Admin is crashing out

PLUS: What are ChatGPT-generated book reviews for?

Greetings from Read Max HQ! In this week’s newsletter, two items:

• On the extended Elon Musk crash-out and the utility of seeing U.S. politics through the prism of forum drama; and

• an examination of ChatGPT-authored book reviews, and whether they would actually be published in magazines.

A reminder: This edition of Read Max is subsidized for all you freeloaders by the newsletter’s dedicated paying subscribers, whose generosity funds all of the obsessions, bugbears, irritations, and rabbit holes that lead to columns. All this writing (and the associated “staring blankly into space”) is a full-time job, and only thanks to people who sign up as paying subscribers am I able to treat it as such. If you get value out of this newsletter, please consider upgrading your subscription--at $5 a month you can think of it as “buying me a beer” in exchange for roughly 20,000 words of … well, whatever it is that I do.

Trump vs. Musk

One of the many stupid long-running theses of this newsletter is that what we think of as “social media platforms” are mostly just million-user message boards, and as such retain--especially among the most frequent and visible posters--many of the ancestral folkways, customs, and cultural conventions of their forum forerunners. No development of the last few years has been as vindicating of this theory as Elon Musk’s purchase of Twitter in 2022, almost immediately after which he assumed, as a poster, the extremely familiar persona of “forum mod”: A self-imposed “main character,” over-eager, easily wounded, and above all, corny. But it wasn’t until Thursday that we saw the fullest expression of Musk’s message board-ness, thanks to his adoption of the most classic forum activity: The crash-out.

A brief recap: Musk announced that he was leaving his ambiguous role within the Trump administration last week, and very quickly turned on the president’s agenda-setting “Big Beautiful Bill,” describing it as a “disgusting abomination” that would “drive America into debt slavery!” Trump, speaking at a news conference, said that Musk was suffering from “Trump derangement syndrome” and that he “would have won Pennsylvania regardless of Elon.”

In a reply to the right-wing crypto-influencer account “Autism Capital” on Thursday, Musk decided to set the record straight:

What followed has been described as a kind of back-and-forth: “An afternoon of escalating attacks,” in the words of the Wall Street Journal, or, as put more succinctly in an X.com trending topic, “the girls are fighting.” This is true as far as it goes, but I still think it’s a somewhat misleading characterization, given that Trump’s participation was limited to three crisp “Truths” in response to the dozens of tweets and retweets from a clearly aggrieved Musk. This was not, really, a “fight,” or, if it was, it was an example of that one-sided species of fight fostered by message boards--the crash-out, in which one participant gets so mad they enter a kind of trance state and post themselves into thrilling, embarrassing oblivion. This is a bit clearer if you visit Musk’s page and read all his tweets from Thursday, which display many of diagnostic criteria for crash-outs, among them:

• Unclear, flailing attempts at “dunks”

• Vague, overdramatic insinuations

• Petulant, wounded threats

• Loud proclamations of being not mad, actually laughing

• Attempts to rally support from the lurkers rubber-necking the crash-out

• Accusations that the object of your crash-out is a pedophile

• And, finally, backing down without having accomplished anything at all:

This is, of course, a highly visible political dispute between two of the country’s most powerful men, as well as a highly entertaining reality TV-style feud between two celebrities with a knack for attention-getting stunts. But I want to underline that this is more than just another example of the ongoing and near-complete convergence between “electoral politics” and “television entertainment.” It is also an example of the ongoing convergence between “electoral politics” and “forum drama,” driven by the overwhelming mediation of politics by various message board-like social-media platforms.

In this case, Musk has, consciously or not, stepped into the (astonishingly familiar despite its specificity) role of “overworked forum admin whose offline substance abuse and desperation to be liked by other posters has driven him to madness,” while Trump assumed (a long time ago, if we’re being honest) the stock role of “eccentric but predictable message-board regular whose strange obsessions and unique formal style have turned him into a kind of valued cult Character despite his obvious mental deficiencies.” (I would argue that his decade-plus performance in this role is as important to understanding Trump’s media presence as his status as “reality-TV star”; I would also say the fact that we elected a message-board eccentric to the presidency is both (1) a world-historical tragedy and (2) about what you might expect from a politics structured as a forum.)

