Aaron Tucker is one of the creators of ChessBard
This interview follows up on a post from last month about the digital poetry generator ChessBard. It was conducted over email.
E Martin Nolan: I suck at chess. But the poem I made by playing the ChessBard intrigues me. The first time I “poetified” the game, the poem ended with the line, “mate conqueringly all boxes hidden.” Another iteration of the game included the line, “binary butt blends behind binary.” This made me wonder about the amount of control you as designers had here. For you, what’s the most interesting thing about the way the ChessBard works? Is there anything in the process that has surprised you?
Aaron Tucker: We’ve tried to generate a machine that needs no editing and takes as much of the control away from Jody and I as possible. Largely, this meant getting the language templates that the ChessBard is using (in combination with the source poems and more complex word pools) to best reflect “literate” (and interesting!) results. After the first few iterations of this project it became very apparent that simply filling in “NOUN ADVERB VERB NOUN” only worked with certain verb forms and certain noun forms—a player notices now that the verbs only show up in simple present tense and nouns are only ever singular; similarly, we had to cut certain conjunctions and other words out of the language pool, as they were creating “disrupted” poems. The nuts and bolts of this process can be found at our project description. Earlier versions of the machine can be found in our archive.
Jody Miller: Building off that, typically, when I am surprised (as a programmer) by the output of my programs, it’s a bad thing. In this case, it’s wonderful. Between the two of us, we know every single thing about the mechanisms underlying the ChessBard, and still it retains this capacity to generate surprising poems.
AT: It’s true! Despite all the time we’ve spent in the guts of the thing, the poems produced are quite often startling in how they often link together, line to line, stanza to stanza. There are always hilarious and delightful phrases, but the most satisfying poems are the ones that somehow drag the same thought or vocabulary all the way through. As Bill Chamberlain wrote about Racter, it gives the appearance of thinking. Or maybe it is thinking! Like Racter, the ChessBard is often a surprisingly whimsical poet.
EMN: I’m stuck on the word “thinking” here. Is the “delightful” or “whimsical” nature of the poems the result of your programming, or is that a limit of the kind of thinking a program like this can do? That is, is there an inevitably limited tonal range for the ChessBard based on the random nature of its compositional process? Has it created many sad, or moving, poems? We might be getting into Yoshimi Battles the Pink Robots territory here, but could you design it to make sad poems?
AT: Well, the process itself is not really that random—it has a strict set of rules and templates and pools that is governed by computer code, which is itself is very unforgiving of ambiguity or randomness, and the poems have to pull from all that in very specific ways.
JM: And there certainly are some design limitations affecting the poems the ChessBard can generate in its current incarnation. There are indeed a fixed set of stanza templates, and a fixed word set to draw from.
AT: However, with that scaffolding, the playing of the games themselves generates the works and that is where it gets messy. Could we design it to write deliberately sad poems? Sure! At least superficially—we could go into the language pools and add all sorts of dying puppies and such—but I think that would lead to really stilted and, for lack of a better word, unnatural sounding poems. I’m not sure “thinking” is the right word either, but there is some choice going on, despite all the strict rules (especially in the playable version), and those choices lead to some sad, moving poems. Right now, the ChessBard is making some choices from its finite tools, but at its best, the poems produced do sound more natural than the previous version, somewhat accidental, and more emotionally engaging when read.
JM: A hypothetical, more flexible future version could include enough knowledge about the English language to generate poems more truly its own. Adding to that, we could use machine learning techniques to generate poems in the style of a particular genre or poet by feeding it the appropriate training corpus.
What would happen if ChessBard was played by Christian Bok’s “Xenotext”
EMN: The ChessBard translates chess moves into poetry. This assumes poetry has “moves.”And of course, it does. I’ve often wondered if a baseball pitcher’s pitch sequence, or a hockey player’s shootout moves, could be translated into poetic moves.
But I have to wonder, what is lost when we break poetry down to “moves?” What is gained? I guess this extends beyond the ChessBard, to the role of play—which you use to connect poetry moves to chess moves—in the creation of poems. On the ChessBard’s Poetics page, you refer to play in terms of the rules of play. External rules can open up possibilities, but they can also take agency away from the poet.
So, the question: With all that in mind, what has the ChessBard revealed to you about poetry “moves” and the role of play?
JM: The possibilities are indeed endless—you can express a hockey technique or any other phenomenon using a variety of expressive mechanisms—prose, poetry, music, 3D printed objects, etc. It’s just a matter of translation—of expressing the thing another way.
AT: There are interesting experiments doing similar things, such as Box Score App, or Graeme Burnett’s work. The word that jumps out at me is “translate.” The metaphoric act, at the base of poetry, is to translate one thing or action into another; it’s not just a juxtaposition (as some textbooks might insist), but a transformation that requires, for lack a better word, common material between the two. There is a need to reduce two very different complex things to some common, elemental denominator. This is a necessary task in order to find some bridge from author to reader, finding a common space that is not necessarily the most literal matching, but rather the one that is likely to resonate across those divides of object-object, reader-text/author.
This also means finding common languages. For me, the most satisfying parts of making the ChessBard were finding out how to bridge between big bundles of language, the languages/codes a computer understood and the rules/codes required of a chess game. Such a fascinating symbiotic ecosystem.
JM: Applying rules to the creative process seems to me a long-standing practice—haiku, sonnet, limerick, the English language itself, and so on. Writing is always constrained by something—deadlines, format, grammar, etc. In many ways, embracing rules can be freeing. Admittedly, the rules of chess sounds like a bit of a stretch.
AT: But still, I think, then, the movements of play, at the core, are the common material that drives the ChessBard. They are both loose (the infinite words in English; the near infinite amount of possible chess games and moves) and confined (the words in our language pools; not every chess piece can move anywhere at any time). They are performative, in that one MUST play in order to generate a poem.
