Andrew Branch

“A Robo-Poet Does Not Scare Us”


Read afterthoughts to this piece from Ryan Walach.

Fat Robot: The Efficacy of Ad Hominem

It is an underappreciated irony of development since the industrial revolution that automatization has preceded the automaton. While horseless carriages waited for the invention of the engine before disposing of the quadruped, labor markets have been eager for a shortcut: if the robots long-considered slow at replacing humans will not come, humans might instead be coaxed into devolving, and working in ever-more-machine-like ways.

The hallmark of this drive to produce robots through psychology rather than metallurgy is the production line. Literature, however, has by no means been spared. In the long-untranslated 1953 article “Un robot-poète ne nous fait pas peur,” Boris Vian quips that Victor Hugo could be considered a robo-poet; an even more extreme example would be that of book packagers such as the Stratemeyer Syndicate, whose armies of in-house writers pumped out repetitive children’s titles under a few pseudonyms-cum-human-brands. Pulp genre writing, however, was not Vian’s real beef; it had been around for at least a hundred years when his article was published, and would have made an untimely a target for aesthetic invective. By 1953, mindless literature had come and gone from high culture chic (surrealism made a movement out of the practice of writing “automatically”). Less than a decade before the first stirrings of the OuLiPo movement, art without thought was poised to return. The robo-poets in question were not machines, but single-mindedly formulaic human writers.

The allegorical nature of a 1953 article purporting to be about poetry-writing robots is likely unsurprising: although Vian cites real technological advances, literal robo-laureates were then, as now, far off. To understand who the human perpetrators of robo-poetry in the French literary scene of the early 50’s were, it is necessary to acquaint ourselves with the background of Boris Vian. In the small part of his diverse life that was devoted to literary journalism, his primary critical project was making fun of Jean-Paul Sartre.

Vian’s great coup was to have most of his writing about Sartre published in Les Temps modernes, the avant-garde magazine at which Sartre occupied a variety of roles and of which he was always considered the intellectual heart. The following excerpt from a list of suggestions for the improvement of Les Temps modernes, which Vian published in Les Temps modernes, should give an adequate sense of the sort of irreverent eye-poking he specialized in:


6. Insertion of photos of pinup girls (but good ones, not just any old ones).

7. Publication of a few articles on existentialism, which is, it seems, a new style of dress that is currently quite fashionable.

The sarcastic suggestion that Sartre’s magazine needed to increase its coverage of existentialism, and the dismissal of the same as a sartorial fad (one cannot help but suspect that Vian would have made this pun if the word existed in French) demonstrate the level of critical discourse in question. Vian was emphatically unwilling to rise to Sartre’s (pretentious) level, and preferred to hurl abuse from a proverbial peanut-gallery he somehow managed to relocate to the main stage of the show itself.

To situate the rebuff of robo-poetry that we have translated below in Vian’s wider corpus, and to identify Sartre as the robo-poet, casts light on the vague but strident affirmations of “Un robot-poète ne nous fait pas peur.” Its assertion that “we can think, therefore be, and precede essence itself” is a détournement of the famous existentialist claim that existence precedes essence: what Sartre proclaims as the human condition is in fact a limitation of robo-poets, which is to say of Sartre as a thinker.

The term “destructive criticism,” employed by Vian in his aforementioned Temps modernes article, seems an apt label for his treatment of existentialism: the positive formulation he opposes to “existence before essence” is not as obvious as his contempt. What would it mean to think before being in a poetic sense? The simplest answer is to take poetic “being” as a metonym for the practice of poetry and to understand the article as an accusation that existentialist fiction seeks to reignite the Surrealist ideal of writing without thought; existence before essence becomes prose before reflection. If that is indeed Sartre’s project, Vian seems ready to concede its success:

…in the realm of the vague, the bizarre, the ephemeral, the abstruse and the dream-like, the robot will beat us every time. It, in effect, will pick one term or another free from the influence of a lived past, which imposes bad reasons for choosing upon us humans. It will be truly free, whereas when an especially original structure sprouts from our pen, it’s probably just because we’ve become a little too intimately attached in our readings of Mallarmé or Jarry.

The suppression of conscious thought does, apparently, lead to more original writing.

The discussion of an individual’s “lived past” and “reasons for choosing” as aspects of the artistic process makes the target of Vian’s ire explicit: consciously choosing one’s words to evoke one’s experiences is precisely what Surrealist writing rejected. Yet Alfred Jarry’s famously sui generis Ubu Roi, invoked in the last sentence, was first written as a critique of an overweight and over-pedantic teacher whom the author encountered during his own schooling. The first blow to the robo-poetic ideal of mechanical creativity amounts to an argument from authority: to claim that originality must be produced by strict refusal of personal experience is to claim that Alfred Jarry was not original.

