Stop telling me what you think I want to write

Plus: Ukraine war at 4, remarkable rhetoric from a murderous dentist, and more

Good morning! It’s Wednesday, and here’s this week’s post.

---

A take I haven’t written elsewhere

Stop telling me what you think I want to write

(via)

Subtitles are bad for comedy. I know the youths love reading their television, and I’ll usually assent when a younger friend or relative asks to turn on closed captions, because I am nice. But I will not assent when we’re watching a comedy. The timing is never quite right, and I will not have each surprise and punchline ruined. It’s better to miss a word here and there than to spoil every laugh.

You know what ruins in a very similar way? These damn AI writing suggestions.

They’re not wholly novel, I know. Text messaging apps have offered some sort of autocomplete function forever. But recently—in the last few months, I think?—they’ve gotten so aggressive. And across platforms! I very much implicate

Substack

in this, by the way; it wanted “implicate” in this sentence to be “implore” on the first round and “imply” on the second, then it thought “implore” should be “implication” and “imply” should be “implicit,” and—be quiet!

Seriously, can I have a thought to myself? Can you not interrupt? Can you stop guessing? Let me say it. Just let me say it! I know you know what kinds of words I like to say. I know you could chronicle all my words so much more accurately and completely than I could. But please, please, leave me alone. Let me finish my own thought, or else it is not my thought.

It’s bad enough when Word and Gmail do this—the incessant invitations to “draft with Copilot” or use Google’s AI assistant to disgorge a slop clone of my voice and tone when scheduling a meeting or rejecting a pitch. Google, at least, has backed off a little since I repeatedly deleted its invitation emails and closed its top-of-inbox offers. But Microsoft is relentless. As far as I can tell, Copilot cannot be entirely exorcised from my machine.

Then there’s Substack. I am not in the habit of complaining about Substack on Substack, as I often find examples of the genre to be precious, winking, and self-congratulatory. Just once, though, I will do it, because Gmail doesn’t claim to be fostering creativity and the free exchange of original ideas. Substack does. So why is it trying to write my pieces for me? Why does it so often refuse to let my train of thought to run on its own track? Stop fiddling with that switch.

Some writers want this, I realize. Here’s

Noah Smith

way back in 2022 (he’s referring to himself in the third person because it’s a co-written post):

Much of the task of nonfiction writing involves coming up with new ways to phrase sentences, rather than figuring out what the content of a sentence should be. AI-based word processors will automate this boring part of writing—you’ll just type what you want to say, and the AI will phrase it in a way that makes it sound comprehensible, fresh, and non-repetitive. Of course, the AI may make mistakes, or use phrasing that doesn’t quite fit a human writer’s preferred style, but this just means the human writer will go back and edit what the AI writes.

In fact, Noah imagines that at some point, his workflow will look like this: First, he’ll think about what he wants to say, and type out a list of bullet points. His AI word processor will then turn each of these bullet points into a sentence or paragraph, written in a facsimile of Noah’s traditional writing style. Noah will then go back and edit what the AI wrote—altering phrasing, adding sentences or phrases or links where appropriate, and so on. An iterative, collaborative writing loop where an AI coauthor masters different parts of the cognitive stack than Noah himself, not dissimilar to the co-writing of this article.

I don’t know if that’s how he works now. The LLMs have probably risen to the level of competence he had in mind, but maybe he didn’t end up going this route.

All I know is I do not want to go this route, and I don’t want to be incessantly nudged in that direction. I don’t want to edit an LLM’s writing; I want to write. I don’t find phrasing work boring; I find it as necessary a part of writing as the ideation stage. Getting the wording right is part of refining ideas. If a turn of phrase is wrong, it may mean the idea underneath is not quite right. Iteration should not be outsourced. Leave me alone to think it through myself.

Or, at least, give me the option to be left alone. Provide the aggressive, intrusive autocomplete if you must. I’m sure there’s a big market for it, and Substack is on record as being in the AI optimist camp. That’s fine. But let me turn it off, ideally with a single, easily located toggle.¹

Finish other people’s sentences. Let me have mine to myself.

---

Intake

• The Anatomy of Murder, by Dorothy L. Sayers et al. I am delighted to report I’m still coming across new Sayers-involved mysteries thanks to these Detective Club volumes. This is my latest acquisition. I’m only in the first chapter, which is an account of a real murder in Australia in the 1860s. The people involved are all pretty middle class: The victim is the main teller at a large bank, and the murderer is a dentist. Presumably they are reasonably well educated but not the product of elite institutions. What’s interesting, then, is the strikingly formal and florid way they write, as the account includes excerpts from letters and diaries which were revealed during the trial. A sample, from the murderous dentist:

Do not rouse the demon that I know lies dormant in me. Beware how you trifle with my love. I am no base slave to be played with or cast off as a toy. I am terrible in my vengeance; terrible, because I call on the powers of hell to aid their master in his vengeance. God, what am I saying? Do not fear me, darling love. I would not harm thee, not thy dear self, but only sweep away as with a scimitar my enemies or those who step between my love and me.

• “I teach at Harvard. Store managers see me as a threat,” by Reginald Dwayne Betts for The New York Times

• “A theory of media that explains 15 years of politics,” Ezra Klein interviewing Martin Gurri for The New York Times. I quoted Gurri’s book in Untrustworthy—interesting to see it popping up here now

• “The deeper question raised by the NIH grant overhaul,” by Yuval Levin for The New Atlantis

[I]t sure seems more likely that in this domain, as in others, the Trump administration has decided to grandstand rather than govern. They have taken an idea that could easily have been the foundation for a plausible reform and have chosen instead to make it the foundation for a social-media campaign divorced from administrative realities, at odds with the law, likely to be reversed, and sure to further raise the temperature of our politics.

• “Douthat on belief,” by Alan Jacobs at The Homebound Symphony

But I am not at all convinced that a move from, say, atheism to Wicca is necessarily “a step in the right direction”—i.e., once you’ve entered the genus-town of “religion,” you’re closer to the species-house of Christianity than you were before. Indeed, I wonder whether many people might be less interested in Christianity as a result of such a move, since they might plausibly think that as long as they’re operating within the genus, does it really matter what species they prefer? (The “We all get to God in our own way” line has had a very long run and doesn’t show any signs of slowing down.)

---

Output

New work:

• Ukraine war at 4 | Defense Priorities (newsletter)

• Admin-heavy season for me at CT, so not a lot of fresh writing right now

Newly relevant work:

• NATO made a mistake by promising membership to Ukraine | Reason, July 2023

• When more troops means less security | Reason, July 2023

• Of course Washington and Moscow are fighting a proxy war | Reason, April 2023

• Pentagon link reveals 14 U.S. troops in Ukraine | Reason, April 2023

• The war in Ukraine has no end in sight | Reason, February 2023

• The GOP split on Ukraine aid isn’t really about Ukraine | Reason, January 2023

• Are we sure America is not at war in Ukraine? | The New York Times, June 2022

---

1. If this exists, I have not been able to find it, neither by poking around in settings nor by googling. If you know about it, please tell me! I want to be wrong here.