This story is over 5 years old.


There's Not Much 'Glitch' In Glitch Art

Algorithmic artist Daniel Temkin tells us about exploring the "wilderness within the machine."

Artist Daniel Temkin has been creating and discussing glitch art for over seven years. In that time, he's exhibited in solo and group shows, and had his work featured in Rhizome and Fast Company, amongst other publications. For Temkin, glitch art is about the disruption of algorithms, though algorithmic art is a bit of a misnomer. He prefers "algo-glitch demented" in describing the methods, aesthetics, and philosophy of glitch.


In January, Temkin published a fascinating glitch art essay on NOOART titled "Glitch && Human/Computer Interaction." There he laid down the philosophy and "mythology" of glitch, which had really started in a series of email conversations with Hugh Manon. Though there is no shortage of writings on glitch art, many aspects of the these texts didn't address what Temkin loved most about how it is created.

"The glitch aesthetic may be rooted in the look of malfunction, but when it comes to actual practice, there’s often not much glitch in glitch art," wrote Temkin in the essay. "Yes, some glitch artists are actually exploiting bugs to get their results — but for most it would be more accurate to describe these methods as introducing noisy data to functional algorithms or applying these algorithms in unconventional ways." This, he said, doesn't make it traditional algorithmic art (algorithm-designed artworks), but a more demented form of it—algo-glitch demented.

Over a series of email conversations, Temkin elaborated on some of his conclusions in "Glitch && Human/Computer Interaction." Aside from highlighting some of the best algo-glitch demented art, Temkin also talked about bad data, image hacking, and why computers are no less "image makers" than humans even though they aren't sentient (yet).

MOTHERBOARD: Aside from being an artist working in glitch, would you say that you've also sort of become a philosopher of glitch or algorithmic art, if there is such a thing?


Temkin: There's tons of writing on glitch, much of it very good (, for instance), but some aspects of glitch theory didn't jibe with what really interested me about the style. Originally, Hugh Manon and I started a long email conversation about glitch, which evolved into our 2011 paper. It ranged across glitch aesthetics, methodology, and issues around authorship, while delving into glitch's ambivalence about error—the way the glitch is possible because of software's ability to "fail to fully fail" when coming across unexpected data.

We questioned why computer error is so emphasized in this form when nothing is really at stake in a digital file (a deleted but endlessly reproducible JPEG has none of the aura of an Erased DeKooning), and what it means to purposely simulate an error, something that ordinarily has power because it is unexpected and outside of our control.

Ted Davis, FFD8 project

These issues stuck with me, until I considered Clement Valla's familiar quote about his Postcards From Google Earth project: that "these images are not glitches… they are the absolute logical result of the system." It was a familiar quote, but in this instance got me thinking about how most glitchwork can be described the same way—as products of perfectly functional systems.

I wrote my recent piece for NOOART, arguing that glitch's preoccupation with error doesn't always serve it well, that it limits the scope of what's produced and how we talk about it. Bypassing computer error opened new avenues of investigation about our relationship both with technology and with logic systems more generally, and got at what interested me more about the style we call glitch.


In the NOOART essay, you write: "Some glitch artists are actually exploiting bugs to get their results — but for most it would be more accurate to describe these methods as introducing noisy data to functional algorithms or applying these algorithms in unconventional ways." Can you elaborate on that point?

In the paper, I discuss JPEG corruption, one of the fundamental glitch techniques. Introduce bad data to a JPEG file, and you'll see broken-looking images emerge. I use this example because it's so familiar to glitch practice. JPEG is not just a file format but an algorithm that compresses/decompresses image data.

When we "corrupt" a JPEG, we're altering compressed data so that it (successfully) renders to an image that no longer appears photographic, taking on a chunky, pixelated, more abstract character we associate with broken software. To the machine, it is not an error—if the image were structurally damaged, we would not be able to open it. This underscores the machine as an apparatus indifferent to what makes visual sense to us, at a place where our expectations clash with algorithmic logic.

Daniel Temkin, Dither Studies #2, 2011

The excitement of altering JPEG data directly is the sense of image hacking—making changes at the digital level without being able to predict the outcome. This becomes more apparent in other glitch techniques, such as sonification, which add layers of complexity to the process. Giving up control to a system or process has a long history in art.

Gerhard Richter describes committing to a systematic approach, veiling the work from conscious decisions that may ruin or limit it. As he puts it, "if the execution works, this is only because I partly destroy it, or because it works in spite of everything—by not detracting and by not looking the way I planned" [p179, Gerhard Richter, Panorama]. In digital art, we often function in an all-too-WYSIWYG environment. Glitch frees us from this, bringing us to unexpected places.


Can you draw a distinction between generative art (which can feature algorithms) and your concept of algo-glitch demented?

I call it algo-glitch demented, as opposed to algorithmic art (which I understand meaning generative art that uses algorithms). I'll have to paraphrase Philip Galanter and say that generative art is any practice where the artist sets a system "in motion with some degree of autonomy," resulting in a work.

"Glitch is a cyborg art, building on human/computer interaction. The patterns created by these unknown processes is what I call the wilderness within the machine."

