Subjectivity with AND|OR binaries… Just what is “poetic computation”, anyway?
I’m attending the School for Poetic Computation (SFPC) Fall intensive program this October through December 2016, in New York City. When asked to explain what this school was about, I’ve often lazily, even bashfully, given a poor, shorthand answer—something like, “It’s a digital art program”. I know this answer doesn’t adequately explain the school and the way it weaves together code and physical computing; radical theory and historical context; poetry and logic; community, collaboration, and exploration of alternative models of learning and teaching.
The school’s motto is “more poetry, less demo”. I’ve come to understand how this is meant. “Demo” is a way of referring to showing off what technologies can do—technology for its own sake, for its novelty, for a “wow” factor. “Poetry” can mean a lot of things, but in the motto it explains at least that the creative act is foremost, the technology and tools used for it simply in service to the expression.
I’ll take a closer look at a series of related—ahem—binaries to explore these two words, “poetic” and “computation” and see what they might mean to me, used together. I hope to return to these thoughts as touchstones for my work at the school, and future work. Hopefully they can be of some use to you, as well.
COMPUTATIONAL or NOT
At its most basic, computation is the act of taking and remembering an input, performing arithmetic and logic operations, recording answers, and delivering an output from those answers. The standard test of whether a system is computational is the Turing Test, named after British mathematician Alan Turing. Best known for cracking Nazi Germany’s Enigma cryptography machine, he also theorized about a machine that could express all logic. Mathematical proofs have shown that his model is indeed the minimum needed for a machine or system to carry out all logical operations. Basically, a system is “Turing complete” if it can: provide a means of reliably recording and retrieving values, carry out sequentially a very basic set of “if recorded value is A then record in its place value B” instructions, and keep track of position or “state” within the set of instructions. Nowhere is it specified that computation requires a screen, a mouse, a keyboard, a CPU, or even variables, functions, data types or features generally associated with computers and programming today. Just some forms of memory, instruction, logical operation, input, output. A basic calculator without memory is not a computational system, while a system involving (well, an infinite supply of) abacuses and a set of written instructions could actually be.
ANALOG or DIGITAL
Turing’s American contemporary Claude Shannon (who, incidentally, did his research at Bell Labs, once housed in the same Manhattan building that is now Westbeth Artists Housing and home to SFPC) was the pioneer in reducing all logic to electronic logic gates using arrays of transistors. Open or closed; allowing or restricting current flow; on or off; one or zero. Information transmitted via analog signals, subject to degradation due to noise from the line or medium it traveled in, could be reduced to sequences of binary digits that could be transmitted far more reliably. Take a look at Adam Westbrook’s (of Delve.tv) excellent short video on this revolution and how we came to see all information as representable with just these two numbers.
SIGNAL vs. NOISE, LITERAL vs. SUBJECTIVE
Analog recordings and transmissions are said to “suffer from” noise. Isn’t poetry and the poetic more of a subjective, noisy, or analog form of expression? Isn’t it the opposite of the objective, binary, perfect world of computation? Computers rest upon a logical foundation which must be objective, predictable, in order to function. The languages we have built to perform tasks with them tend to be relentlessly literal and exacting, as anyone who has written a computer program and hit a syntax error from a single character typo knows. On the other hand, a poem, poetry seems to demand from the audience a non-literal, subjective interpretation. So how, then, poetic computation?
COMPUTER-AIDED OUTPUT vs. COMPUTATIONAL PROCESSES
There is an obvious but unsatisfying and incomplete answer. It’s evident that computers can be tools for individual, subjective, creative production and interpretation. We produce images, video and film, animations, games, sculptures, with computers. We edit and collaborate on texts with them. We document our creative work, write novels, make digital photographs with ample room for subjective interpretation using digital tools. We creatively select, interpret and display data. Arguably, we multiply our effectiveness at all of this by using computers to extend our memory and our ability to reason about complex realities or ideas. But I don’t think these uses are necessarily poetic computation. Poetic computation is about bringing poetry and subjectivity to arranging and carrying out these core processes of input, memory, instruction, state, and output.
So, let’s leave the OR, or oppositional, binaries for now, and take a look at some pairs joined with AND. This is an exploration of the possibilities for poetry, subjectivity, and noise in the means of arranging and carrying out these defining functions of computation. They are of course not the only options for poetry and subjectivity in computation.
