Moth Generator (@mothgenerator) is an interactive, multi-faceted, collaborative digital artwork by Katie Rose Pipkin and Loren Schmidt. The following statements illustrate its complexity and set the stage for an eventual preservation plan for this work:
Moth Generator is:
- A Javascript drawing program that creates images of imaginary moths by translating text into numbers
- A Twitter feed where moths are regularly published and @replies are used as moth-generating text
- A collection of computer-generated moth images and names, including looping animations created from generated moths and reused for other purposes
- An element of a complex virtual world project
- A collaboration between a game designer and an artist whose work deals in large part with code and bots
- A relatively well-known example of a Twitter bot
what is a bot?
To better understand Moth Generator in context, we might begin with a Wikipedia summary: A bot is a “software application that runs automated tasks (scripts) over the Internet.”
Woolley et al. (2016) point to the ubiquity of bots performing invisible work on which the internet runs, and also to “an emergent kind of bot, capable of interacting with humans and acting on their behalf, [that] is playing a more active role in our everyday lives.” Of these, social bots (or socialbots, as this New York Times article spells it) are some of the most visible thanks to their interactions with users within the constraints of social media platforms. Twitter bots are a specific instance of social bot: Twitter accounts that publish posts and/or interact with users based on running certain kinds of inputs through programs.
Twitter bots come in many forms, offering social and political commentary (often accidental), absurd juxtapositions, and variations on jokes, to give a few examples. As Woolley et al. write, “Bots can be useful for making value systems apparent, revealing obfuscated information, and amplifying the visibility of marginalized topics or communities.” Protest bots (per Mark Sample) certainly do this, but so might art bots not explicitly dedicated to social commentary. Describing some of her bot work, Pipkin has said:
Looking at my source code for those projects now is very much like looking at an outsider-art approach to computer science. Which is, I suppose, what they are.
When asked in an interview, “What, indeed, are bots?” Pipkin pointed to their descent from other forms of automata: “Digital bots (especially bots that live in social spaces) fit into this long history of objects granted almost-humanness. They fill in for a part of human action, the slice of person granted to digital representations of ourselves.” Woolley et al. write, “One distinguishing feature of bots is that they are semi-autonomous: they exhibit behavior that is partially a function of the intentions that a programmer builds into them, and partially a function of algorithms and machine learning abilities that respond to a plenitude of inputs.” Bots and their designers perpetually navigate the boundary between intent and accident, including ethical issues involved in giving bots more or less free rein.
Woolley et al. are clear about the upsides to the thoughtful use and study of bots, even as they call (in their “bot-ifesto”) for critical attention to and policy-building around them: “By embracing messiness, those who make, use, and interact with bots open themselves to the automated creativity, innovation, and unpredictability so central to the web.” That Pipkin and Schmidt operate in this emerging space contributes to the significance of Moth Generator as an ongoing work of digital art.
how does @mothgenerator work?
The program behind @mothgenerator is a Javascript drawing process and set of rules. Components of the program include translating text to numbers and drawing individual moth parts like wing shapes, markings and coloring, legs, and antennae. Taking text as input, the program converts each letter or number into an integer. These integers are the input for separate processes forming an image of a moth from “some 50-100,000 strokes” according to a ruleset defined by the artists. (See interviews by Levy,Nguyen, & Voon for additional background.)
Text input for moth creation has multiple sources. The bot periodically generates and tweets moths whose names are randomly pulled and combined from a set of about 10,000 Latin moth names and 4,000 English names, all real. Pipkin used a web crawler to build the two name pools. An example is below (and more appear throughout the post):
juniper-tabby-abbreviated
bopteryx hodgannascus pic.twitter.com/7pLAkAK0w2— moth generator (@mothgenerator) March 5, 2016
Another extremely compelling source of input is tweets from other users. The artists have programmed Moth Generator to use @replies to @mothgenerator to “seed” new moths. A recent example is below:
@wileysegovia the Bernie Sanders in 2016! moth pic.twitter.com/ez9lo3BJ0r
— moth generator (@mothgenerator) March 19, 2016
stakeholders and significance
Assessing and unpacking the significance of Moth Generator requires identifying major stakeholders in its long-term preservation. What each stakeholder values about Moth Generator must be considered — and negotiated between — in order to produce, execute, review, and update a preservation plan.
