“If you don’t fuck with me don’t be around me.”

– Lil B

Dudes, I’m sorry. I can’t take discussions of literary taxonomy seriously. I enjoy saying obvious things, and this is one: Each piece stands on its own. The most woodsy/introspective water! light! rain! poem can be pretty good sometimes. A harsh text concept bomb is often harsh in a bad way. There’s good and bad art and cool and uncool people all over, and on the whole I feel OK about the aesthetic scene. We have Netflix, for one thing, which is amazing. Colbert is really funny. Hey man, have you seen Party Down? Have you seen Ghost Dog? Hey have you seen True Detective? Have you seen Planet Earth?

What were we talking about? Right, conceptual poetry. I, for one, am not interested in questioning/challenging notions of originality, creativity, or authorship. It’s 2014. These battles have been settled for the moment. I reckon the culture has caught up to our ‘esoteric’ ‘appropriative’ structures over the past decade in profound ways. ‘Everything’ is ‘free’ on Tumblr and everywhere else people post quotes, PDFs, screen shots, image macros, videos, etc. Even closer to pure conceptual ideals are the proprietors of in-browser pirate TV stream sites, Scene rippers and EPUB sharers, mass data leakers, and the most prolific literary artists of all damn time: spammers worldwide. (Spam appropriation by poets is A-OK, but we need give credit where credit is due.)

A cool thing about conceptual writing is that you can take a picture of your dick, download several dozen others’ in 2 minutes, grid them up, title it “One of these cocks is mine,” and that’s a landmark piece. You can attempt to lift every object in your apartment using your dick, record the results, and that’s a landmark piece. You can use a crowd of low-paid workers to translate Moby Dick into an emoji sequence, and that’s a landmark piece.

Pretty much all ‘good’ poem art (text-based or otherwise) is really silly and insular. When I say silly, I mean its benefit can’t be quantified neatly. It is fun. It is arch, hilarious fun for cool aesthetes. It is a dope pastime.

You know what else is a dope aesthete pastime? Literary taxonomy. It’s fun to draw formal distinctions and say who’s in and out of what camp/classification/whatever. But every category is entirely arbitrary, and every interpretation of a shared category is itself arbitrary. I am a member of the set of poets named Steve, along with Steve Benson and Steve Roggenbuck. We are a movement.

What is Flarf? Who is a Flarfist and who isn’t? I just made a poem using search results and it’s cute and a little upsetting. Is it Flarf or is it just Flarfy? Or is it flarfy? Most of us maintain a mental list of who is/was on the Flarflist and refer to similar works by others as generically ‘like Flarf.’ It’s a slightly touchy thing. But poets in the Netherlands, apparently, never picked up on that limited use. In 2009 the Dutch publisher De Contrabas released Flarf, een bloemlezing (Flarf, an anthology), including works by Ton van ‘t Hof, Mark van der Schaaf, and six other writers who aren’t ‘Flarfists’ as we use the term. In 2012 the Dikke van Dale Dutch dictionary included “flarf” as a lowercased noun.

Are the Dutch in the wrong? Of course not. There isn’t money to be made from the Flarf brand, so there’s no reason to defend the trademark. So, further, is there anything wrong with a naive undergrad referring to his/her poem as ‘flarf’ rather than ‘Flarf-ish’? Not really, though it’s something you might correct if you’re that kind of pedant. Flarf is different things to different people.

So defining conceptual poetry here in some attempt at authoritativeness is not only pointless; it’s also boring. We’ve all had these conversations before, and we like the books, collections, and magazines that we like. That’s that. I really don’t care about the fray – the long FB threads, ‘that article’ that ‘everyone’ is talking about this cycle. We can declare the death of this and that until we turn graygreen, and we can scold each other and be aggressive on Facebook, but what’s the point? I don’t know if it’s my distance from the university teaching market or what, but fights over poetic politics seem completely empty. Who cares what other people like? Poetry is the opposite of money.

We can be more constructive. We can still be friends. Be my friend?

Now, in the spirit of collegial good times, I’m going to talk about conceptual poetry and its place in the literary landscape. I won’t, however, use the term again – both because it’s so fraught and because at this point its repetition is grating. Instead, let’s start by sorting works based on objectively quantifiable measures. Text length is a quantifiable measure, as is word count. Character frequency and word frequency are objective measures; an English text will contain more instances of the sequence “the” than a text in Spanish. If a text contains the sequence “linnet,” it is more likely a Romantic lyric than an Imagist poem. Trying to make genre distinctions based solely on objective measures is a fool’s errand, but indulge me in this bit of silliness.

