Tyler Hobbs on QQL and the Future of Generative Art

Alex Estorick: Tyler, what does “QQL” stand for?

‍Tyler Hobbs: QQL (2022) doesn’t actually stand for anything. I tend to select names for my work that don’t impart a clear meaning. I want the viewer to approach the work initially with a neutral mindset and for the work to establish itself visually. The approach that Dandelion and I took this time for QQL was simply to pick a name that had three letters that sounded good, without assigning a particular meaning to them. We actually had a little generative workshop to create the name that involved a bit of programming as well as curating. QQL was the output we liked the best.

‍AE: You’ve described the project as “an experiment in generative collaboration.” How has your experience of Art Blocks prepared you for this project, and what did you want to do differently this time?

‍TH: This project absolutely would not have happened without Art Blocks and Fidenza (2021) and seeing that model play out so thoroughly. This project in some ways goes back to pre-Art Blocks generative art and in other ways it moves forward. Before Art Blocks, most generative artists, including myself, would have a curation step in their artistic process. So once the algorithm was complete, I would generate hundreds of images, curate a handful of my favorites, and turn those into finished works via unique prints, for example.

With Art Blocks there is no curation step — every single output ends up in the hands of the collectors. And that really places some strong constraints on the kind of algorithm you can craft because the minimum standard of quality for the outputs has to be extremely high. That normally prevents you from taking certain risks with the algorithm.

What was attractive about adding curation back into the equation was that it would allow for a riskier algorithm. Both Dandelion and I were interested to see what a riskier long-form algorithm might look like.  

That also tied in really nicely with the involvement of the community because, with QQL, it’s not me or Dandelion curating the output, it’s the entire collective community. That also dramatically shapes what kinds of outputs get selected, how many outputs get viewed, and the entire approach to the artwork. It’s a very different model. But without Fidenza and Art Blocks I wouldn’t even be thinking along these lines — QQL grows out of that style of work.

‍AE: The community has been talking a lot this year about the craft of generative art and artists’ preferred techniques. How does the QQL algorithm differ from that of Fidenza? ‍

‍TH: So QQL and Fidenza share one core component, which is an underlying flow field that structures most of the image. This isn’t surprising as flow fields have been a really core part of my work for many years. Where QQL differs is in how the structure is established. Fidenza tends to be much more scattered and randomized. It’s a little more uniform in how the components can be distributed around the image. But QQL takes a lot of care to establish larger structures that interact with the flow fields in various ways that tend to have a lot of emergent possibility. 

QQL is designed entirely around supporting emergence and really encouraging it, and so some of the choices in the algorithm provide a lot of flexibility, but they tend to create larger-scale structures that are mostly absent from something like Fidenza.

‍AE: That reminds me of a recent conversation with Iskra Velitchkova, Sofia Crespo, and William Mapan whose work often seems to tread the borderline between non-human machine intelligence and more-than-human natural systems. You also speak of emergence in a way that joins the machinic with the organic.

‍TH: The relationship between the natural world and the digital world has been a core focus of my work. Throughout my time as a generative artist, it’s the most fascinating element of the work because the computer has such alien primitives compared to the analog world that we come from. The computer also tends to impose a very particular aesthetic on works that are created in a natively digital way, and I think programming is the most natively digital way to create. 

Computers like to do things perfectly and are engineered to do things perfectly. They like straight lines and clean surfaces. The analog world has none of that — nothing is perfect; everything is broken and dirty and wobbly. I’m fascinated by the imperfection of the analog world and how the digital world tends to lack that, and I enjoy investigating that artistically. I think that other artists have seen the same thing, which is why it’s been so popular lately. 

‍AE: The reality, it seems to me, for artists working with technology is that the systems you are working with invariably end up inflecting the systems that manage our lives. It would be very easy to have a conversation about generative art that focused solely on formalism and aesthetics, but do you ever reflect on the political implications of your work?

‍TH: I regard programming and software as a superpower, which if you apply it to almost any field has the capability to drastically change it. For a long time, people have resisted applying that power to artwork, which is why I think early generative artists were really shunned by art critics. But if artists aren’t playing with these tools, then that means that they are being utilized purely by large corporations that are attempting to sell you products, or else by the military. Isn’t it a much nicer situation to have artists as some of the first people to get to experiment with AI and really develop our relationship with it?

Art is part of how we establish healthy relationships with new technology, and programming is the ultimate technology. That’s why it is critically important for artists to be deeply involved with it.

Historically, I’ve viewed the computer as a tool. Sometimes I refer to it as a collaborator, but mostly in the sense that it imposes its whims on me because, like any medium, it has both strengths and weaknesses. Generative art tends to shift the role of the human toward a more curatorial role. QQL is a great example of this. The algorithm can spit out endless pieces of artwork, but it really takes the curatorial eye to identify and assign meaning and value to those outputs. AI art tools [like DALL-E] are very much the same — you’re absolutely curating the output and you might even be curating the input, as some artists like Helena Sarin do beautifully. As tools become more capable of doing things themselves, our role shifts to evaluating the output and finding meaning in it.