EMN: Let’s dig into that “fascinating symbiotic ecosystem.” This gets close to the way I think about music and poetry. They share the property of temporal rhythm. What they don’t share is the semantic meaning inherent in language and the melodic nature of music.
AT: Right! There were some projects that were key to the initial imagining of the project, including Duchamp/Cage’s Reunion (put on at Ryerson University) and Chess Music. And to hear chess players talk about the game, they use similar terms, especially when discussing a well played game, combination, or a specific player. Reuben Fine’s quote sums it up: “Combinations have always been the most intriguing aspect of chess. The masters look for them, the public applauds them, the critics praise them. It is because combinations are possible that chess is more than a lifeless mathematical exercise. They are the poetry of the game; they are to chess what melody is to music.” Fine’s use of “combination,” as both a chess term and a creative and aesthetically pleasing gesture, tells me that we aren’t necessarily talking about differences, but are looking for overlaps, similarities.
EMN: Still, even though music and poetry can tap each other’s exclusive realms, the fact remains that music can do things poetry can’t do because it doesn’t have to mean anything, at least not in the way language is generally assumed to mean. So, does the ChessBard free the poet from a semantic burden (the need to mean purposely)? Or does the game of chess itself have semantic content that can be translated? If the moves of chess are simply expressed in a different (poetic) form, does that mean that the semantic content that comes out in the poem is somehow linked to the semantic content of the chess game? This all may come down to the question of what is being shared between the game and the poems? Or is it not a matter of sharing, but of using the game to prompt a poem that is in the end unrelated to any meaning inherent in the game?
JM: In my mind, the game is the poem, and the game and the poems generated are two ways to look at it, which I suppose lands me on the side of believing there is indeed semantic content somewhere in the game itself.
AT: I’ve been spending more and more time lately, as I try to produce a more contained version of the ChessBard (hopefully a manuscript of a famous player, Duchamp perhaps) and, after translating large batches of an individual’s games, I can see definite consistencies across the poems (opening moves, opening words) that reoccur in melodious ways. It would be fascinating to get a professional chess player to play a “sad” game and see if it produced a sad poem—I suspect the poem would be sad if we knew some of the history of the game itself, but otherwise, would be quite separate, growing from the seed of that game, but not in a cloning manner. The game and resultant poem are intertwined, but each bear characteristics distant to their individual form. True, 1:1 translation is impossible and not really the point of the ChessBard. Still, I tend to find the overlap between the two forms in how repetition leads to evolution (of an idea, form, etc.). This is a technique that nearly all the music, chess, and poetry I enjoy depends upon. In this way, the poems produced by the ChessBard repeat and evolve in really similar ways to the games themselves and start to distinguish specific players (or times, places, schools), just as we might use the same tool of unique/authorial repetitions to identify and label poets.
EMN: None of this is possible without computer technology. How do you see the ChessBard contributing to poetry’s growing use of digital forms (Patricia Lockwood’s twitter feed, online publication, code poetry, poetry apps, etc.)?
AT: We’re hoping to add to the conversations and works in the growing field by hosting what is essentially an infinite poem generator, potentially as big as the Internet. To me, that’s where a lot of really interesting digital poetry (and digital art) comes from: using what is distinct and strong about the technology itself—whether that be the dense connective collaboration made possible by networks of networks, the elegance of a well executed script or the pure, brute force thrust of a powerful computational machine. I’m less interested in digital forms that simply mark-up a poem with sounds or hyperlinks, or poetry that calls itself digital poetry simply because it exists online (as an e-book might). As with any art form, the medium is just as important as the content; by this I don’t just mean the distribution system (Twitter, a website), but the ways in which the computer technology is doing something unique to its own distinct capabilities in the process of making an art object (the poem). And if the ChessBard is successful, it is because it plays to all the different strengths of its machine and human authors and readers.
JM: I think the growing ability of new software techniques to deal with natural language have enormous potential to interact with fields like poetry, comedy, and creative writing in an incredible way. The ChessBard is a somewhat introductory version of this concept, but I expect to see much more of this type of development in the future. I think there’s a lot of room for interesting takes on computation-based, or computation-augmented, human poetry.
EMN: You’ve presented the project to university students. What was their reaction? Could you see their interest in poetry grow before your eyes?
AT: I’ve used it in my own Creative Writing classes at Ryerson at its various stages to show, a) what a poetic project looks like at its different points (even if its something as potentially esoteric as a chess-to-poetry translator); and b) to talk about conceptual poetry and what exactly students think a poem is (and does). Within those confines, students were inquisitive and acrobatic in explaining how the ChessBard fits (or doesn’t fit) with how they understand poetry. In a more recent digital literacies lecture, I tried to focus more on the conceptual end and work through questions of authorship, especially as their (and our) lives become more and more enmeshed with digital devices and languages— in what ways do we understand how a computer “writes” or “reads” or “thinks?” From this, I try to challenge them to explain whether what the ChessBard is doing is poetic or not. Some students, raised on older traditions of authorship as a harnessing of a unique feeling that is birthed and molded from a poet’s soul, bristled at the idea, but most went along with it, and, dare I say, enjoyed it. In fact, roughly 100 students played games of their own against the ChessBard, which I’ll be collecting (hopefully) within the next month or so.
EMN: What would happen if the organism housing Christian Bok’s “Xenotext” (D. radiodurans) played the ChessBard?
AT: I’m fairly certain that’s the prequel to the movie Her.
JM: I would die a happy man if I get to see ChessBard vs a. D. radiodurans bacterium in some kind of televised exhibition chess match.
There are so many interesting combinations here. Growing biological sheets out of chess games, for example, is a fantastic thought. There may well exist a poem you could grow and wear like an organic hat.