The artists
Pipkin and Schmidt, the artists behind the bot, are the primary stakeholders in preserving Moth Generator. It’s possible to draw out key concerns from interviews, talks, and blog posts about Twitter bots in general and this project in particular.
Although this project is typically referred to as a Twitter bot, Pipkin has described Moth Generator as a drawing process for which Twitter acts as publishing platform (interview by Fader, 2015). In other words, the drawing program at the heart of Moth Generator is the key component, and @mothgenerator is just one of many possible instances of its implementation. As further evidence of the drawing program’s importance, Schmidt emphasized a link between Moth Generator and entomological illustration in an interview: “Though our own moths are digital, the way they are constructed has more in keeping with a drawn moth than a photograph, and this was an important touchstone” (Nguyen). This connection is expressed not only in the generation of moth shapes from many thousands of marks, but also in how constraints and rules for how moths can look are continually adjusted. Strict adherence to real moth shapes and textures are not the goal.
The bot’s relative autonomy — outside of occasional tweaks to the drawing program and a collection of error screenshots — is another key characteristic from the artists’ perspective. Pipkin has said that she does not monitor her many Twitter bot accounts: “This means that their notifications never reach me; the things that are said to them (or their own replies) are often invisible to everyone but them” (interview with Bucher, 2015). Users interact with the bots and sometimes with each other, their conversations tied to bot tweets through Twitter’s data structures. Interactivity contributes to this project’s significance, not only in the field of social bots but also within the artists’ respective bodies of work. Pipkin has said, “I’m proud of the project in general, but the reason it is my favorite is because it is my first bot that uses the @-reply functionality of Twitter in a constructive way.” The artists may write the rules of the drawing program, but “bots also have their own secret lives outside of intention.”
In addition to being a process with a life of its own, this Twitter bot fall within what Pipkin calls work-by-generation: “a fundamentally similar, but shifted process from that of work-by-hand; rather than identifying and chasing the qualities of a singular desired artwork, one instead defines ranges of interesting permutations, their interpersonal interactions, how one ruleset speaks to one another.” This echoes a frequent argument within the field of software and digital art preservation — or anywhere source code is in play — that preserving potential is a key function of documenting and preserving. To Pipkin and Schmidt, Moth Generator is as much the potential for new moths as it is a collection of them. Pipkin has hinted at moving away from this kind of visible, published output in her experiments with bots: “I have an interest in biological emulation and in the hidden data that Twitter links to every tweet (perhaps my next bot will not be readable on the Twitter webclient, but instead comes alive in an API call?)” Preserving Moth Generator’s potential may mean capturing the kind of underlying Twitter data that allows Pipkin to reuse and remix her own work in the future.
Finally, it’s important to note that Moth Generator was not meant to be a standalone program or bot. Schmidt and Pipkin created it as one element of a larger virtual world project based on “wandering through a generative landscape of interaction” (Levy). They conceived of the virtual world as a second installment of a web-based project called inflorescence.city in which texts and map-like images regenerate each time a URL is refreshed. Moth Generator’s embeddedness should be reflected in any preservation plan.
The community
“Community” is an admittedly broad term for a group comprising audience, co-creators, researchers, and critics. I’m inclined to group them together because, while their precise expectations of Moth Generator are bound to vary, what they value about the project seems to overlap quite a bit.
Moth Generator’s audience includes Twitter users who follow and/or interact with @mothgenerator. Those who follow @mothgenerator on Twitter may do so because they enjoy seeing new moths on the regular. @mothgenerator’s context in a Twitter timeline makes the occasional moth an unexpectedly lovely moment amid news, networking, arguing, and other normal Twitter activities. Many of the articles, interviews, and blog posts about this particular bot note the beauty and weirdness of the moth images and names.