Of course, by popular account Jarry’s work was extremely original. It so departed from the strict rules of classical French theatre (and the more subjective confines of good taste) that it remains notoriously controversial to this day. Thus, Vian’s final admission, that his ideal writer might be derivative of Jarry, is almost oxymoronic. Even the formulaic imitation of art as shocking as Jarry’s would likely produce great agitation in an audience; to deny such reader response is to prefer a strictly formal understanding of artistic originality that amounts to detached academic nitpicking. Such is the world-view of the robo-poet: s/he considers originality in the way a computer program checking undergraduate papers for plagiarism considers originality, and congratulates herself on simple variation in word combinations. To imitate Jarry mobilizes a truer form of originality in that it invokes in an audience the sense of awe before innovation.

Besides his reference to Jarry, Vian’s major examples of innovation all hinge on the idea of the person as polymath. Notably, the passage that parodies existentialism’s central slogan simultaneously promotes Descartes’ Cogito and a list evoking the diversity of human endeavor: “We can make love, read, play the piano, swim, and even construct robots. We can think, and therefore be…” This association of dilettantism with Cartesian dualism is a distinctly original gesture in aesthetic philosophy, which posits that material existence and specialization are both hindrances to full intellectual expression.

A machine, and particularly a machine of the 1950’s, is intimately and obviously bound to physical processes in its production of information. This makes it ineligible for selfhood in the Cartesian sense: a punched-card computer can never be “a substance whose essence or nature is only thinking, a substance which has no need of any location and does not depend on any material thing.” For Descartes, the division of mind and body is necessary for a being to be certain of its own existence. Yet Vian is neither a robot nor a philosopher interested helping robots prove their being to themselves. For him, the robot is an other whose function is the production of thought, and its flaw is that it can never present its essence to its reader as anything but a consequence of its existence. Thus, Vian interprets Descartes against the grain: it is not that thinking proves being, but that if the being of texts is allowed to take precedence over the thought of their makers, the possibilities of mental expression become limited and over-determined. A machine for writing poetry might be defined by the fact that it writes poetry, but good poetry should be defined by the fact that it is the reflection of something more expansive than a poem (a mind).

The limitations on robot thought, we have seen, do not apply to the poetry it produces. While Shakespeare and a computer might turn out identical sonnets, the fact that Shakespeare might subsequently go for a swim, whereas the robot cannot, is important. Such an objection to robo-poetry recalls romanticism, insofar as it privileges artistic genius as a process over art as a consumable product. The idea of the poet as an inspired being, and the popular attribution of his output to his idiosyncrasies, is reconceived not as a para-artistic phenomenon but as art itself. A robot might, for example, produce a Van Gough painting, but it would be unlikely to cut off its own ear, and even if it did so the gesture would be unsurprising, because it would be a function of design and programming. This, in turn, would mean that the robot’s identical “Starry Night” canvas was not art. For Vian, that recurrent question of amateur reading (“what was the author’s intent?”) is a legitimate component of literature, which the robot can never achieve, because a program is fundamentally deterministic. The robot might be perfectly expressive, but to an audience its expression is too perfect and certain to be considered intellectual.

The human robo-poets, then, are those who attempt to suppress intent, individuality, and error from their own work. Any sort of constraint-driven writing meets this criteria: the first reference to Hugo likely draws on his over use of the alexandrine as a strict poetic meter. The case against Sartre finds less formal support, but still relies on the idea that the author’s thought lacks the infinite possibility that characterizes human minds; this limitation makes his writing predictable and uninteresting no matter its actual content.

Vian’s most extended critique of Sartre appears in his novel L’Ecume des Jours, where he satirizes a popular philosopher called, bluntly, “Jean-Sol Partre.” The character Chick makes the mistake of joining a sort of cult of personality around Partre, when a personality is precisely that which Partre lacks: “How would one not be interested in a man like Partre?... capable of writing anything at all, on any subject at all, with such precision.” These qualities, which Chick lauds, directly contradict Vian’s ideal of human essence as expressed in his rebuke of robo-poetry: “We are not perfect, but we are very adaptable.” Sartre’s perfection and precision make him inadaptable, in spite of his wide range of interests, because he moves from subject to subject without himself changing. He is bound by the true existence of the things he writes about, and lacks an independent intellectual essence that could allow a reader to speculate as to why he writes what he writes. His existence precedes his essence in the sense that the texts he produces cannot be informed by considering the nature of the producer behind them. Like a robot, Sartre is nothing more than the producer of texts signed “Sartre.”

