Last Week's New Yorker Review: November 20, 2023
In any event, Picasso and Chaplin did not get along.
Last Week’s New Yorker, week of November 20
Must-Read:
“The War on Chaplin” - Louis Menand takes pity on the little fellow. A straightforward history of Chaplin’s exile, one that so carefully sorts through the political and personal travails that rendering a judgement seems beside the point — one doesn’t need the received wisdom if there’s enough evidence to make up one’s mind. Menand’s rendering of the media environment avoids making the past look either too familiar or too bizarre; the heckles of gossip rags are placed alongside the shady political workings they helped inspire. Menand reads Chaplin as just an average liberal, willfully misunderstood because his art was iconoclastic, which feels about right. Still, I’d have preferred more on Chaplin’s politics and less on his sexual proclivities; Menand makes too many guesses about things there’s no way of knowing (“There is a real evidence problem here,” he says, only halfway through unspooling the evidence.) Mostly, this story is fascinating for those interested in why, and how, the culture industry occasionally decides to eat its own. Menand’s conclusion that Chaplin’s trouble was, in part, his individuality feels correct, and borne out by other cases: Welles, Sturges, even Keaton. In their own unique ways, none of them could pass the test.
Window-Shop:
“Made You Look” - Jackson Arn stays in Vegas. A dual review of The Sphere and Heizer’s City is a clever idea, and Arn draws the comparison aptly — they’re “trying for the same things but in opposite ways.” Most of this review is written in a somewhat irritating register I’d call mock-critic; Arn is all too aware that what he’s looking at isn’t exactly art, and he has to remind us of that by striking a performatively arch tone. (“Neat stuff.”) There’s also an inadvertently hilarious moment in which Arn tries to give two examples each of highbrow, lowbrow, and middlebrow culture, and manages six textbook examples of the middlebrow — fitting for a magazine with a complex about its own middlebrow nature. Once Arn finally dispenses with the glib scene-setting, though, his conclusions are quite wonderful; he locates the poignant in surprising places and nails the transcendence of “glitching” in immersive art: “it’s when the work ceases to be one size fits all, and yields, finally, to interpretation… illusion mixed with disillusion can be more intoxicating than either.” You can see the art critic tearing up — at last he has something to really look at.
“The Human Comedy” - Helen Shaw clowns around town. Sobelle’s show seems hard to put into words; Shaw gives it a go and provides some interesting context about “the French clowning tradition of Jacques Lecoq” but mostly I wished there was less of this review and a bit more space given to the two shows covered in the second section, which are linked not just by clowning but by a theme of “resolve in the face of adversity.” Shaw shines at rapidly summing up the crux of these productions, especially Eric Dyer’s “wild-eyed… holy failure,” and I suppose the slimmed-down Goings On page means her wonderful blurbs need to go somewhere. I’d still love another graf on each.
“Wild Reeds” - Alex Ross oboes about with James Austin Smith. Smith’s concert sounds super cool — an evening of contextualized musical history with “the feeling of a live documentary.” Ross brings his usual steady style, and doesn’t try to make more of the event than it warrants — but neither does he quite manage to bring the East German history to life. There’s just not enough time for that.
“Your Lying Eyes” - Daniel Immerwahr deepfakes around and finds out. If you’re going to read one A.I.-related piece from this somewhat misbegotten A.I. issue, this is the one — but you really don’t need to read much of it. The beginning and middle are full of repetitive place-setting, reestablishing things an astute reader (such as yourself) will already know, and giving a fairly stock history of image manipulation. But stick around, or skip forward, as the final two sections are a thoughtful, novel, and clever analysis of the deepfake’s misleading reputation and its true importance. It does seem that Immerwahr is cribbing a lot of this from the wonderfully named book under review, Walter Scheirer’s “A History of Fake Things on the Internet,” but why read the book when you can get the witty precis?