Expectations of these stakeholders may include the ability to @-reply Moth Generator and receive their very own custom moth in response, as well as playing around with wording to generate dramatically or subtly different moths. Other forms of interactivity include sharing moths with others by @-replying other Twitter users, retweeting or favorite-ing @mothgenerator tweets, and potentially even interacting with other users who also reply to or share a moth. For Moth Generator’s audience, Twitter’s interface and functionality may be central to its significance. Given recent uproar about potential changes to Twitter’s 140-character limit and the introduction of a timeline sorted by algorithm (not chronology), capturing some representation of how @mothgenerator currently appears is important. While the character limit change was ultimately rejected, it’s hard to predict how future tweaks to Twitter’s appearance and functionality will affect the experience of creating and interacting with Twitter art. A preservation strategy for Moth Generator will help document this particular moment in bot-making and Internet art.
Finally, it’s possible to see Moth Generator as a digital collection, since what users disseminate when they share a @mothgenerator tweet includes an image and a short bit of text. It’s an unruly and distributed collection, however. Still images produced by Moth Generator may exist in many different places online, including all over Tumblr and in this animated iPhone background.
auras forever?
Pipkin wrote in a post on Medium:
A digital object appears across a multiplicity of screens both at once and forever … There is no distinction between original and copy; they exist at the same scale, have the same aura.
This is debatable as a reality, if not as an artistic approach. We’ve seen over and over again that metadata — what allows this connectivity and shared aura to persist — does not seem inclined to perpetuate or replicate itself. Deliberately collecting, describing, and monitoring source code, documentation, and actual tweet data for long-term preservation will go a long way in supporting researchers studying Twitter art and artists remixing Moth Generator in some way. Pipkin and Schmidt value the autonomy of their bot, but they also seem to trust in its ability to operate indefinitely. Planned obsolescence, glitching, and decay don’t seem to figure in their vision of Moth Generator’s future, so there’s no reason at the moment to think that Moth Generator should not be preserved.
Moth Generator is not only a drawing process, a virtual world feature, a Twitter bot, or a collection of digital images; it is all of the above at once. Assessing the significance of this project means recognizing how its components and manifestations interact and designing a preservation strategy around those interactions.
sources
Bucher, T. (2015, December 22). About a bot: Interview with Katie Rose Pipkin. Humanity+ Media.
Dewey, C. (2016, January 6). Twitter’s rumored character-limit change is bad news for artists and botmakers. Washington Post.
Fader, L. (2015, November 9). 12 weird, excellent Twitter bots chosen by Twitter’s best bot-makers. New York.
Irwin, M. (2013, August 9). Techno-artistic: Katie Rose Pipkin and the new art. The Austin Chronicle.
Levy, L. (2015, August 5). How to make a moth. Studio 360.
Mears, J. (2015, March 29). All about that bot. Digital Public History.
Nguyen, C. (2015, July 16). This Twitter bot automatically generates imaginary moths. Motherboard.
Pipkin, K. R. (2016, March 12). Selfhood, the icon, and byzantine presence on the Internet. Medium.
Sample, M. (2015, October 3). A protest bot is a bot so specific you can’t mistake it for bullshit. samplereality.
Urbina, I. (2013, August 10). I flirt and tweet. Follow me at #socialbot. New York Times.
Voon, C. (2015, July 16). A Twitter bot that generates beautiful, imaginary moths. Hyperallergic.
Williams, J. (2015, September 9). Q+A with Katie Rose Pipkin. University of Texas at Austin Department of Art and Art History.
Woolley, boyd, Broussard, Elish, Fader, Hwang, et al. (2016, February 23). How to think about bots. Motherboard.
I like how you start out by listing out the various components of what Moth Generator is. That kind of list is very useful at helping to explicate all of the potential components of it that might be preserved. With that said, I don’t really have a concrete sense of what.“ An element of a complex virtual world project” means. Could you unpack that a little more? Also, on the point about it being “A collaboration between a game designer and an artist whose work deals in large part with code and bots” it seems like a key point there is that it is part of both of these artists portfolios of work, that is, it is part of the continuity of their work across various projects. Which fits great with your last point, that it is an exemplar of the genre of work that is a twitter bot.