Vian unapologetically embraces the intentional fallacy, and Sartre is a terribly uninteresting author for those interested in reading authors into their works. Returning to the parodic version of the philosopher: “[My book] will be quite boring to read, said Jean-Sol Partre, because it bores me a lot to write it in the first place.” Sartre would not be bored, we are led to imagine, if he were a worse writer, and an amateur pianist and swimmer as well. If he were a being for whom thinking truly preceded his texts’ existence—someone, in other words, for whom thoughts weren’t merely the raw materials of textual production—his writing would be a person’s rather than a robot’s.


Vian’s strategic attitude toward the problems he seeks to remedy also commands our attention. Despite contextual clues that make existentialism an obvious reference, his article is, at heart, a diss track against an artistic school that goes entirely unnamed. It is as though the certainty that history will forget the robo-poets produces in their critic a proleptic amnesia that prevents mentioning them individually or in detail; this makes for an inspiringly blasé dismissal of literary fads that fail to deliver substantive innovation. The insult is compounded by the climactic exclamation: “Get lost, FATTY robot!”, which recalls Ubu and the portly high school teacher on whom he was based. Vian’s criticism culminates in a moment that is at once derived from a personal experience (Jarry’s), derivative, and a slap in the face.

Whether or not we agree with Vian’s criteria for judging what constitutes aesthetic “substance” is largely irrelevant, and to dwell too long on it is to miss the point of the article: “Un robot-poète ne nous fait pas peur” is not a study of the robo-poets, it is a manual for their destruction. The method is simple: “explain to it how it functions, and you will have it humble and at your mercy.” Debating the merits of existentialist aesthetics is irrelevant: to identify its underlying principles is to invalidate them, as though they were a magician’s tricks. This critical principle is derived directly from the tenants of the robo-poets themselves—all their striving to write without analyzing their own writing practices crumbles before an act of interventionist criticism: thinking about and explaining their writing for them.

In short, when Vian says to Sartre: you are a boring nouveau-surrealist, it is an attack—not only because of the undesirability of being boring or a surrealist, but because of the undesirability of being explainable.

A Robo-Poet Does Not Scare Us

There you have it, dear Parinaud: the dangers of the demi-culture. You need only read a morning paper and see that M. Albert Ducrocq has constructed a robo-poet to be immediately shocked. Then again, is there anything extraordinary about it? The last century had Victor Hugo. And so?

Please note that I have no idea how it works, this robot of Ducrocq’s. But I know that one has the right, and even the duty not be surprised by this sort of news, at least since the advent of electric tortoises1 and Ashby’s electro-thingamajig (I’ve been trying to find the name of this machine for three days, but contrary to the assurances of Charles Trenet2, I remember authors’ names much more easily than those of their inventions). There are now tons of machines for selecting various things in relation to tons of others things, in ways that we find obscure. They end up producing some sort of personality (or perhaps a sort of liberty, if you prefer; you know, Parinaud, I’m no snob).

Let us suppose that Albert’s robot, unlike Ashby’s, does not choose a resistor through its internal workings (I believe that that’s what Ashby’s “...” does. There was a certain Wheatstone involved, if I remember correctly). Let us suppose, then, that it chooses words from a word bank that has been provided to it, and exclaims them in a great nasal voice, or states them more discretely in binary language, which would then be translated. As Albert has a gift for electronics, he has even been able to get his robot to respect certain rules of good composition (he’s no simpleton, you know, Parinaud, he also writes really good books without the help of robots). Suppose now that hateful electronic limitations are imposed upon this poor thing such that it must first produce a subject, then a verb, then a complement: at that point the Robot would be making sentences.

But come on, Parinaud, you’re a child to be dazzled by this, when the most minor writer does the same thing all day long! You know translation machines are being studied. Imagine someone writes: “I’m a little green rabbit” and translates the sentence into French. Replace “I” with another subject, “am” with another verb, “a” with another article, et set-era. Provided that your conversion table is clever, you’ll get the better of Malcolm de Chazal3. A machine could do it!

There is a point, however, that you must not forget: it is relatively easy to get a robot to do all that, on the condition that you provide it with neatly pre-made words. Ah! I see you thinking you’ve triumphed over the machine, and I can already hear you exclaiming: “But Larousse4 is the poet!” Tremble, Parinaud. If the robot were provided with letters, it would just as happily make words.

We wouldn’t necessarily understand these words. The collection of fabricated words would be many times more complete than the miserably embryonic vocabulary used by repressed human wordsmiths. New letters might even be invented (the newspaper article indicated as much). The possibilities for a robot are endless.

You see what all this is leading to, and you are nervous. With good reason. To extricate ourselves from this bind it is important that we give an extremely precise meaning to everything we write; in the realm of the vague, the bizarre, the ephemeral, the abstruse and the dream-like, the robot will beat us every time. It, in effect, will pick one term or another free from the influence of a lived pasts, which imposes bad reasons for choosing upon us humans. It will be truly free, whereas when an especially original structure sprouts from our pen, it’s probably just because we’ve become a little too intimately attached in our readings of Mallarmé or Jarry. The robot will exhaust all possible two-tense and tri-partite combinations and deliver us texts without syntax, for which it will assume sole authorship.


