Tyler Hobbs on Algorithmic Aesthetics

Using the essays of Tyler Hobbs as a reference point, Peter Bauman (Monk Antony) spoke to Hobbs about coded generative art's aesthetics.‍
About the Author
Tyler Hobbs, Thin Air, 2020. Courtesy of the artist

Tyler Hobbs on Algorithmic Aesthetics

Using the essays of Tyler Hobbs as a reference point, Peter Bauman (Monk Antony) spoke to Hobbs about coded generative art's aesthetics.

Peter Bauman: I’ll start with a foundational question: Where does the art mainly lie in your work? Is it the results? Is it the process? Is it something else? In my Framework article, I characterized you as a “Resultist” but that was my designation. How would you describe yourself?

Tyler Hobbs: Yeah, it's a wonderful question. It's one I've been thinking about recently for other reasons, for other writing that I'm doing as well. At the time, based on the bits you quoted there, I think it's definitely fair to say I was a Resultist and in many ways I still am.

This is the way I like to precisely define where the art is now: it's the potential output space of the generative procedure combined with the curation model.

That is to say, I think what the artist has really crafted is that output space and all the things that can happen in it. Obviously, they don't necessarily have control over exactly what happens but to a certain extent they do because they're also able to choose the curation model. For example, with Fidenza, I was developing it with the knowledge that every single output could be part of the final result set. That potential output space was designed with that curation model in mind—versus QQL, where I know that it's going to undergo heavy curation, sometimes really extreme curation. The algorithm can have a completely different type of output space and be successful and to ignore the curation model would be an oversight. So I think that's where the art really lies but I think the results are probably the best way to judge the success of those other things—the tangible, observable effects. I think the art has already been, in some ways, created before those final outputs are even seen.

Tyler Hobbs, Loxodography 0.26, 2019. Courtesy of the artist

Peter Bauman: I'm curious how your views have evolved on this. Has it occurred over time or is it something you've recently reconsidered?

Tyler Hobbs: It's definitely evolved a lot over the years. Starting out as an artist, my algorithms just didn't have a lot of complexity. I was really just looking for one output and not trying to think so much about what defines the art form and things like that early on anyway. Early on, I did start to think about algorithms that are interesting for multiple outputs and trying to get a wider range of outputs. I think getting to real, proper long-form work is where this started to be a more interesting question. With Fidenza and the other long-form projects that came out at that time, you're really looking at the full output space and you're seeing enough examples from it that it can be a sophisticated thing. If you're just showing a handful of cherry-picked outputs, it's a little bit harder to think about the entire output space. Long-form really brings this question forward. 

When I started working on Incomplete Control, I wanted a much different output space. It's only 100 outputs so it's a more focused project. One of the things I really wanted was for the output space to be more continuous and have fewer discrete separations. Fidenza, for example, has things like spirals and non-spirals or soft shapes and non-soft shapes. There are these distinct divisions in the output space. Incomplete Control is where I started to think a little bit more about the quality of that whole space and how it's connected. Can you move smoothly and continuously from any one output to any other output? Then QQL is obviously where curation started to be a more interesting question. I started to think more about what role that could play and how that would affect the design of the output space. So it was really over the course of years and multiple different projects that my views evolved.

Peter Bauman: You mentioned how this process paralleled the evolution of your technical abilities. In 2018, you wrote that “technical abilities are only important so far as they allow us to create images that speak to us.” Do you still feel the same way? What does your work gain from this emphasis on imagery that speaks to us over these technical abilities?

Tyler Hobbs: Yeah, I would say that I really still agree with that. I don't think that a highly technical process makes the work good on its own or meaningful on its own. Oftentimes, it can be a distraction for the artist, especially with a medium like this, where a lot of us are coming from a technical or engineering background. These are the things that, as engineers, we tend to really love—these fascinating technical procedures and aspects. I strongly feel that when you're making artwork, that's the wrong thing to emphasize and focus on. If you're trying to make effective artwork, technical aspects should take a backseat and just be in service of the work and the concept. The work and concept come first. The more technical skills you have, the more helpful they are in executing those things but you should never put the cart before the horse. I still feel that way and the artwork that's always resonated with me most is the artwork that I can enjoy directly without needing to learn the backstory.

If I'm at a museum, my preferred way to go through it is to basically only look at the artwork. If I'm going to read the card next to the work, at most I'm going to read the artist's name, the title and the year. I like to try to take whatever else I can away from the work just by observing it. Not all work is successful or really approachable in that way but I really like the art that is successful that way.

