O flesh, how art thou fishified

Emotion canceling

Information science presents signal and noise as a fundamental dichotomy, presuming that these can ultimately be separated cleanly. This leads to a view that language in its multivalence should ideally be reduced to unambiguous code that is pure signal and no noise, and that technology should aim for that goal of abolishing the need for interpretation in favor of pure, direct transmission that operates like a reflex, so that talking to someone would be like pushing their buttons and pulling their strings. The degree to which communication might convey a range of feelings is the degree to which it fails; it should be purified into a purely “neutral” tone of voice, like a Tao Lin novel.

Softbank, a name often associated with wasteful funding of inane tech startups, was in the news this week for announcing an “EmotionCancelling Voice Conversion Engine” (“AI-driven,” as everything must claim to be right now) that purportedly alters the tone and pitch of spoken words in something close to real time, so that— to take up the example Softbank was eager to promote — when someone is yelling at a call-center worker, their fury will come across as muted and less antagonizing, if not as flat and affectless as might be ideal. This way companies can continue to infuriate customers without worrying so much about the psychic damage their already overtaxed workers might face as a result, in this brief interim before the companies can replace those workers entirely with machines.

Dubious emotion-detection software is already used to evaluate workers and size up their breaking points. Brian Merchant called attention to a company that plans to use emotion detection “to detect when a call center worker was on the brink of ‘losing it’ — and play them an AI-made montage of family photos set to their favorite song to calm them down.” The point is that employees should not be free to have spontaneous emotions while working; those feelings should be instead be brought under management’s control and deployed as a company asset. The “emotion detectors” actually inflict feelings on a captive population, compelling them to behave in ways the systems are designed to induce.

Obviously this kind of monitoring technology is ultimately anti-worker. It’s also anti-consumer. Assuming emotion-cancellation technology works anything like Softbank claims (a very iffy assumption), it fits with what Adam Kotsko complains about in this post:

The market was supposed to incentivize businesses to offer attractive products, at competitive prices, in a convenient format, and then customers were supposed to respond to those positive signals by rewarding them with their business. Now businesses increasingly take actively customer-hostile actions — locking up products, replacing paper menus with cumbersome QR codes, and of course chronically understaffing everything, which is the root of all of these issues — and insulate themselves from any feedback.

I tend to see “customer service” as a way of entangling us in our identity as “consumer” and reinforcing the hold consumerism has over us, making us dependent on the conveniences and petty tyrannies that can come along with shopping, as if these were the best pleasures society had to offer. But I think Kotsko has a point here that technology and monopoly conditions have freed companies from accountability, from having to meet customers’ expectations, even when those expectations are circumscribed by what sorts of satisfactions are imaginable in a consumerist society. Companies are frustrating the emotions of entitlement that they have worked so long to foster in their customers and insulating themselves from the consequences, leaving that anger to accumulate and ultimately be discharged in other ways.

Emotion-cancelling technology, like noise-cancelling technology, would addresses the symptom and not the root causes; it assumes an angry world and would do nothing to mitigate that. In fact, if anger stems from disconnection, misunderstanding, and misreading of people’s intentions, emotion-cancelling technology would certainly make the situation worse. (Perhaps this is the selling point for VR helmets, which could filter out all external emotional expression, allowing the desocialized helmet wearer the impression that they are in full control of their own solipsistic feelings.) It would contribute to a situation in which no one bothers to attend to the cues that help them understand what another person is feeling, because those cues have been filtered out of communication, which would be reduced to a universal deadpan. The skills people might otherwise develop in interpreting tone would further atrophy, and other people’s feelings would generally be understood as just noise.

But it is just as likely that emotion-cancelling technology would push us to become more creative in how to convey our emotions. Angry customers would develop icier ways to convey their contempt, and so on. Most people don’t regard other people’s feelings as noise; they see them as one of the most valuable things in the world, the only information worth anything in the end. Achieving a level of communication where we are not mutually cancelling one another’s emotions in various ways is about the best thing we can hope for out of life, and it’s unlikely that most of us will want technology that actively works against this. If we come to assume that every voice we hear over a phone is as manipulated and altered as every image we see on one, we’ll likely start to look harder for forms of direct “real” contact rather than give up on the idea altogether.

Outward watchings of life

I recently started a project where I hope to read all of Shakespeare’s plays in copies I find in free “little libraries” or for 50 cents or less at thrift stores. It’s not that I couldn’t just take the Riverside Shakespeare down off my shelf and read through them all in a few months that way, but I felt like I had to make a game of it or else I wouldn’t bother. And it seemed like I should bother: Given all the terrible books I’ve read in my life, I feel like I should at least make time to get through all of Shakespeare once. This method is even forcing me to read the most obvious and most popular works first: It may be a long time before I come across a cheap copy of Coriolanus.

