When English and Computer Code Both Feel Like Foreign Languages
"I am Ill at Ease in a Room of People Speaking Quickly and Fluidly"
Exactly I and exactly the world
Fail to meet by a moment, and a word.
–Laura Riding, “The World and I”
Computers lack skill with the interface—language—that humans use to tame and conceptualize the world. I sympathize with computers in that regard. I was four when I first spoke in English. I was six when I first programmed in Logo. Brain development varies wildly among children. Math and science came relatively easily to me; human language has always been harder. Foreign languages come slower to me than to most. English feels no more like a native language to me than Logo and C. I am ill at ease in a room of people speaking quickly and fluidly.
Perhaps as a consequence, I have kept my feet in multiple social environs simultaneously, most often through a combination of humanities and technology work. I read Ralph Ellison while learning C, Robert Musil while writing a compiler, James Joyce while working at Google. The communities were complementary. At Google we could have a long discussion of New York subway optimization problems—where to stand on the platform for the emptiest car, when to give up on waiting for a train. In academia, abstract and abstruse discussion of the sociopolitical implications of Hegel’s master/slave dialectic were mumbo jumbo to tech wonks. This theorizing constituted a style of loosely analogical thinking that didn’t mesh with the precise engineering of my workday.
These two groups by and large thought ill of each other. My caricature, while exaggerated, is not too far off: to tech wonks, humanities scholars build ill-founded castles in the air with meaningless words to prove that nothing means anything. To the humanities scholars, tech wonks are imprisoned by a positivistic mind-set that leaves little capacity for context, speculation, or modes of thought that cannot be reduced to logical form. Each side tends to be remarkably uncharitable to the other. The playing ground is hardly level, however; the tech boom and increasing centrality of computation to life gave the tech culture a sense of relevance and financial success that contrasted to the glut of low-paid labor in academia. Academics can no longer ignore tech, and fads like the “digital humanities” as well as technology studies became new mechanisms of propping up the precarious system of American higher education.
To my mind, the two domains were equal—and equally foreign. The exactitude of computer science provided me with useful checks on linguistic hot air. Humanistic fancy, however, enabled me to figure out what I was doing in this technocratic labyrinth, and to ask myself why I was doing it and where it was going. I no longer program full-time, but I miss the mental practice it gave me, which served to focus my mind in a rigorously geometric fashion.
I didn’t belong to either community. I was a poor fit whether the topic of conversation was econometrics or Hegel. When I think of my social life in childhood, I remember silence, punctuated by occasional, apprehensive experimentation and subsequent failure. I dub it failure not because I failed to fit in, which is every child’s experience, but because none of my experiments seemed to result sufficiently in my being myself. My words were never quite right, and people never seemed to understand what I said in the way I had intended. I felt poorly served by the English language, and well into my teens I saw few models for how I could bend and mold it to the shape of my mental images and feelings. My favorite authors became those, like Virginia Woolf and Ralph Ellison, who were practiced in ventriloquizing through a wide variety of voices and putting each one into question through its proximity to others. But before I had those reference points, I found myself silent. Most of all, I was watching: I was Camus’s judge-penitent in The Fall. I was the watcher in Kafka’s “At Night”:
Deeply lost in the night. Just as one sometimes lowers one’s head to reflect, thus to be utterly lost in the night. All around people are asleep. It’s just play acting, an innocent self-deception, that they sleep in houses, in safe beds, under a safe roof, stretched out or curled up on mattresses, in sheets, under blankets; in reality they have flocked together as they had once upon a time and again later in a deserted region, a camp in the open, a countless number of men, an army, a people, under a cold sky on cold earth, collapsed where once they had stood, forehead pressed on the arm, face to the ground, breathing quietly. And you are watching, are one of the watchmen, you find the next one by brandishing a burning stick from the brushwood pile beside you. Why are you watching? Someone must watch, it is said. Someone must be there.
I marveled at practiced speakers who could give the same stump speech on whatever topic of their choosing (the mind-body problem, the Republican majority, the Turing Test) without so much as thinking about the words they were saying. Words never stuck in my mind—only ideas, and ideas can’t be spoken without words.
Heinrich von Kleist suggested that ideas of the mind were alive, which made them unable to be captured with a rote set of words. Ideas that admitted to a single permanent expression were dead—in the air and on the page.
Speech then is not at all an impediment; it is not, as one might say, a brake on the mind but rather a second wheel running along parallel on the same axle. . . . For it is not we who know things but pre-eminently a certain condition of ours which knows. Only very commonplace intellects, people who yesterday learned by heart what the state is and today have forgotten it again, will have their answers pat in an examination.
As I grew more fluent in English, I still found it hard to keep up with the parade of styles required by social norms. In college, I found that the freewheeling mode of essay writing I preferred did not make a good impression on most professors. There were exceptions, which made it that much more baffling. Most did not welcome the confused, inchoate flood of ideas that my teenage self liked to set down, and I was asked to adopt a very precise and academic subdialect to get high marks. This dialect varied by discipline and even by class. Many of my humanities classes were exercises in deriving the dialect required and forcing one or two ideas into the necessary framework, throwing away 90 percent of what I had found interesting. But faced with the choice between that and the seemingly impossible task of creating work that was simultaneously satisfying both to me and to the professor, I thought a good grade would serve me better in life. I could write eccentrically in my spare time. By the time I needed to decide between a job in programming or going to graduate school for philosophy or literature, the choice was obvious.