On one hand, we could try to master the robot by being as rigorous as possible ourselves. On the other hand, we could attempt to create alternative logics that it would be completely unable to grasp. Creating alternative logics leads, unfortunately, to apparent chaos (relative to the current “order” of things). And we, mankind, know contradiction only by hearsay, for we have at our disposal neither simultaneity nor singularity. A robot, however, can have twenty simultaneous functions and still keep a happy heart, and thus it breaks our balls once again.

But the trouble with our two solutions is that they are contradictory, which, as I said, humans don’t handle well.

To provide a bit of reassurance, Parinaud, I will, despite everything I’ve already said, remind you of the fundamentals. After all, the functioning of a robot depends on what we put into it. And even if it has free will, it’s only because we intended it to have that. If it is a poet, it is only because Albert is a poet-maker. Isn’t that the better thing to be? Ducrocq, maker-of-poets. To think that others become soldiers, or, barely better, butchers.

I know this letter is tiring you, my friend. It’s a bit heavy. The style is tortured to the maximum degree. Forgive me, because you know my reason for doing it: we don’t want just any old robot to see itself reflected in our profound meditations, and any shortcuts in form and reasoning risk producing such a result. We are ultimately struggling in Vian5; realize that, sooner or later, robots will do things that we will be unable to do. There’s only one thing for it: let us neglect everything else and cultivate our versatility. Some robots will be poets, others chefs, others mathematicians—fine. But to be all three at once what a volume they would need!6 We are not perfect, but we are very adaptable. We can make love, read, play the piano, swim, and even construct robots. We can think, and therefore be, and precede essence itself. We can laugh. Oh, I don’t deny it, some robots will laugh better; but certainly not the same robots that cook, and do math, and the rest.

The world is in the grips of theories propagated by scoundrels who want to turn us into workers and, worse, specialized workers: Parinaud, we must refuse. We must know everything; know what’s in the belly of this robot; be specialists in everything. The future belongs to Pico della Mirandola7. Mirandola-ize! Belittle this robo-poet using your knowledge of cybernetics, explain to it how it functions, and you will have it humble and at your mercy. In order to do everything you do—provided you were well raised—this robot would have to weigh tens of tons, the poor thing. So let it come, and in a scornful voice, with a lofty look, shout at it: “Get lost, FATTY robot!”

No being with feelings can resist that sort of thing, and a robot that wants to lose weight is screwed, because robots don’t waste away like us, by losing mass. They become weak, anemic, and in one shot they break down; and they repair themselves, like crabs8. One last piece of advice: don’t drive yourself crazy. When the world is full of robots, there will be nothing so easy as to invent one designed to hate its own species. Thus, we’ll become Neros, and our hands will be clean as we play homemade fiddles while we watch the robots’ sheds burn at our feet while they writhe in the embers like uppity ants, as we listen to the majestic accompaniment of a song composed by a two-year-old juggling prodigy raised between the paws of a tiger outside of the civilized world.

1 Not, as Vian’s general whimsy might suggest, an authorial invention but rather a popular wind-up toy of the era.

2 It is not wholly apparent which of Trenet’s works Vian has in mind when he makes this remark, but the popular song Souvenir d'un chanteur à voix is a meditation on nameless yet stylistically memorable “singers one often hears on the radio.”

3 Chazal was famous as a writer of aphorisms, and his mention sets a low bar for the machine's abilities: it might produce some good sentences, but not a Proustian opus.

4 The eponymous name of a standard French reference dictionary.

5 The original reads “nous luttons contre des moulins à vian,” a pun where the author interjects his own name into the expression “lutter contre des moulins à vent.” This French expression, derived from Don Quixote, means to fight imaginary enemies, and by extension to fight an unwinnable battle.

6 We might think that Vian’s remark, which apparently refers to the famous computers of mammoth proportions in use during the 1950’s and 60’s, has been obviated by microprocessors; but if we stipulate that the Google search engine is the epitome of computational multifunctionality, googling “Google server farm” should quickly disabuse us of the notion that Vian was wrong to say versatility comes at the price of space taken-up.

7 A Da Vinci-esque epistemologist of the Italian rennaisance, who was both a promoter of the human quest for knowledge and the author of a collection of eclectic theses, which he believed could provide a basis for all knowledge.

8 Vian might have found it amusing that the self-healing properties of crab shells have now been co-opted by the automobile industry in an attempt to develop paint that grows to cover scratches automatically.