‍AE: I sometimes jokingly refer to QQL as a grail machine. Do you feel that there is a grail formula? If so, is it the golden ratio? Are we dealing with a neoclassical logic that reveals an underlying structure to the world?

‍TH: I think what’s most surprising to me as an artist and also as a close observer of how QQL is unfolding is that there’s an intense variety of shapes that a grail or masterpiece can take. This is true not only at a formal graphical level, but also at a higher level. Some people really go into QQL looking for a type of purely graphical abstract composition, while others are much more interested in finding representational elements that happen to match the world out of pure chance; others go into it with a sense of humor. So there is no solid formula for how to produce something amazing, and what I’ve come to realize is that we evaluate artwork in an incredibly high-dimensional space.

We have basic ideas around things like contrast, rhythm, texture, and negative space — tools that artists sometimes employ to evaluate the quality of a piece of artwork. But there are so many exceptions to each of those rules that it’s quite clear that they are just heuristics, and not true guidelines. You can talk all you want about composition but then look at a work by Jackson Pollock or Agnes Martin that breaks all the rules and love it more than, say, an Edward Hopper painting that follows all the rules. 

I’ve long had this idea of trying to create the ultimate generative art algorithm that produces an endless series of continually interesting outputs, but what you learn as an artist is that to accomplish that, you really need strong AI, as well as the knowledge of all human culture and art history. We’re just not capable of encoding that type of evaluation into a generative algorithm yet. 

So there is no formula for producing a grail. And that’s honestly a good thing, because that’s what keeps it interesting.

‍AE: What strikes me about QQL is the sheer variety of high-quality outputs: from the monochrome to the densely patterned, and from figurative to abstract. How has the community of collectors responded to the project?

‍TH: An interesting trend so far is the extent to which people are looking for outliers in the algorithm. There might be an image that is incredibly well-balanced with all the colors playing nicely, as well as a great rhythm and composition. But if it’s not a type of rare output from the algorithm, I think people are less likely to appreciate it. What I’ve noticed is that people get most excited by something that is rarely seen from the algorithm, even if it might score lower on some of these other metrics. I find it really fascinating that you can’t evaluate the quality of a given output without knowing the rest of the body of the algorithm and what it tends to produce. What’s also beautiful is that, as people spend more time with the algorithm, they tend to develop a more particular taste about what they enjoy. It’s really fascinating to see that evolve. 

‍AE: It does feel like there is a certain gamification of aesthetics involved here, that the project operates as both art and game.

‍TH: I never considered it to be a game. But it did become apparent that people were using it like a game and getting enjoyment out of playing with it. I’ve heard from many people who have never created generative art or any type of artwork before QQL who have spent days or weeks generating artwork and learning about this world. I think it has a beautiful way of entrancing people without necessarily being built to do that.

‍AE: It also makes me wonder about the potential therapeutic role that such activity might play. We published an article earlier this year by Peter Bauman about how generative art decentralizes creativity, and it seems to me that QQL offers even more creative agency to the collector. Are we seeing a progressive decentralization of creativity?

‍TH: I think that’s a lovely idea. To be honest, I haven’t thought of it in those terms explicitly, but hearing you say it, it makes a lot of sense to me. Creation has always been a multi-party story, and a lot of artists are familiar with the concept that the work is finished by the viewer. 

The creator is never simply placing something into the viewer’s mind, they’re offering a suggestion and a starting point which the viewer completes. Projects like QQL make that much more explicit. 

Even something like Fidenza makes that more explicit — there are 999 outputs, but those are not all considered the same. Viewers and collectors have collectively decided that a couple dozen outputs really represent the algorithm. Those are given the most meaning. QQL takes the artist’s hands off the reins even more. Of course, there are trade-offs to that. I can’t impose such a particular vision on what the work might be like. But what I gain is that it contains the vision of thousands of people and a much more diverse set of perspectives than I can offer myself. I won’t say that that’s better or worse than any other model, but it is an interesting development.

‍AE: We’ve started to see a number of generative artists collaborating. What made Dandelion an ideal collaborator for you? 

‍TH: Dandelion was an excellent collaborator for this project for a couple of reasons. First, they are a really brilliant individual, particularly in the conceptual space. They also have a very natural understanding of Web3 and its collective ethos. Furthermore, Dandelion brought skills to the table that I would have struggled greatly to execute — work involving smart contracts and building out the QQL infrastructure. They also made some great contributions to the visual elements of the artwork, and we were able to work smoothly as a pair on developing QQL’s visual language. 

I think QQL is just the beginning of this type of model, and that the decentralization of creation can be pushed even further. Dandelion and I have already talked about other models that go further than QQL and, at this point, I don’t think we’ll be the only ones with those ideas. 

I think it’s also a really natural fit for Web3 and the culture of giving power to those who didn’t have power before. 