I always wanted my work to be enjoyable and appreciable, even without knowing the backstory—even if the viewer doesn't know that computers are involved or programming is involved or what generative art is.

Ideally, they don't need to know any of those things and the work is still meaningful to them and impactful. A lot of this, for me, came from Agnes Martin and her writings and attitude toward her own work. She was a big proponent of inspired work. She liked to say that the work needs to be inspired and that you can't force it there by technical or intellectual means. That really resonated with me and I think it's very visible in her work and how effective her work is. That's something that I try to bring into my work as well.

Tyler Hobbs, Fidenza #99, 2021. Courtesy of the artist

Peter Bauman: I love Agnes Martin and how she thought systematically about painting. A Guggenheim exhibition she was in with Frank Stella, Systemic Painting, really celebrated work organized around rigid rules and simple repetition that resisted formal interpretation. Formal interpretation can be quite challenging in generative work as well with its specific characteristics. You mentioned taking away what you can through observation alone. Can you walk me through your observation process or the steps of (formal) analysis but for generative work? It can be when you curate your own outputs or consider the work of others.

Tyler Hobbs: Yeah, it's a good question for sure. For generative work, I want to know a few things right away. How many iterations from that algorithm are being used? What's the model? Is this an open, infinite output? Were these curated? Were these direct from the algorithm? If they are, did the artist take ten, one hundred or one thousand of them? After getting an idea of the size and dimensions of the output space, it's an instinctual response for me in terms of aesthetics. Being so involved in the craft, it's hard for me not to think about things in terms of how artists put together a particular process, procedure or effect. What really determines for me if the work is successful is that more quick instinctual response. There are plenty of works that are highly technical, yet I don't find them particularly effective. We could talk about formal elements but they're so specific to each work and there are exceptions to every rule that I could come up with. 

I think some of the qualities that I'm judging about the output space that almost always apply are the balance of unity and variety. You could obviously have low variety and there's not enough interest in the full set. It needs to be worth it to see all of the images in the output set. Ideally, the work should be compelling enough for you to want to look at every single image in the output. If it's not, then you may have generated too many of them. Again, that's an ideal, as my own work probably falls short of that. I also think about how cohesive the outputs are. Is there a unifying theme, motif or element that really makes these things feel pulled together? Or did the artist achieve variety just by stuffing random things into a bag and calling it an algorithm? I think there's a big difference between those things. Those are some aspects of how I judge the generative qualities of the work. Then there are also more direct, formal things that can be analyzed in terms of color, composition, texture and so on.

Peter Bauman: Can you elaborate on the formal elements you consider, such as balance and emphasis?

Tyler Hobbs: Yeah, absolutely. When I'm working, these formal considerations are always top of mind and different artists will weigh those things differently. A project can completely discard some of those considerations and still be really interesting and successful, having a really strong particular viewpoint. The formal elements of visual artwork are a suggestion or very basic mental model that you could adopt. In almost all of them, though, you can take the standard best advice and do the opposite of it and it'll be just fine too. That one's a little bit harder to pin down.

Tyler Hobbs, Drops III, 2016. Courtesy of the artist

Peter Bauman: Does your approach to formal analysis with coded generative work differ at all from your approach to traditional practices?

Tyler Hobbs: There's a big gulf between generative art and more traditional media, mostly in the sense that with more traditional media, the artist is really focusing on a single output. They're able to craft one particular instance of these elements working together. Generative artists, essentially, by definition, are challenging themselves to abstract these concerns and make them more powerful and flexible to see if they can be adapted to a wider range of conditions. I think that's a strictly harder task. In a lot of ways, I think that generative artists get a little bit of leeway on how polished some of these elements are because it just hits the limits of what you can do at an abstract generative level. I think there are so many other strengths in the generative art form that counterbalance that; for instance, the ability for surprise.

That's something I always take into consideration when I'm evaluating traditional art versus generative art. For traditional art, I have a higher bar for how polished I expect some of those elements to be versus generative work because of the difference in difficulty in achieving that.

Peter Bauman: I see generative aesthetics as encompassing the concerns of traditional aesthetics as well as the concerns of digital aesthetics. That makes it, obviously, a very complex subject. If we think about it in that way, are there any other criteria specific to generative art that impact your assessment of a collection? Do you think we need a special criteria for valuing generative art’s aesthetics or can we just borrow from traditional or digital aesthetics?