Computer languages are frequently variations on a theme. There are a few paradigms, such as the procedural, imperative structure of C (and its progenitor Algol) or the functional structure of LISP, which subsequent languages have refined and fine-tuned to bring out certain strengths at the expense of others. Imperative languages are structured as a sequence of commands that affect their surrounding execution environment; functional languages are structured as the evaluation of algebraic functions. The paradigms match different tasks better or worse than one another. So it is with human languages. Certain modes of speech and thought work better than others in certain contexts, so I learned how to speak in different settings. Algorithms can be expressed in different programming languages. A sorting algorithm is the same whether in C or LISP, even if its code is wholly different. I did not wish to give myself over to one particular dialect so that my use of language became predictable and ossified. I balanced the tech world with graduate school and tech articles with fiction.
I found myself enduring the constant push-pull of withdrawing from a social context while respecting its norms and its participants. My wife has observed that in raw form, my writing often reads like it was translated from German—all complex noun phrases and dependent clauses. I am aware that I am putting forth a great deal of effort to communicate simply and clearly. The autistic hacker Meredith L. Patterson explained her social difficulties in this way:
While LiveJournal and Twitter have taught me to translate my thoughts into English faster and faster over time, translation is still hard and it still isn’t realtime. . . . those childhood lessons about not being so rudely direct have stuck.
Unlike Patterson, I lacked any dominant instinct. I had different reference points for this sort of social autodidacticism, one being Denis Diderot’s Rameau’s Nephew, in which a bon vivant discusses how he has learned the arts of conversation, flirtation, and social parasitism through hard work and study:
I find in writers a digest of everything one ought to do, and everything one ought not to say. . . . I am myself, and that is what I shall remain; but I behave and talk in a socially acceptable manner.
And by the way, you shouldn’t suppose that I’m the only reader of this kind. The sole merit I claim here is having accomplished systematically, through clear thinking and rational, accurate observation, what the majority of others do by instinct. That’s why their reading doesn’t make them better than me, but instead, they go on being ridiculous, whereas I am so only when I mean to be, and then I leave them far behind me; for the same art that shows me how to avoid ridicule in certain situations, shows me also, in other situations, how to achieve it at a superior level.
Then I bring to mind everything others have said, everything I’ve read, and I add everything of my own invention, which in this domain is surprisingly abundant.
I used to look down on this sort of apparent fakery, prizing some ideal of honest and earnest self-expression. I slowly came to see, though, that it was not fakery but communication that I was after, and that only by learning such mechanisms could I be understood. To put it another way, I gave up feeling entitled to be understood on my own terms. And so I adopted stylistic dialects just as a computer runs different applications and compiles different languages into machine code.
The character of Rameau’s nephew is a general-purpose machine that runs varieties of social interaction so ware. I have strived to be similarly adaptable. But to be a universal social machine is also not to have the immediacy of instinct and the rushed emotions of saying what I mean and meaning what I say. That translation puts a kink in the process. I am slower and more error-prone than machines that run a particular social language natively. The writer’s eternal demon—do these words mean what I think or are there better words for it?—was with me long before I became a writer.
The hesitation I feel in using particular words or idioms, and the uncertainty I have that they ever truly fulfill my intentions, are not merely private matters for me. Even as I negotiate between my words and my thoughts, there is a much larger, but parallel task facing society: the primitive, dissonant relationship between our words and our computers’ code. We have been teaching computers how to translate crudely between human language and computer code. What they are able to understand and misunderstand now has the power to shape our lives.
1. Alan Sokal’s Social Text hoax remains the perfect illustration. While I find much of worth in the larger fields of philosophy and literature, Sokal picked very ripe targets that had, for the most part, become exactly the self-parodies that critics thought them to be. That some academics still get defensive when it’s brought up indicates just how close to the bone he hit.
2. To exercise that part of my brain I now turn to half-understood books on quantum physics, machine learning papers, and Stephen Lavelle’s diabolical puzzle game, Stephen’s Sausage Roll.
3. Or, as the cartoonist Eddie Campbell puts it, the tug between the Scylla of compromise and the Charybdis of failure.
4. A writer’s writer, Diderot had a heavy in influence on Goethe and Hegel, who cited Rameau’s Nephew in The Phenomenology of Spirit, which in turn in influenced Lionel Trilling’s Sincerity and Authenticity.
5. Ingeborg Bachmann’s short story “Simultan” (translated as “Word for Word”), a portrait of the inside of a translator’s mind, is as good a depiction of this non-native mind-set as I know.
From Bitwise: A Life in Code. Copyright © 2018 by David Auerbach. Excerpted by permission of Pantheon Books, a division of Penguin Random House LLC. All rights reserved. No part of this excerpt may be reproduced or reprinted without permission in writing from the publisher.