“In Front of Their Faces” - Eyal Press interrogates the facial-recognition technology that enables police profiling. This is fairly well-trod ground — Google brings up similar pieces in Wired and the Times. Press doesn’t add a ton; there’s no deep philosophizing about surveillance and justice (for that, see Theo Anthony’s fantastic All Light Everywhere) nor is the central anecdote remarkable enough to justify the piece on its own. If you don’t already know this story, this is as good a place to get it as any; Press keeps things brisk but comprehensive.
“Reality Bites” - Inkoo Kang flips for The Curse. Some nice descriptions — Fielder’s “vaguely befuddled, still-buffering affect” — and Kang’s enjoyment of the show mostly relieves her from nitpicking, which isn’t her strong suit. (There’s still too much nitpicking here for my taste.) Everything is in balance, but Kang still tends to write from a remove that keeps me from deeply connecting with her points. More feeling!
Skip Without Guilt:
“Metamorphosis” - Joshua Rothman networks with A.I. pioneer Geoffrey Hinton. Continually draws an extended metaphor between various aspects of Hinton’s fascinating life and various aspects of the A.I. problem. This means there are really three parts to evaluate: Coverage of Hinton’s life, coverage of A.I, and the lines drawn between the two. The first is, on the one hand, never so compelling that it would warrant coverage if Hinton weren’t a leading thinker on a major talking point — but still pretty compelling. Hinton is one of those weirdos who can’t help thinking and acting very much like himself, whether he’s capturing dragonflies and chopping down trees on his private island or reckoning with the losses and tragedies that speckle any long life. The A.I. coverage, on the other hand, never feels especially unique — it’s mostly competent, but never surprising, as though Rothman, who usually covers ideas, has tamped down some of his own ideas to make room for Hinton’s.1 As for the lines drawn between the two — I have mixed feelings. They’re often lovely parallels, as when the line “to have a feeling was to want what you couldn’t have” speaks both to alleged computer “feelings” and to Hinton’s grief over his late wives. But the story ends up feeling a bit trampled by them — if Hinton’s story is interesting mainly for what it says about his ideas, and his ideas are interesting mainly because they illuminate his story, then… why are we reading, again? The problem compounds itself, so that a piece which starts both grounded and compelling ends up somewhere hazy and dull. (“An A.I.… might be able to imagine a place like [Hinton’s cabin.] But would it ever need one?”) This feels like a recurring issue with this magazine’s tech coverage: Details demand metaphors, which evoke narratives, which overwhelm details. Sometimes, facts accumulating needs to be enough — the explanatory scaffolding can obscure the structure.
As for Hinton’s ideas, my guess is that Rothman focuses on the one that can best be turned into a tagline: A.I. can feel! But this seems mostly like a case of defining your terms in ways that make for good thought experiments — sure, A.I. can feel, if your definition of “feeling” is deliberately broadened to include the things A.I. can do. I don’t find those sorts of thought experiments that compelling, but they’re not useless — they’re essentially Hinton’s specialty; neural networks basically began as a thought experiment. That doesn’t mean every thought Hinton has will change the world.
“Begin End” - James Somers wonders if he’ll need to teach his kid to code. Frustratingly mediocre prose in service of a fairly unconvincing argument. I’m not code-literate, so I can only gesture vaguely toward the weaknesses I see in Somers’ idea that “centaurs” (basically just a chess-nerd term for cyborgs) will come to dominate programming, and that programmers will be left tending to software’s “design, its configuration, and its maintenance.” Proposing possible futures is always a fool’s game2 and Somers’ scrying covers an extremely small slice of what coding does in the present — many of the tasks he talks about feel basically like examples you could find in a coding cookbook, or by browsing github and making tweaks. GPT is good at speeding up simple tasks, and Somers feels nostalgic for this part of the process, which may be lost. But he makes this extremely narrow argument in ways that give it the feeling of something broader — a “tsunami” that threatens to wipe out coders entirely, a field being “swallowed at a gulp.” There’s a bit of dishonesty here, verging on clickbait: Somers has us anticipate a panic that he never quite justifies.3
Somers’ taste for fuzzy figures of speech doesn’t help matters. Coding and GPT-coding is variously compared to a Harry Potter broom, “an instrument you must learn to play,” “a spell language,” and “the root of”… something. There is a slightly disturbing reference to “GPT-4-sized holes.” I was also annoyed by Somers’ incredulity at coders’ supposedly cushy working conditions, which included a starting salary slightly higher than they’d asked for, “world-class health care and parental leave,” and the fact that “if the pressure to deliver ever got too intense, a coder needed only to speak the word ‘burnout’ to buy a few months” — a weirdly euphemistic way of describing basic accommodations for mental health. Somers sees something “wrongheaded in all this,” but I see the labor conditions every field should strive for — Somers is just describing what it looks like for a worker to actually be respected. Whether or not the robots are really coming, it can’t hurt to unionize.