Tyler Hobbs: Yeah, excellent question. I'm going to use a more specific term and talk about algorithmic artwork here. I think generative art is a broader category and I think you're primarily talking about algorithmic artwork here. So I'll use that term to be precise. I do think that algorithmic art has its own distinct aesthetics and capabilities, and that you, as a viewer, absolutely should take those into consideration. When looking at the work, it's almost like asking if photos should be judged by the same aesthetic criteria as paintings—quite obviously not. They have their own strengths, weaknesses and limitations in terms of what they can capture. There's also new territory that they can explore that other media cannot. I think we've just scratched the surface of what algorithmic art can show us because it's such a powerful way of working. Yes, I think some of the new doors that are opened revolve around precision, scale and complexity. Algorithmic art has a completely different handle on these things than, say, painting or photography. Absolutely, those should be taken into account.

Peter Bauman: Would you say there is such a thing as a digital aesthetic?

Tyler Hobbs:

Peter Bauman: Those are some of the things—precision, scale, complexity—that we would be missing out on if we were to only consider traditional aesthetics when assessing coded generative art?

Tyler Hobbs: There's that and then there's the inherent qualities and bias of hardware and software that put us in a completely different aesthetic space as well. Everything that comes out of the computer tends to be perfect by default. It takes more information to describe an imperfect object than it does to describe a perfect object. The amount of information needed to describe a rectangle is significantly lower than the amount of information needed to describe a lumpy, hand-drawn oval, for example. Computers are really geared for that perfect representation of things. That's the path of least resistance. That's what comes out by default and is the foundation that everything else is built on.

There are going to be massive effects from that. I don't think we should expect digital art to look and feel the same way as painting, which comes from a world with almost the opposite properties: it's messy and chaotic by default. It's hard to achieve perfection in order. That's a massive aesthetic difference between the two.

Tyler Hobbs, Holmberg 2, 2018. Courtesy of the artist

Peter Bauman: Some artists whose work more directly embraces the digital aesthetic think that painterly aesthetics are working against the computer in some way. I think part of it is like you described: it is so much harder to code something with an organic feel and texture than it is a perfect square. Does this characterization strike you as accurate—that you’re in some way fighting the computer by adopting a painterly aesthetic?

Tyler Hobbs:
There are a few ways to think about this. First, I do think it is a mistake to focus on replicating specific analog media as an end. I don't think that digital work needs to pretend to be something it's not. Striving for realistic recreation of those physical media is somewhat harmful to algorithmic work discovering itself and its own representation. For me, an important distinction is that you can feel that way and still appreciate qualities of the analog world that are not present in the digital world.

I think it's natural, at least for me, to look at the digital landscape and marvel at how different it is from the analog—how full of rich detail the analog world is—and to imagine how much it could change the digital world if we achieved a similar level of richness.

It's probably much more interesting to find new ways to achieve that level of rich detail in the digital world in ways that are more native. And I think we really will get there. I think for a lot of artists, things like paint, pencils and so on are sort of obvious first models for study that they can reach for. You can look at paint and you can observe all these wonderful messy qualities that it has and think this is a great thing to study as a source of richness, as a model of richness. Let me algorithmically play with some of these properties. I think that maybe the best way to do that is to learn how to abstract those qualities rather than mimic them. It's much more difficult to abstract them than to mimic them. That's why we see a number of artists go for the more direct recreation. I've been trying to move away from that for a number of years but I think that's why some artists are in that space.

Peter Bauman: Continuing on with this idea of algorithmic aesthetics, how can we better communicate or demonstrate the advantages that we've been talking about in this conversation of algorithmic aesthetics in generative art to the traditional art world? They tend to have reservations not only about the work itself but also about anything related to NFTs. Can you talk about those challenges?

Tyler Hobbs: The most effective way is to make good artwork. I think work that's good enough speaks for itself. Not everybody wants to see or hear that. Some people are always going to be closed off. They're going to have prejudice that's hard for them to overcome. This is just human nature, unfortunately. If the work is good, if the work is powerful, if the work is meaningful, then it will speak to people. It doesn't need to speak to everybody but over time, it really will find its audience and its connections. I don't think that it's so necessary to consider that in particular. I think artists are better off just focusing on their work and continuing to make it as good as they can.

I do think some of the ways in which art and artists in the space often fall short may be in areas like display. Thinking that it's good enough to put their work on a 16:9 screen, even if that doesn't match the aspect ratio of the artwork, and to just put that screen on a wall and call it a day.

There's a certain laziness that I think the traditional art world doesn't accept in terms of presentation.