It also really takes advantage of the power that smart contracts have to offer, and I know that artists can’t resist exploring something like that. So I’m quite confident that we will see many more works along these lines. The tough part is that it’s not easy to pull off a project like this — it’s quite a big undertaking. So I think it will take some time before it’s a really accessible art form.

‍AE: You’ve spoken about the potential for QQL collectors to wait years to mint. How important is time — which is so central to art on the blockchain — to this particular project?

‍TH: A component that really interested Dandelion and me from an early stage was adding the element of time to the exploration of the algorithm. Prior to Art Blocks, whenever I was curating outputs, by the time I reached the later stages — say 500 images in — I became much more particular about what I enjoyed from the algorithm and what I thought was high quality or special. My takeaway is that this type of curation is heavily influenced by time and experience. The more time that you spend with the algorithm, the more particular your tastes become. 

We were really interested in introducing time to QQL in order to guarantee that people would really acquaint themselves with the algorithm and the type of output it produces. And because it’s such a high-variety algorithm, it takes an incredibly long time to familiarize yourself with it. We were curious to see whether people would make a meaningful effort to investigate the undervalued corners of the algorithm. 

‍AE: Web3 seems to have created a state in which artists and collectors are increasingly blurred. It sometimes even feels like artists need to be collectors in order to succeed as artists. The NFT certainly seems to have financialized digital aesthetics, but I also wonder whether we’re moving toward an aesthetics of finance, which is to say an ecosystem where the real creativity lies in the creation, or gamification, of markets than in anything else. What is your perspective?

‍TH: First, you’re absolutely right, NFTs bring money clearly to the forefront and marketplace websites like OpenSea and Archipelago include sales records right next to the artwork. So as things stand right now, art and money are incredibly intertwined, and I think you could argue that, in some ways, that makes for much healthier markets. But as you’ve stated, that also raises the question of how much the money skews the artwork and how much it enters the artist’s consideration regarding how a project is executed. Different artists will have a different take. Some might adopt the approach of Andy Warhol, for whom: “[...] good business is the best art.” That doesn’t really excite me but I think NFTs offer a lot of tools for it. 

One particular element that I do think is interesting is the ability to sell seeds. In QQL a seed determines what art comes out of the algorithm, and the person who discovers a seed has an exclusive right to mint that as an NFT — nobody else can come in and use that seed without their permission. What we have done is given the ability for the user to transfer or sell that seed to somebody else. 

Participants in QQL — people from the community who are generating artwork — actually have an ability to earn money from their creations. And, to us, that was a really positive development. 

They are essentially becoming a parametric artist within a very particular medium. This has created a new market and possibilities for artists of a new and particular kind to appear and to work.

‍AE: Earlier this year, you exhibited a large print of Fidenza #61 (2021) at Phillips in London. What struck me was that the physical output displayed marks that I’ve not seen produced in painting or printmaking for that matter. This was a physical version of a born digital object, which provoked an uncanny sensation. I know that you’re a big fan of plotting and printing and a work’s physical materiality, but are you surprised at how your work can look so different when it is revealed in the physical world?

‍TH: I have a lot of thoughts on this. The first is that digital display technology is currently far behind printing in terms of resolution, color accuracy, scale, and matching aspect ratio. For all these reasons, it’s really challenging to provide ideal viewing conditions through a digital display. Prints make that much easier to achieve. So for purely practical reasons, I often elect to print work when I want it to be shown in its best form. That’s not something I’m dogmatic about. If we find ourselves ten years from now with better digital displays, I think I’ll be happy to transition from prints to a more natively digital viewing format. 

However, I do have a great love for creating physical works. I love the messiness and the imprecision and all those things that my analog body craves in the physical work. I also love to use my hands and to get more physically involved in the creative act. That’s something that working with the plotter and painting allows me to do. Artists have always, maybe accidentally, found a way back to painting. 

Generative art has been a component of many other art forms for hundreds of years, and every medium has more or less of a generative approach to it. With watercolor, for example, you’re electing to yield control to the medium of water, which is a more chaotic element. And in return, you get these unpredictable and uncontrollable elements that add life and magic to your work. I think that people will soon start to look back at other art forms and really recognize the generative component that was formerly implicit.

What’s interesting for me is that it required computers and digital and programmatic thinking to arrive at a new form of painting that actually felt fresh. 

There are definitely some outputs that are better on a digital screen, which has to do with the backlighting, but also our expectations around what belongs on a digital screen and what belongs in a print. Because I earned my living as an artist selling prints for a long time, I became very sensitive to what types of work would print well. And so I think I have naturally evolved to a place where most of my work looks good printed by default, but there are absolutely exceptions.

‍AE: Finally, what can you tell us about your forthcoming collaboration with Pace Verso?

‍TH: So Pace and I are working together to show an exhibition in spring 2023 that will focus on QQL, and I’m looking at creating a series of paintings that are derived from QQL-generated images. I’ve always been fascinated with taking steps back and forth between the digital and physical worlds, and this will be another exciting step down that road for me.