“Infinite Art” - Anna Wiener self-generates with Holly Herndon. Incredibly frustrating. Feels the need to summarize the entire history of A.I. generated art, while cutting away every so often to Herndon’s biography, which isn’t particularly compelling. (She may be one of the most interesting artists in the A.I. world, but is she really interesting enough to warrant coverage without the A.I. hook?) That structure takes about three-quarters of the article, and the last quarter, where Wiener ought to tell us what she thinks, or at least what Herndon really thinks, is essentially a series of depressive shrugs — things could be bad (“the culture could become a void”) but maybe they’re fine (“the successful art work… morphs, moves on.”) There’s no real thesis here, and not even a presentation of Herndon’s arguments as cohesive — her musings just sit around, sagging under the weight of the extended recap of everything A.I. they’re asked to hold up. In theory, Herndon’s work fighting for artists’ rights could be the crux of the piece, but that work doesn’t actually seem to be what Herndon is interested in (try as she might to appear skeptical, she’s such a true believer she’s going to A.I. orgies in castles) and she doesn’t have much to say about it, so Wiener can only stretch the topic so far. Mostly, this just made me tired. A.I.? Alright, Igeddit.
Letters:
Regular correspondent Michael dropped a remembrance of Bayard Rustin, the subject of a piece by Adam Gopnik: “I first heard of Rustin in the early 2000s when there was a somewhat contentious debate over whether to name a new high school in West Chester after him. The vote was eventually approved, but reading back on the Philadelphia Inquirer's coverage of the issue you get choice quotes like, ‘“This whole project was a complete farce from beginning to end,” said [June] Cardosi, who has been opposed to Rustin from the beginning because of his sexual orientation.’”
What did you think of this week’s issue?
Maybe it’s for the best: The one paragraph where he lets himself muse is rather dire, with a Garden-of-Eden metaphor that ruminates on ‘exile’ in ways I find distasteful — the reality of exile is all around us, it’s not a theoretical to gesture at.
at least, if the goal of the game is “accurately predict things in ways that facilitate current understanding and future progress.” If the goal of the game is “make a quick buck,” then visions of the future are one of the best strategies available!
I should cite a discussion of this article in the Today in Tabs Discord as key to my understanding of some of this.
Lots of thoughts on this week's issue.
Overall, the AI pieces missed more than they hit. I thought the facial recognition article particularly redundant, bringing literally zero new insights into this space that I haven't already read in at least 6 other publications. The Hinton article did not justify its length and the metaphors were clumsy.
I did enjoy Daniel Immerwahr's piece on deep fakes. Unlike those two articles, this did at least bring an interesting angle on this topic even if I think Immerwahr was being too narrow in photos specifically. As the other AI article on generative artists have pointed out, the deepfake voice / music clone is on a different trajectory than images, even if these still mostly aren't to trick people (although I've heard of some phishing scams that have successfully done this).
Finally, the article on the eulogy for coding was decent. As some folks on Hacker News noted, the author wasn't coming at this as an expert (the specific example of getting random entries from a dictionary he cites is easier than many coding interview questions at tech firms). But the overall sentiments mostly rang true, even if the prose could be better. I doubt coding is fully going away but it certainly is changing a lot.
I had previously read a review of that Chaplin book in the LARB so it didn't feel as if Menand had brought that much to the picture here. I'd put the piece on the Sphere or deepfakes as my must-read for the week over it.