I think visual artists historically paid much more attention to presentation, at least for good work; they often paid much more attention and detail to exactly how the work was displayed and those considerations. With video art and new media artwork, often they are very specific. The presentation layer itself is part of the artwork. You read the description of the work and it says, "a twelve x twelve CRT TV and twelve-minute one-channel video" or something like this.

I do think that focusing on improving that presentation layer can go a long way towards improving the quality of the work and how it's received.

Tyler Hobbs, QQL #221, 2022. Courtesy of the artist

Peter Bauman: I love that point and do forgive us because recently we did crop your Fidenza on that cake. [laughs]

Tyler Hobbs: [laughs] Yeah, well, it's a cake.

Peter Bauman: You mentioned this emphasis on visual presentation that may be lacking. That reminds me of one of my favorite essays of yours, “Intellectualism Has Hampered Generative Art,” where you wrote about the balance between intellectualism and visual communication in the medium. You called for “the newer generative artists” to focus on the latter because it was what “really matters.” It's now been five years and there’s been an enormous number of new generative artists in that time. So how do you think they've done?

Tyler Hobbs: I've been really pleasantly surprised. I think the average quality of generative art coming into the world today is significantly higher than it was five years ago. I do think that artists are less focused on the technical aspects and more focused on the visual results, which I'm really pleased to see. Maybe that's because the tooling has just become more accessible and standard. You have things like Processing and p5.js that are good enough for a lot of artists. Fewer artists are needing to develop custom software in order to create their work. There are still artists out there who have a very technical focus and I think that's fine. There's enough space for a range of different artists and attitudes. I think, on the whole, I've been very pleased to see the direction that generative art has gone in the last five years.

Peter Bauman: I imagine when you wrote that, you didn't think you'd be assessing that statement so quickly. The next question is antithetical. We've been talking a lot about emphasizing aesthetics or the importance of visual communication. But many contemporary artists have intentionally moved away from traditional notions of beauty. What are some of the limitations of focusing on aesthetics when our conception of them is constantly changing?

Tyler Hobbs: Yeah, it's a good question. Do certain aesthetics have an expiration date? Yes, they do. I think that's been the case forever and I don't think that's really going to change. It's probably healthy for all artists to accept that their work has a limited time range in which it will speak to an audience. Very few works escape that. I don't think that aesthetics are the only meaningful thing to pursue in artwork. I think it's perfectly valid to pursue conceptual ends, political goals or social goals with your work. All of those are perfectly valid for me. They're not as compelling. I'm more motivated by aesthetics and visual communication and emotional impact in my work. I think every artist makes work for their own reasons and it's almost impossible to make good work for somebody else's reasons. I simply choose to follow my internal motivations, which are more aesthetically focused.

Peter Bauman: It's a criticism I've heard from gallerists, for example, and I think it often at least leads to fruitful dialogue. But, finally, I'd like to ask about generative aesthetics, as it was traditionally known by mid-century Information Theorists and Cyberneticists like Max Bense. You've said in a 2019 talk that “when we try to write rules about artwork, we're trying to make more explicit attempts to describe what it is that we appreciate about artwork.” You expressed doubt at that time that algorithms had the potential to make those determinations, while Bense and even artists like François Morellet argued that they could. That was what they thought of as generative aesthetics—to craft a science of aesthetics. Since 2019, a lot has changed in terms of AI advancements. Have these changed your views at all? Do we just need better AI for some kind of science of aesthetics to become a reality?

Tyler Hobbs: There has been a lot of progress. With generative tooling, particularly, I was able to have a handle on low- to medium-range aesthetic concerns. But I think it's completely ill-equipped to deal with high-level aesthetic concerns and I don't think that it ever will be, even with strong AI, because I think it gets to the point where art is very tied in with humanity and what it means to be human.

It's always so hard to define exactly what art is. But I think art is made by humans, for humans. Maybe that's a reasonable definition or maybe it's not. At the highest level of consideration, everything about what it means to be a human in this day and time plays into how we evaluate artwork. Everything about society—our values, our norms and our traditions—all these things come into play when we look at and evaluate artwork.

For a generative system to be able to do that, it would have to be a human. While the tools may have a better handle on some of those simple formal elements, I don't think that they can have a handle on the highest-level questions of meaning and motivation in artwork.


Tyler Hobbs is a visual artist from Austin, Texas, who works primarily with algorithms, plotters and paint.

Peter Bauman (Monk Antony) is Le Random's Editor-in-Chief.