We’ll continue by considering a less trivial measure: information density. The text string “aaaaaaa” contains less information than “Ishmael,” though they’re the same length. The string “abcdefg” falls somewhere between. We can write a short script that says “Print ‘a’ seven times,” or “Print ‘a’ and iterate through the alphabet until you reach the seventh letter.” For that matter, the string “ishmael” can be packed into fewer bits than “Ishmael” if we use an encoding scheme that doesn’t include capital letters.

These are theoretical distinctions. In practice we’re all using Unicode or ASCII to represent text, so every character in a non-compressed text file takes up the same amount of space. (This isn’t entirely the case in Unicode, as UTF-8 and UTF-16 use variable-width encoding – multiple bytes for less common characters – but it’s close enough).

Any ‘normal’ English text – i.e. language written by humans – is about 50% redundant. That is, it can be losslessly compressed to about 50% of its original size via Huffman coding and/or LZ77 compression. But a ‘weird’ text can contain more or less information than plain prose. A truly random string of alphanumeric characters cannot be compressed. If there’s no redundancy, there’s no possibility for compression.

A text that contains a lot of repetition, on the other hand, can be compressed by more than 50%. To illustrate, I’ve just created a document containing 70,000 repetitions of the sentence “I will not make any more boring art.” That’s 560,000 words, on par with The Anatomy of Melancholy. My uncompressed text document comes to 2.59 Megabytes, while a ZIP archive of the file is only 7.75 Kilobytes. That’s a compression ratio of 334 to 1, and we could have done better with a simpler scheme. That 7.75 KB includes some extra baggage, i.e. various metadata.

Rob Fitterman’s This Window Makes Me Feel and Josef Kaplan’s Kill List are two examples of works that can be compressed by more than 50%. Nick Montfort’s 40,000-word palindrome The Utterance of the Petulant Child is sparser yet: “Mom, Mom, Mom, Mom, Mom, Mom, Mom ... Mom, Mom, Mom, Mom!” Cassandra Gillig’s Such Chainz begins with the lines “2 Chainz / 3 Chainz / 4 Chainz / 5 Chainz” and continues for hundreds of thousands of iterations. These works by Montfort and Gillig contain practically no information at all. Each can be fully explained in a single sentence. If their texts are someday lost, it will be easy to reconstruct them perfectly.

Does Such Chainz make for a rich reading experience? Would I curl up with it on a chilly evening and ‘get lost’? No, of course not. But still, I like it. It’s a good piece. Will I purchase a copy? Probably not, though if Cassandra gives me one I’ll display it proudly. Does it bother me that the piece isn’t commercially viable, that Gillig can’t make a living through this kind of poetic labor? No. It’s just a pastime.

As long as we’re throwing around measurements, let’s bring them together. Let’s imagine mapping individual literary objects on a coordinate plane based on observable features. On the x-axis we’ll plot text length, measured in characters. On the y-axis we’ll plot total information content, measured in bits. If we create a scatter diagram for many, many texts, we’ll end up with a big cluster of ‘normal’ texts forming a diagonal band extending up and to the right. If a ‘normal’ text is 10,000 characters long, we can expect its information content to look something like this: 10,000 characters * 7 bits (in ASCII) * 50% = 35,000 bits, or 4.375 KB. If a 10,000-character text contains only 1 KB of information, then it will fall in the sparsely populated region south of our diagonal band. Likewise, a string of non-compressible random digits will fall north of the band.

If someone else were to create such a scatter diagram of ‘literary’ works, I would make a point of investigating the lonely outliers. What’s more, if I saw a pattern of works with similar densities outside the main band, I would be curious about where they came from and what they have in common. If I saw a few really long texts off in the distance, I’d want to check them out too. That’s just me.

But hey, why stop at two dimensions? Let’s add an axis for word count. This measure will correlate strongly with character count, but there are sure to be aberrations. We can also add an axis for ‘number of line breaks,’ which in conjunction with word count will let us roughly sort verse from prose. We can add another axis for ‘number of words that don’t appear in a standard dictionary.’ Let’s also tack on axes for the geographical coordinates of the work’s composition.

Since we’re spitballing, let’s include qualitative axes as well: ambiguity, ambivalence, irreverence, humor, sincerity, dreaminess, referential opacity/transparency. (Props to Charles Bernstein’s “Poem Profiler,” a teaching tool listing several dozen criteria by which poems can be described.) We might as well throw in ‘quantity of text copied from elsewhere’ and ‘degree to which text was composed in collaboration with software.’ We might even include a measure for ‘level of information contained in the work’s generative spec sheet.’ In those terms, Kenneth Goldsmith’s Day is a work of low-information writing. We can say “The New York Times, September 1, 2000, all text from upper left to lower right, every page,” and, as long as the Times archive is still around, a future editor could do a decent job of reconstructing the full book.