POETIC INPUT and NOISE
How could we allow for input to our computation system, that allows for subjectivity, “noise”, and exploration? Input to a computational system may be the means with which to enter data to be operated upon, or the instructions for the operations. We could introduce a “noisy,” analog aesthetic to the input of data or even instructions, allowing intentionally for corruption. The resulting instructions and data must be taken objectively by our instructions interpreter, but there are a lot of possibilities for interesting “garbage’ input. There is a whole aesthetic of “glitch” in the electronic art world. Or, we could make our input dependent on subjective actors like people or other living beings. Present and even historical, aggregated human activity is a rich source of subjectivity. Fo example, we might ask a room full of people to use words to describe an object and somehow combine the descriptions to make an objective, computationally exact set of data upon which to operate.
POETIC INSTRUCTIONS and MULTIPLE EXPRESSIONS
As with natural, or human, languages there are many ways of expressing the same thing with the programming languages we have invented. We create instructions for computational systems that ultimately must compile to combinations of logical operations. But, authors of instructions have all the leeway of our language to create programming languages or ways of expressing instructions. It is not at all non-computational to express instructions with shapes, sounds, or dance steps as long as there is an interpreter that can turn these into logic. In fact, 0 and 1 are not required as the representation of logical true or false in a Turing complete system. (I’m looking forward to participating in the “world’s slowest computer” sponge computation exercise with Ramsey Nasser later in the program).
Even with more standard programming syntaxes there are many ways to write, and even read, the same essential instruction. There is no reason that the instructions themselves might not be an interesting work of art with a layer of subjective interpretation beyond the compilation to logic.
POETIC ARITHMETIC and CHANCE OPERATIONS
There’s not a way to get subjective with logical or arithmetic operations directly, but computer-simulated near-randomness can open up some poetic opportunities. Chance is actually an objectively determined choice from a set of options, but the determination of how and how much to insert chance as a value in a system. There’s a very interesting interplay between authorship as a feature of poetic subjectivity, and chance as a major agent of non-exact determination in computational systems!
POETIC STATEFULNESS and COMPLEXITY—A computational system must reliably track an objective state, or know what is the current and next instruction. The need for reliability here reduces the wiggle room for poetry. Subjective state might allow for multiple or overlapping states. It reminds me of the idea of intersectionality, getting away from defining ourselves with either/or, of mutually exclusive definitions. A or maybe instead B, is not a valid computational state, but a D representing the same value as A and B and C is a valid computational state. A single state could capture the idea of two-thirds A and one-third B. Increasing the complexity of objective states provides a way to simulate and approach subjective state, and to better express the poetic.
Studies of emergence in biological, ecological systems, and even physics show that new properties can be emerge from complex systems and relationships that are not properties of simpler systems with the same set of actors. There is a sense that increasing the complexity and number of actors of an objective system actually introduces the possibility of scientific subjectivity. But I’m not going to go too far down the quantum “woo” path here….
POETIC MEMORY and CORRUPTION/ERASURE
People tend to think of poetry as an expression or exploration of a moment, as vertical, where prose is typically seen as horizontal, a sequence in time. Much of our use of computers has to do with long-term memory and data storage. There is an ethos of infinitude around this computer-aided memory, like we can store and recall everything from and for all time. All knowledge and activity, stored forever in “the cloud.” I like the idea of placing the idea of poetic computation in opposition or contrast with this sensibility or value system. It would be difficult to perform operations of much complexity with few bits of memory, but people found ways to be creative back in the days of the Commodore 64 and magnetic cassette storage devices. Making a system with very constrained memory might be a way to introduce subjectivity. What should be remembered, and what must be forgotten? Erasure, and corruption of memory, are ripe for poetry.
POETIC OUTPUT and SUBJECTIVE INFORMATION
Ultimately a computational system should produce an output to be usable, human readability is certainly not a requirement for Turing completeness! The human beholder is of course a subjective, and fallible, interpreter of data. Information is not data ,but occurs as the integration of data, or stimulus, by a subjective consumer. Some being is informed by the output of a system. We can display output from an objective, logical system through any kind of filter to allow for subjectivity. A step in the system might be for subjective beings to record some impression, which are then used as further input. A game of “telephone” as part of a computational system. In some ways, this may be the easiest way to conceive of subjectivity and poetry in a system. Ultimately, if we are making computational systems to communicate or produce something in the real world, the result will be interpreted subjectively. I think the key in thinking of poetic computation is to account for and work with this in creating the system.
PERSPECTIVES
Finally, I want to point to some questions and answers to approach “poetic computation” from other students and instructors in my SFPC cohort.
Dan Wei: Define poetic computation by what it is not. What is computation without poetry?
Taeyoon Choi: A transfer of untranslatable feelings; saying a lot with very little
Katrina Allick: Breaking complexity down to the basics to apply new values and potentials
Medhir Bhargava: a mental framework/approach
Ruby Childs: impractical computation
Agustín R Anzorena: “soft computing”. Poetic originates from the artist and is not “readable.” “Aesthetic” is an outer manifestation, is readable.
