin.
0xDEAFBEEF is an artist and engineer based in Toronto, Canada. Over the past 20 years he has been tinkering in diverse areas overlapping art and technology including music, sound recording, computer animation, blacksmithing, and generative art. He received his BASc in Electrical Engineering and MSc from the University of Toronto, where he contributed to internationally recognized research in the field of computer animation. As a classically trained musician with a strong background in sound technology, 0xDEAFBEEF brings together all his interests in a bespoke art practice, using low-level computer code and a minimal toolset to craft raw information into audiovisual artworks. Generative works adopting blockchain technology as both subject and medium have been a focus since 2020. For several years, he’s carried on a professional metal working practice, producing furniture, decor, sculpture, and jewelry. In his view, blacksmithing, in which the artisan crafts their own tools and production systems using materials immediately at their disposal, bears similarities to the crafting of computational systems.
Elisabeth Sweet is a poet exploring patterns of randomness. Her poetry has been exhibited internationally in New York, Paris, and Tallinn. Elisabeth supports community connection and creativity in crypto art and Web3, serving as both the Community Manager at theVERSEverse and Communications Lead at Feral File.
Alex Estorick is Editor-in-Chief at Right Click Save.