In this bumper SPAM Cut, Hannah Lee Nussbaum responds to two poems written by Heather Christle — 'The Running of Several Simulations at Once May Lead to Murky Data' and 'Learned Has Two Syllables and I Only Have One' — published here as a diptych in Granta’s August digital issue. Working alongside Christle’s language, Nussbaum considers why we need weird lossy metaphors created by defective machine learning algorithms, the rotten state of metaphor in late-stage human language, the possibility of applying a reverse Turing test to your best friends, and why things feel so good when we are told they are only simulations.
> 'Learned Has Two Syllables and I Only Have One' is the first poem in this diptych by Heather Christle, although the poem isn’t a staccato mono-syllabic exercise, and there are some duo-syllabics and a few sneaky tri-syllabic words in there too. The title of the poem made me try to cram each double into a single (ma-chine as m’chine) and each triple into a double, but manually speaking, this was a total failure when I tried to read the poem out loud like this.
> Christle is conservative with syllables in this poem, because too many would ruin the machinic pace of the piece, would disrupt the leaden plainness of the language being used. Each stanza acts as a divestment from metaphor, and there’s a crystalline, Ouilippan thing happening here. Christle’s words are tautologically sterile, are trying to mean only what they mean, although this is difficult, because even the most ordinary language is metaphorical by nature. Like how happy is a metaphor for up and sad is a metaphor for down, which Natasha Stagg points out in an essay she wrote in 2016 called 'Internet as Horror', which I read around the same time as I read this poem.
> Christle’s clipped, germ-free language makes me think of Stefan Themerson’s semantic writing, which would not say ‘horse’ but would rather say ‘a solid-hoofed, plant-eating domesticated mammal,’ and it seems true that her language practices the kind of estrangement required when you are doing childcare. The kid will inevitably point to something, like a hair on your arm or a blemish on your face, and say what’s that, what’s that on your arm, what’s that on your face. And you, inevitably, will be really stupefied, stupid feeling too, because your sculpted human brain is trained in abstraction, not literalism, not low-level classification. It’s shame, it’s the abject body, you will probably tell the kid, and you’ll give them an orange cracker and pat their skull, and they will grow up surrounded by words and images which they have been taught to classify as broad political and cultural concepts.
> But when you speak like a computer, or to a computer — which is what Christle seems to be doing in this first poem — you must necessarily turn away from this human meat language of ours, which is supersaturated in allegory and metaphor. Ours is a language deployed towards symbol-heavy populist speechifying, where words are all units of metonymy, where words are constantly circulating and evolving like memes, where each single syllable is pregnant with history and culture, and in fact I would even argue that each syllable in its own right acts as a tiny self-driving metaphor, a little sound island that makes us think of this or that. In late-stage human language, meaning has gone totally viral and each sound has a thick crust around it. Context accumulates and accelerates as words are repeated. Late-stage human language is apparent in terms like “globalist,” “states’ rights,” “locker room,” “inner city” — all saying and not saying, all totems of the way metaphor and abstraction have ossified our words into compact, lazy symbols. Christle’s poem pokes at what’s at stake in moving backwards or outside of our lazy regime of abstraction. Which is, certainly, what is required if we are to approach the problem of machine learning as it relates to language, which is, on the level of content, what this poem is directly about. And maybe it’s not a problem, I’d like to add, but an opening.
> I imagine that she originally wrote a decadent and highly pigmented poem, or at least thought of one, then took a palette knife to it and attempted to strip it back into its constituent zeros and ones. The difficulty — the one she calls a readjustment/no more/painful than/a thicket — is that all of this abstract, high-level information is lost when you strip an image or a word or a whole poem down to its back-end code. New metaphors are created. Metaphors of misidentification, of confusion. An imperfect algorithm might accidentally categorize a red robin as a smear of blood. A long man’s face might be sorted into the column category. To backpedal words into non symbolic, bag-of-formal-qualities territory creates a moment of reverse emergence — concepts are stripped back into their constituent aesthetic facts, and a culturally innocent machine might well make connections between these constituent aesthetic facts, and new, stranger metaphors will be the upshot. The result is inevitably a weird realism, a Picasso-esque reality, which sounds sensual to me, do you agree?