What makes algo-glitch demented is how we misuse existing algorithms, running them in contexts that had never been intended by their designers. Furthermore, there are moments of autonomy in algo-glitch, but this autonomy is not what defines it as algo-glitch; what's more important is the control we give up to the process.

You call glitch art a collaboration with the machine. That's an interesting point because the human is conscious of this, while the machine is not. Or, do you have another way of looking at that collaboration?

Machines are not sentient, but they are image-makers. Trevor Paglen, in a recent Frieze Magazine piece, says we are now or very soon to be at the point "where the majority of the world’s images are made by-machines-for-machines," and "seeing with the meat-eyes of our human bodies is increasingly the exception," refering to facial-recognition systems, qr code readers, and a host of other automation.


One of the most compelling ideas to come from James Bridle's New Aesthetic is how we can treat the machine as having a vision—even as we know it's not sentient—and just how strange this vision is, that does not hold human beings as its audience.

Jeff Donaldson, panasonic wj-mx12 video feedback, 2012

Glitch artists have been doing this for a long time, treating it as an equal collaborator and seeing where it leads us as we cede control to broken processes and zombie algorithms. Curt Cloninger describes it as "painting with a very blunt brush that has a mind of its own;" in this way, glitch is a cyborg art, building on human/computer interaction. The patterns created by these unknown processes is what I call the wilderness within the machine.

Can you talk about glitch as mythology? I've never heard it described as such.

I'm probably being a bit obnoxious there, using mythology to describe the gap between how we talk about glitch and what we're actually doing. There are several strains of work within glitch or that overlap with glitch. There is Dirty New Media, which is related to noise-based work; materialist explorations; the algo-glitch I've emphasized in the JPEG example; and what we might call "minimal slippage glitch" (a term that arose in a Facebook discussion between me and Rosa Menkman).

Minimal Slippage fits a familiar contemporary art scenario of the single gesture that puts things in motion and reveals something new. It's great when things actually work this way, but when this language is used to describe work made by manipulating data repeatedly, there's a problem.


I also take issue with the term glitch art. I don't propose we replace it, only to be more conscious of its influence. If we produce work with other visual styles using glitch processes, why limit ourselves to work that has an error-strewn appearance? This connection begins to seems artificial. I kept this in mind with my Glitchometry series. I use the sonification technique to process simple geometric shapes (b&w squares and triangles, etc.) into works that range from somewhat glitchy to abstractions that fall very far from a glitch aesthetic. They emphasize process, the back-and-forth with the machine, and an anxiety about giving up that control.

Clement Valla, from “Iconoclashes” 2013

With Glitchometry Stripes (an extension of the Glitchometry work), the results are even less glitchy in appearance; this time using only sound effects that cleanly transform the lines, ending up with Op Art-inspired, crisply graphic works that create optical buzzing when scrolled across the screen.

You mention Ted Davis's FFD8 project in your essay. What is it about the work that you like?

FFD8 is JPEG image hacking, with protection against messing up the header (which would make the image undisplayable). It's a gentle introduction to glitching, but it illustrates how it works, which encourages one to go deeper. I'm suspicious of glitch software that does all the work for you, essentially turning glitch styling into the equivalent of a Photoshop filter. With FFD8, enough of the process is exposed that folks starting out in the style might decide to take the next step and mess with raw files directly, or build their own software, or discover some new avenue to create work.


What's your opinion on something like the iPhone's panorama function, which, if you move the camera fast or in unexpected directions, creates glitches? It's movement-based as opposed to other types of glitch.

I think someone will come along with a brilliant idea of how to use it to do something fresh and interesting. One interesting work that uses photo-stitching (although not on the iPhone) is Clement Valla's Iconoclasts series. He loads images of gods from the Met's collection and lets Photoshop decide how to combine them, creating improbable composites, many physically impossible. It works because of how carefully the objects were photographed. Each is lit the same way with the same background. Many of these religious relics come from cultures where it was believed that such objects were not created by human hands. Now an algorithm, also not human, decides how to combine them to construct new artifacts.

Daniel Temkin, Glitchometry Circles #6, 2013

Where do you feel you've been most successful in your own projects?

I never trust artists to tell me which of their works are more successful. [laughs] I'll tell you the theme I'm most interested in. Much of my work revolves around this clash between human thinking and computer logic, and the compulsiveness that comes from trying to think in a logical way. My own experience with this comes from programming, which is my background from before art. Glitch gives me a way to create chaotic works as a release from the overly structured thinking programming requires.

As a few examples of work that deals with this, my Dither Studies expose the seemingly irrational patterns that come from the very simple rules of dithering patterns. They began as a collaboration with Photoshop, where I asked it to dither a solid color with two incompatible colors. From there, I constructed a web tool that walks through progressions of dithers.

In Drunk Eliza, I re-coded the classic chat bot using my language Entropy, where all data is unstable. Since the original Eliza has such a small databank of phrases, yet so clearly has a personality, I wanted to know how she would seem with her mind slowly disintegrating, HAL-style. Drunk Eliza was the result. The drunken responses she gets online have been a great source of amusement for me.