From this perspective you might say that Musk was always destined to lose this fight, for the simple reason that maybe the biggest sin in forum drama is getting actually mad at the regular trolls. To blow up and yell at the characters is an unacceptable loss of face in message-board etiquette: It’s an admission that you can be trolled even by the most familiar and predictable figures. Yes, Musk thinks he’s doing politics, not fighting on a B.B.S. But you can’t win fights against posters like Trump on the forum itself. The baseline culturally proper response, if one was necessary, to Trump’s press conference would be for Musk to have QT’d the video with a single “lol.”

ChatGPT and F.O.B. Voice

Another week, another claim that ChatGPT-generated text rivals human-authored writing. This one comes from Times (of London) columnist James Marriott, who got “three paragraphs of a Chat-GPT review of The Rachel Papers by Martin Amis… I think you could submit writing of this standard to a magazine and get published”:

About this as a piece of writing you can really only say the same thing that has been said about every bit of intentionally “impressive” generative-A.I. production over the last few years: It’s amazing (to paraphrase commenter Kalen) that a computer can do that! But the object under consideration is, to people with experience and discernment, not really very “good”: overwritten, cliché-ridden, in some places simply nonsensical.

But I think the claim about a magazine publishing writing “of this standard” is in some ways more interesting. Because, in a different era--say, 30 years ago, during one of the many golden ages of magazine publishing--I think it would probably be true. In better times for the magazine industry, there was higher demand for a particular kind of glib (but not actually humorous), knowing (but not actually smart), fluid (but not actually stylish) text--what my friend Mahoney calls “F.O.B. voice,” for front of book, the pre-features section of a magazine for which, depending on the magazine, editors might end up cranking out 150-to-500-word nuggets of smooth blurb prose about new books, movies, news stories, gadgets, restaurants, or whatever.

F.O.B. voice is as a rule smooth and clichéd and often only semi-coherent, because it needs to be reproduced quickly and without much effort by overworked writers and editors on deadline. It’s also superficially impressive to most readers, thanks both to the packaging that surrounds it, and to their standards for impressiveness, which are quite a bit lower than professionals. For all these reasons, and additionally because it’s obviously been trained on archives of magazines written in F.O.B. voice, it’s unsurprising that ChatGPT takes naturally to producing in F.O.B. voice.

And unsurprising, in turn, that writers and editors would turn to ChatGPT to produce more F.O.B. voice at the press of a button. A few weeks ago, a freelancer admitted to using ChatGPT to generate a (very F.O.B.-voice) “summer reading list,” featuring a number of nonexistent books, for a “promotional summer section” to be used in a number of papers. But as Martha Bayne wrote at the time, this fiasco--as embarrassing as it was--was not precisely a tale of noble journalism betrayed, so much as the death throes of a business model in terminal decline:

Syndicated content is nothing new. King Features Syndicate, the Hearst-run operation that produced the “Heat Index” package, also syndicates “Dennis the Menace,” Sudoku puzzles, and columns by the “Car Talk” guy. And while I am beyond sympathetic to the pain of all involved — I am, after all, an author with a book coming out this summer, and have about a .001% chance of seeing said book included in any summer reading lists — it’s also silly to pretend that this stuff has ever been anything but lowest-common-denominator filler. Did anyone actually read Parade magazine? The “Heat Index” disaster is as much a cautionary tale for the business side as it is for editorial.

Ironically, just as ChatGPT and its ilk have gotten astonishingly efficient at generating F.O.B.-voice text about whatever subject imaginable at no marginal cost, the actual demand for F.O.B. voice is rapidly vanishing. By my count there are fewer than half a dozen genuinely good general-interest magazines still publishing in the U.S., and at their best this text is not good enough to be published in any of them (certainly not as written). But it’s also a bit too smart and wordy to be published in any of the remaining not-very-good general-interest magazines. A small sliver of the industry, rich and proud enough to pay real writers, remains; but the portion of the magazine and newspaper industry that might once have been interested in L.L.M.-produced book reviews is in the midst of an accelerating collapse. It’s very interesting that a computer can make that--but what you even do with a ChatGPT-generated review of The Rachel Papers?