> Said another way, these new lossy metaphors produced by still-too-dumb technology might help us make connections between seemingly disparate words, tease similar properties out of culturally dissimilar symbols. In this way, machinic metaphors might well be a tool of world building that verge on magick — they can make our sense and they can re-arrange it.
*
> If the activity of cutting language away from abstraction — doing language like a sort-of-dumb machine would do it — is the move happening in the first poem, Christle’s second poem does something tangential, cock-teasing but not straightforwardly delivering on my guilty desire for facile conceptual twinning between the two texts. The second poem — 'The Running of Several Simulations at Once May Lead to Murky Data' — moves with a protagonist who imagines her meaty human companions — the real ones she is eating dinner with around a real table with real salt shakers — to be virtual (or machinic, or programmed, or otherwise computationally choreographed). The protagonist invites us to join her in an uneasy case of pretending, an induced brain-game that makes everything look different. A reverse Turing test that coats real humans (which are typically bland and predicable social organisms) in the dazzling gloss of life-likeness (the amazement with which we hear a machine speak in a women’s voice, as though we have never heard it before, so clear, so feminine). The poem knows that when a virtual object is life-like (what verisimilitude!), it is eons more astonishing than the real thing, because the real thing is yesterday’s news.
> And so Christle’s brain-game (this poem) allows real things to take on the glittery mystique of the virtual or the simulated: artificially I will induce this feeling in myself, the speaker tells us, pretending/—until it is real—that each person/is speaking from a highly naturalistic script,/having carefully rehearsed each/tiny gesture. What intrepid attention to detail! What finely tuned mockups! I am reminded of a short story written by Ben Marcus in 2013, Notes From the Hospital, in which Marcus describes a hospital on an island — a fastidiously fashioned space in which the air is breathable, the scale is one-to-one, and even the most advanced scrutiny cannot reveal the setup to be constructed and forged — so close is it to an actual hospital, with all of the bodies and walls and smells therein. But this hospital isn’t real, we are told — it’s made by a technically masterful artist — and so the thing feels miraculously life-like, accurate, while still retaining some of the impossible and strange and utopian feelings we associate with and assign to things we know to be virtual.
> Simulation is at the very centre of what poetry is, in the sense that poetry is always a necessarily really inadequate representation of the thing the poet originally tried to evoke. A poem on a page is always a simulation of an original ghost poem, and in this sense a poem is always a record of failure, says poet and critic Allen Grossman (to whom I was led by Ben Lerner, who also writes on this). The actual poem is a failure, but the virtual poem (the poem the poet meant to write) holds within it that feeling of immense potential, the deep, instinctive sensation of a yet-to-be-executed idea in all of its impossible perfection — a schematic, a model, a mockup, a prototype: all perfect ghosts prior to the flaccid not-quite-right of real life execution. Virtuality itself is a way of feeling, a way of looking, and simulation is a sensation, and the sensation is generous, hopeful, rapt.
> The ginger minutiae of social tics and turns, the remarkable talent it takes to speak, to reach for the salt, to be alive (being alive not being the norm) — all of these signals are consigned to the filing cabinet labeled ‘actual’, and so we don’t see the poetics in these small and ceaseless triumphs. Perhaps it takes a brain game, a reverse Turing test, a clever cranial experiment, a poem that writes around these little moves, to entertain the possibility of the actual poem being the virtual poem, the actual friend as the finely tuned machinic replica, and with such skill! Producing perhaps some awe for their remarkable talent/ for portraying with such detailed conviction/ the humans I know as my friends. Can the meat world shock and delight you as much as an imagined version or close approximation of it might? Does the long cold distance have to be so far?
~
Text and Image: Hannah Nussbaum
Published: 8/3/20