Years ago, when I was specializing in literature in graduate school, it seemed pointless to read Shakespeare because there was too much secondary literature to master and it seemed impossible to make any fresh arguments about any of it. I also didn’t have the aptitude or patience for the sort of philological work involved with sorting through the surviving quartos and folios and establishing viable readings of various lines and that kind of thing. But now that I don’t have to worry about having a professional or scholarly interest in literature, I feel free to actually get something out of it.

Anyway, I mention that only to explain why I was reading Coleridge’s Shakespeare criticism, where I came across this passage:

Mercutio is one of our poet’s truly Shakespearean characters; for throughout his plays, but especially in those of the highest order, it is plain that the personages were drawn rather from meditation rather than from observation, or to speak correctly, more from observation, the child of meditation. It is comparatively easy for a man to go about the world, as if with a pocket book in his hand, carefully noting down what he sees and hears: by practice he acquires considerable facility in representing what he has observed, himself frequently unconscious of its worth, or its bearings. This is entirely different from the observation of a mind, which, having formed a theory and a system upon its own nature, remarks all things that are examples of its truth, confirming it in that truth, and above all, enabling it to convey the truths of philosophy, as mere effects derived from what we may call the outward watchings of life.

Ordinarily this would have washed right over me as run-of-the-mill humanism, one of the many truisms set out about “the Bard” as “total artist” (to borrow a phrase from Stephen Greenblatt) and his supposedly singular capacity to capture human nature truthfully and so on. But one of the effects of subjecting myself to so much speculation and hype about “AI” day after day and week after week (to which this increasingly tedious and repetitive newsletter must attest) is that the humanistic truisms start to have a renewed urgency. There is no real justification for taking them for granted, especially when trillions of dollars are lining up to mount an unrelenting assault on them.

Coleridge’s point here is that one can readily mistake data collection for knowledge of other people, especially when one becomes practiced in it. Having a “theory” to direct observation comes to seem superfluous; one no longer needs any “bearings” when coming to conclusions about how to represent things to oneself.

Habits of observation can thereby become a means of living an ever more superficial life, a way to be in the world without a consciousness of its “worth.” Generative models are exemplars of this approach to phenomena (all observation; incapable of meditation); the more prevalent they become, the more plausible it may seem that the world is both easily understood and worthless at it is, and thus suitable for replacement with simulations. Characters like Mercutio, Coleridge argues, exemplify the opposite approach, which is presumed to be Shakespeare’s own.

Mercutio is from Romeo and Juliet (very readily found in thrift stores); he is given lots of tedious, wordplay-heavy dialogue before getting killed by Tybalt in a street fight, which sets the stage for the rest of the tragedy. But above his function in the plot, he seems to be deployed mainly for Shakespeare to show off some rhetorical tricks and “wit.” Coleridge describes Mercutio as “a man possessing all the elements of a poet: the whole world was, as it were, subject to his law of association.” But this association is not the statistical association uncovered by machine learning approaches to language corpuses; rather whenever Mercutio “wishes to impress anything, all things become his servants for the purpose.” His character demonstrates how it is possible to invent new meanings and rhetorical associations, how language can be continually be refreshed and reinvigorated and willfully misunderstood, how it resists reduction to mundane literalness, to signaling information.

I found Mercutio unbearable, as I do most of Shakespeare’s “witty” characters so far, but I accept that this is because they place a much heavier burden on me as a reader. I have to read their dialogue with far more attention, over and over again, to make sure I am grasping the different levels of meaning, or else otherwise why I am even reading these plays? On the page it is as though the emotion of the language has been cancelled, and it is my task to reconstitute it. (Reading Shakespeare is a way to cancel emotion cancellation — a sentiment my high school English teachers probably would have applauded.) Words can’t be separated from the emotions behind them in the particular situation when they are used. The value of the world is in attuning not to the words themselves but the variable purposes behind them.

Mercutio’s dialogue is alive not in the sense of seeming true to life or realistic (it’s in highly compressed and allusive metered verse most of the time) but because it is rich with interpretive possibility; it highlights the uniquely human proclivity to invest language socially with ever-shifting meaning, part of what Greenblatt calls “the circulation of social energy.” Shakespeare’s work in general, pored over as it has been by centuries of readers, critics, performers, and scholars, testifies to language’s inexhaustibility, to the sense that more density and texture keeps being added and no analysis of it can ever be final or total.

“If the textual traces in which we take interest and pleasure are not sources of numinous authority, if they are the signs of contingent social practices, then the questions we ask of them cannot profitably center on a search for their untranslatable essence,” Greenblatt writes. “Instead we can ask how collective beliefs and experiences were shaped, moved from one medium to another, concentrated in manageable aesthetic form, offered for consumption.” That is, instead of seeing words as separable from feelings, we might see something fundamental to social practice in their inseparability, and something fundamentally anti-social about tech companies on their current trajectory.