Argument by Adjective

a visit to my blog archive

Every once in a while, I’ll visit the archives of the various weblogs that I maintained over the years. I don’t do this often—my writing there (and here) was more about momentum than memory. That is, I was more interested in expressing and maintaining a writing habit than in establishing any sort of permanence. On occasion, though, I posted something that stuck with me.

That’s the case here. We’re coming up on the 15-year anniversary of a blog post I wrote back in June of 2012, titled “Argument by Adjective,” which I was particularly proud of. It’s a little dated, so I thought I might update it here, and after that, I’ll talk about why I thought about it this week (spoiler: it has to do with Ai).

Photo by Jason Leung on Unsplash

Argument by Adjective

I’m not going to reproduce the full entry here, but I’ll quote some of it. The post opens with a bit of reflection on the fact that, in the summer of 2012, I was working with graduate students on their preparation for the fall job market. We’d just done some work on cover letters, and I’d noticed a particular tic in my own writing that I cautioned them to avoid.

It’s important to understand that the habit itself is prompted by the genre of the cover letter. For most (academic) job applicants, the cover letter is their first impression, and “the most difficult site at which we need to balance the competing interests of writer and audience.” This is a real challenge. On the one hand, applicants want to make sure they include every potentially relevant piece of information; on the other, especially these days, search committees may receive hundreds of applications, and the prospect of reading a 5-page essay¹ from each of those candidates is overwhelming.

One way of attending to this balance is through compression:

I might have only 1-2 sentences to describe a project where I would prefer 6-7, and as a result, I’ll really try to pack those 1-2 sentences, and make every word count. When I’m doing this, I have a bad habit of building arguments into my adjectives. So for example, rather than including 2-3 sentences about the various perspectives I’m negotiating as I engage in an interpretation of a phenomenon, I might instead simply say that “my project offers a nuanced, rhetorical interpretation of X.” It’s easy to fall into this–it saves space, and it starts to get at what I feel is the value of my project.

This strategy—argument by adjective (AA)—is conducted through a substitution that feels innocent enough in the moment. Rather than taking more words to describe what a project or a publication does, we can translate that predicate into a modifier, an adjective, which implies that it names what that project is.

My archetypal example of AA² is Dave Eggers’ novel, A Heartbreaking Work of Staggering Genius, although Nicolas Cage’s 2022 movie The Unbearable Weight of Massive Talent is a more recent instance. Both demonstrate a tongue-in-cheek exaggeration of self-regard that goes well beyond anything you’re likely to find in an application letter. At the same time,

I do think that we employ that kind of blurby language, albeit more subtly. It’s not really up to me to decide or to claim that my work is complex, sophisticated, nuanced, compelling, effective, interesting, thoughtful, careful, close, detailed, thorough, or heartbreaking.

I had a couple of different pieces of advice, beyond simply refusing adjectives. First, I argued that they should be deployed somewhat like pronouns.

In the same way that you wouldn’t use a pronoun without first establishing the proper noun to which it refers, I try (hard) to use only those adjectives that are earned within the context of the document itself. That means that if I want to describe my project as nuanced, I need at least an additional sentence that communicates (effectively) that nuance. Or that complexity. Or that thoughtfulness. Or that staggering genius.

My second piece of advice, couched specifically in terms of the job application process was that

Adjectives, I try to remind myself, are for interviews. Nouns and verbs get my foot in the door. That’s the (brilliant) piece of (timeless) advice that I (thoughtfully) re/learned this (delightful) summer.

To put it a little more concretely, one of the things I learned (and remind myself of regularly) is that cover letters are less about persuasion than they are description. There’s something of the bait-and-switch in AA, the insistence that the intensity of our desire (at least in the application process) matters more than the reality of our qualifications. Years later, I remain convinced that this is worth my vigilance, even if I tend to excess myself³.

Photo by Enchanted Tools on Unsplash

AA in the Wild

I was thinking about argument by adjective in a very specific context this week, and that’s the degree to which proponents of Ai have taken to the phrase “Responsible AI.” Dave Karpf had a post today where he responds to Adam Mastroianni’s NYT critique of datacenter resistance (“The Environmental Case Against Data Centers is Misguided”). I don’t want to dive too deeply into the discussion itself, but Karpf characterizes part of Mastroianni’s argument thusly:

There are good and bad things about the AI revolution. But what we should really be focused on is responsible use. Everyone, individually, ought to think about the right and wrong ways to incorporate AI into our lives. That would be far more effective than the data center backlash. Let’s engage in this totally different conversation about ethics and personal responsibility.