Listing things out like this reminded me of this great bit in Ian Bogost’s book, Alien Phenolonology: Or What it’s Like to be a Thing, where he describes 11 different things that the E.T. Atari Game. I’ve pasted it below in case it is also useful in thinking about your list.
• E.T. is 8 kilobytes of 6502 opcodes and operands,
• E.T is also it’s sourcecode
• E.T. is a flow of RF modulations that result from user input
• E.T is a mask ROM, memory etched on a wafer.
• E.T. is a molded plastic cartridge
• E.T is a consumer good, a product packaged in a box
• E.T is a system of rules or mechanics that produce experience
• E.T is an interactive experience players partake in
• E.T is a unit of intellectual property that can be owned, licensed, sold
• E.T is a collectible, an out of print scarce object
• E.T is a sign that depicts the video game crash of 1983
I think you do a great job describing what bots are and linking out to various sources to explain them. That sets up a good case for why it is worth thinking about preserving twitter bots as part of an emerging area of digital art.
With that said, it would also be great to get a bit more insight into why Moth Generator over any other given bot. I think you have all the material that you would want to draw from to make that case, but it would be good to be explicit about why this particular bot is significant which I imagine may also have some implications for how you go on to establish your preservation intent. I could imagine that you might argue that the way the bot works is particularly novel, or that the partnership behind it is, or that the particular artists are significant. That is, there are a lot of routes for this, but it would be good to anchor your approach in one of them.
Great point about the functionality of Moth Generator as part of the way people use twitter being key to its significance. This line of thinking could suggest an approach to reimplementation in the way that Rinehart and Ippolito discuss. At the same time, your suggestion of thinking of it as a digital collection, that is the outputs of the work, would also document much of the way that it has worked over time.
This is really great work. You’ve provided some great context on what bots are and how they work, and managed to unpack a lot of the different elements of what matters about this particular one. I think this sets you up well for working up a preservation intent statement.
Honestly, I’m not entirely sure myself what the virtual world project will look like. It’s still in development… but here are some wild guesses.
+ Some kind of alternate reality game with computer-generated landscapes mapped to real locations — carry your phone (or other personal tracking device) to the right spot to unlock clues.
+ A window-based video game: maybe players have on-screen avatars and navigate through landscapes inhabited by moths. Maybe the moths pop up as one travels through the landscape or are elements to be “unlocked” in some way.
+ A space experienced through virtual reality, in which motion or text input generates moths, possibly overlaid on the real space around players.
+ An open-ended Minecraft-type game in which players build terrain for pretty much any purpose they choose.
+ I actually really, really hope it’s something like Walden, A Game, in which the objective is either less direct or less tangible than navigating space a certain way with certain milestones.
All I know so far is that the landscape generates / is generated as you go. So maybe navigating a virtual space sets landscape generation in motion and also triggers moth generation? This might all build on a project like Mirror Lake that slowly draws or grows landscapes based on some code. Despite the artists’ sequel talk, I’m not sure how closely related this new project will be to Inflorescence City, which is in-browser and almost book-like despite lots of imagery. It reminds me most of Project Gutenberg’s stripped-down interface for reading books in HTML.
Anyway, I suspect Moth Generator represents both the confluence of trends in each artist’s work, but also the start of a new direction through collaboration. I chose this particular bot for that reason — it’s part of many different bodies of work — as well as for the many different ways it’s possible to cut, frame, filter, and remix the phenomenon of Moth Generator so it can continue to be generative. I can think of several ways not to preserve it, in addition to more promising directions to pursue, so that’s an added challenge built into the decision-making around this assignment.
Thank you for defining bots a bit and introducing this particular one – I was not familiar with it, but will start following it on Twitter. The images above reminded me of the boxes and pins entomologists or collectors would use to preserve bugs and insects. As individual art pieces, I initially thought of them in those boxes on display in a museum with the made up name and provided text from the user. Though this idea is pretty far removed from it being interactive art on Twitter – the images themselves reminded me of a very traditional way of collecting and presenting bugs.