What we’d end up with, if we had the patience to build such a detailed model, would be a huge scatter diagram in high-dimensional space. If we wanted to, we could easily come up with hundreds or thousands of criteria by which to measure a single work. This, clearly, is nothing new. For Spotify and Pandora, high-dimensional models are a useful tool for guessing what you want to hear next. For Facebook, high-dimensional models help determine which of your friends are in the photo you just posted. They’re used in all sorts of textual analyses by cutting-edge humanists and linguists and market researchers, etc. etc.

The reason I’ve headed in this direction is that ‘visualizing’ a multidimensional literary landscape liberates us from the reductiveness of ‘genre’ as we tend to conceive it. It allows for greater subtlety – and it allows us to account for works that fit most of the criteria of a conventional genre (e.g. ‘poem’) but are set apart by one or two noticeable differences (e.g. incredibly high word count or ‘written by computer’-ness).

If we consider the entire multidimensional landscape, however, we’ll see that it contains huge voids where ‘impossible’ or ‘horrible’ works would lie. For instance, a ballad consisting of 50,000,000 human-written lines is unlikely. It’s possible, and I’d be impressed if it were achieved, but it’s unlikely. A ballad of 50,000,000,000,000,000 human-written lines is impossible. That region of our imaginary map will lie bare forever.

Richard Dawkins used a similar high-dimensional model to describe the process of species evolution in his 2004 book The Ancestor’s Tale:

In the multidimensional landscape of all possible animals, living creatures are islands of viability separated from other islands by gigantic oceans of grotesque deformity. Starting from any one island, you can evolve away from it one step at a time, here inching out a leg, there shaving the tip of a horn, or darkening a feather. Evolution is a trajectory through multidimensional space, in which every step of the way has to represent a body capable of surviving and reproducing about as well as the parental type reached by the preceding step of the trajectory. Given enough time, a sufficiently long trajectory leads from a viable starting point to a viable destination so remote that we recognise it as a different phylum, say, molluscs. And a different step-by-step trajectory from the same starting point can lead, through continuously viable intermediates, to another viable destination, which we recognise as yet another phylum, say, annelids.

One must, I’m aware, tread carefully when using ideas from science and math to describe the unruly worlds of art and literature. For one thing, written texts don’t evolve the way animals evolve. They aren’t alive and they don’t reproduce. Similar patterns may unfold over the centuries (e.g. poetic forms arising, morphing, splitting, grouping in islands, reaching terminal borders), but that’s as far as I’m taking the connection. The model Dawkins uses is general enough to describe similarities and differences among any large set of distinct objects over time. Given the right parameters, we might observe an island of Imagist poems in the same neighborhood as an island of Objectivist poems. Over here you’ll notice several thousand closely related language poems, abutted to the north by a mass of elliptical poems. No one seems to agree where the border should fall.

That’s because there are no borders, just scattered peninsulas and islands of affinity spread across a vast, empty ocean of non-viable literary forms. There are continuities and discontinuities, gradients from one genre to another in some places and abrupt hops in others. There are literary forms which seem dissimilar yet nonetheless fall close together when considering a subset of our dimensional portfolio. There are occasional schisms, points at which populations of writers end up moving in different directions across the map, populating dissimilar frontiers. And still, there’s a temptation to yell across the gulf. “Hey! Hey! What are you doing?! You’re doing it wrong! Poems are made this way, dude! Dude! Hey! Hey! This way! We should talk about social causes! The collapse of capitalism! Climate change! Poverty! War! There’s so much important stuff to talk about over here! Hey! Hey! Do it this way! Knock it off!!!” And I’m sitting at my kitchen table in Concept Land thinking, “Who are these people? Why are they yelling?”

I’m going to do me, and you can do you. As for which aesthetic course to follow, I often think of Antin’s summary of an argument from Discourse on Method: “i believe in taking descartes’ advice if youre lost in a forest and you have no idea which way to go go for it straight ahead because its not likely to be any worse than anything else.”

So let’s choose the new writing we like and share it with others. Let’s keep making anthologies and magazines and hosting conferences. Let’s make essays and monographs and book series and libraries. Let’s hold more readings. Let’s do more interviews. Let’s make luxury editions. Let’s invent terms and start conversations and chip away at the unexplored darkness that surrounds our spot on the map. Let’s keep our tools transparent so others can learn by example. Let’s plan big. Let’s plant big, wet knowledge kisses on each other’s foreheads and go roll around in the info leaves. Let’s rear digital moonlight.