Likewise, I’m not especially interested in litigating the existence of “responsible Ai” as a talking point (you don’t have to take my word for it if you don’t want to). One easy test for AA is whether anyone could realistically hold the opposite position—my guess is that not even aspiring trillionaires with all the subtlety of MechaHitler would claim that they’re trafficking in “irresponsible Ai.”

The word responsible, paired with Ai, functions as an alibi. As Karpf notes, it changes the conversation, focusing our attention on personal usage rather than structural effect or systemic logic. Mastroianni spends three paragraphs of a relatively short essay detailing his own boundaries with respect to the way he deploys Ai, without ever acknowledging the underlying warrant of this rhetorical choice. “I refuse to let A.I. evolve from a tool into a partner,” and so its moonshot estimates of environmental costs “might be worth” the price we’ll all pay?

Artificial intelligence is a technology that can be used responsibly, I suppose, but flipping that anecdotal usage around into a descriptor of the technology itself is a pocket-lining choice of extractive would-be geniuses. And it comes at a time when they have been particularly reticent to assume responsibility for the real-world consequences of their technology, to say nothing of its knock-on market effects that have already warped our economy. And it comes from people whose decisions are guided far less by public good than by private gain. Karpf notes, “You could build a data center that uses basically no water, and that is powered entirely by solar and batteries. But that would cost a lot more money and take a lot more time.” The Ai barons are racing one another to try and claim monopoly share, and they are perfectly willing to light our economy and our environment on fire to achieve it.

One answer, then, is to coin phrases like “responsible Ai,” to introduce, without any actual basis in reality, positive associations with their product, in the hopes that they will settle subliminally into our everyday usage and go unexamined⁴. The other answer would be to behave responsibly, to account for the actual impact of your product, not just its shareholders. That might mean genuine environmental impact studies. Or being forthright about the expense of data centers, rather than hiding it under shell company bushels, and allowing the market to judge more accurately their financials. Rather than dialing down the sycophancy from model to model, building one that doesn’t require you to pull back from the psychosis you’re inflicting⁵. At the very, very least, you might design a product that doesn’t rely on hundreds of millions of pinky swears by its consumers to resist the temptation to break the world for a few extra bucks. If Ai were in any measure responsible, the phrase “responsible Ai” would be redundant, rather than smoke.

Afterthoughts

In my original post, I compared AA to the language typically found in promotional materials, like book blurbs. We are more forgiving with that genre—it’s exceptionally rare to find a book blurb that doesn’t engage in a little hyperbole that way. When it came time to write my own summary for the back cover, I struggled with it for much longer than I probably should have, resorting to a Facebook poll of my friends to get “permission” to describe my own book as “groundbreaking.” To this day, I wince a little at the thought of that, even as I live in a country ruled by one of the most psychotically self-aggrandizing bullies on the planet. As normalized as it’s become, and as exhausting as it is to have to constantly discount everything we read and hear, I still can’t help but feel like it’s important to earn one’s adjectives.

That’s all from me this episode. I need to get back to my work on attention, so I’ll probably work on William James this week. More soon⁶.

1. This is not an exaggeration.

2. It wouldn’t be until quite a bit later that I happened upon a chapter from Ursula K. Leguin’s Steering the Craft. Chapter 5, “adjectives and adverbs,” is barely longer than a page, but it closes with the following wisdom: “I recommend to all storytellers a watchful attitude and thoughtful, careful choice of adjectives and adverbs, because the bakery shop of English is rich beyond belief, and narrative prose, particularly if it’s going a long distance, needs more muscle than fat.” I use LeGuin’s “chastity exercise” in my Style course.

3. And I do. I would estimate that about half of the proofreading I do on my Substack is going through my episodes and trying to prune all of the ticky-tack adjectives and adverbs that I use to hedge my claims. (e.g., I just removed the word “little” from in front of “ticky-tack.”).

4. If this terminological trickery sounds familiar, it’s much like the way that the political Right has routinely engineered the inverse effect, the Mirror World demonization of vocabulary like CRT, DEI, and even the old chestnut of describing everyone to the political left of the mad king as Communists. In fact, we might describe AA (“Sleepy Joe,” “Little Marco,” et al.) as the refuge of the increasingly incoherent.

5. There is absolutely nothing about the tasks that Ai can accomplish that require the deception and simulation with which they interact with us. There’s a reason why we don’t hear stories about folks wanting to marry Siri, Alexa, or the summaries at the top of our Google results. Psychosis and sycophancy is a choice that these companies make.

6. Thanks to this episode, I realized that my standard signoff is in fact, pure adjective/adverb. D’oh.