But, the other interesting aspect to me is how the bot has taken on a life of its own. The artist isn’t even really monitoring the account anymore right? So, capturing the Twitter feed somehow and the interaction seems important. But, how would the feed be preserved? Would you want to curate/edit it in some way? Trevor also mentions reimplementation above – could there be a different way for users to interact with the bot in the future because Twitter may not behave the same way it does today?
Kerry, that’s a really good point: the moths themselves aren’t interactive. At first I thought this might make them the least interesting part of the project, with the least potential for remixing and reuse. But maybe it’s possible to build something from the moths in aggregate, or to use that nice grey background and plain presentation to Photoshop out everything but the moths (wand tool!) and use them elsewhere.
Based on the tools available right now, capturing the Twitter feed could mean using Twitter’s account and/or search APIs to collect all tweets by or mentioning @mothgenerator as structured data (JSON). The many metadata elements associated with each tweet could be used to build something else down the road. So that gets at the networked aspects of the Twitter feed. For capturing how Twitter looks, if I were doing this today I would use WebRecorder to capture a WARC file of the Twitter feed and replay it in a browser later. There are so many tools for collecting social media data that it’s hard to choose just one or two, or to know what else will be available to experiment with a year from now.
It sounds as though the code for generating moth images is independent of Twitter and could potentially input from and output to a different interface. So interacting with the generator in the future might mean putting text into a text box and seeing a picture on the next screen. I tried drawing out an authenticity-access matrix (following Espenschied) but am struggling to come up with a One Ring scenario. Probably this will be history’s most hodgepodge AIP.
I think this is such an interesting project, and I love how you bring out how interactivity is important to this bot and others. The fact that the creator does not monitor that interactivity reminds me of a particular AI-bot-turned-Nazi that appeared recently, and the implications that lie within that– the moth generator, to a certain extent, reflects its audience, and isn’t negotiated by the creator. What is negotiated is the parameters that appeal to certain audience members. Which is interesting– what about moths attracts Bernie Sanders / Trump fans? I think your project has a unique opportunity to allow researchers to connect certain aesthetics and online platforms with certain communities/values/opinions. This might be worth considering as you parse out what significant bits you plan to collect!
Thanks, Alice! It’s true, @mothgenerator and Tay (?) both reflect those who interact with them as much as they build from creators’ assumptions and decisions. I didn’t think about this early on, but am now wondering how much the choice to generate moths was about worst-case-scenario transforming potentially ugly Internet speech into (probably) inoffensive images. As in, maybe treating text agnostically saps or redirects its power.
Actually, my favorite recent moth might be the one generated through a response the #Vote4Trump moth. The original poster replied to @mothgenerator’s tweet so now we also have the Wow this captures trump’s skin tones and hair colour pretty well moth.
Let me quickly comment on the stakeholders statement: I don’t think the artists are the main stakeholders. Typically artists have largely different interests from their audience, no matter how much they are referring to the non-existence of digital originals or aura. This is for example clearly exemplified by the museum-like presentation of the generated moths: https://twitter.com/mothgenerator/status/705676750785802240 — yet the bad quality of the images on twitter (thanks to twitters JPEG brutalism, but handy for the artist) will still make it possible to present the “original” artifacts much later, in 600dpi digital prints. Also, mothgenerator operates very much like a transparent web service, with no source code being available, and a carefully enacted dramaturgy of not-too-much-neither-too-few moths being produces so they stay consumable for followers. So I guess the artists are taking care of that.
This is all to make the case for the users’ perspective on the mothgenerator being much more fragile and worthy of preservation. How do the moths look inside somebody’s twitter timeline? How would users interact with the service? What makes a good, successful moth (like the mentioned Trump specimen)? Is that driven by how the moth looks or how it is embedded in the online dialogue?
Thanks for your comment, Dragan, and excellent points. I hadn’t considered scarcity as an element of Moth Generator, but it’s true: not being saturated with moths does seem to be part of the appeal. Perhaps the bot’s generative potential doesn’t need preservation support while the encounter and interaction experiences do, if we have to set priorities (which: yes). This is also highlighting how I tend to think of digital preservation from a data management or even records management perspective, as in what to do with working files; and how that might not always be a good fit when it comes to artwork. Much food for thought.