Alexander Calder’s workshop
Something I found frustrating when learning how to write was that writing advice tends to be very low resolution. You have someone looking back on their experience writing for decades and they sum it up by saying, oh, well, don’t use adverbs.
I wish there were more detailed accounts of the decisions writers make when they struggle with a particular piece.
And since my general rule of thumb when writing this Substack is “write stuff that I would have loved to read a year ago”, I figured I should do that. A case study of how I write an essay. I’ll do Looking for Alice, an essay about not dating, one of the better I’ve written.
I warn you: this is going to be over-the-top nerdery. It is not for normal human beings. But if you like writing, or enjoy seeing someone analyze their craft, maybe this will be fun and useful.
December 28, 2022
I wrote Looking for Alice at what for me is a rapid pace—I had the idea on the evening of the 28th of December, 2022, and published it on January 17th, 2023. Normally, it takes me several months of revisions before a piece comes together.
This is where the idea came from.
I was on a Zoom call with Steve Krouse, an internet friend who lives in Brooklyn and runs a startup called val.town. We were having a normal internet friend conversation about Seymour Papert and end-user programming and such stuff, when Steve said, “This might be a strange question, but I wonder if you have any thoughts about dating.” He was under the impression that I am happily married (correct) and wanted to know how I did that.
There is a classic writing tip that says you should imagine the person you are writing for—it is even better when that person is asking you questions. As Steve asked questions, I noticed it was a lot of fun telling him about the experiences I had when I met my wife, Johanna. And it seemed he liked the advice I gave.
Maybe I should write it down? I thought.
No, that would be ridiculous—a piece about love.
But maybe?
Substack saves a version history of every post. Looking at the time stamps, it seems like I made a draft right after the call. Then for the next four hours, I opened it several times to jut down ideas in case I wanted to write the piece, which I surely shouldn’t do.
I remember that I was working on a long series of posts (still not finished) about designing software to facilitate social learning, which felt like an IMPORTANT topic. But the important stuff was a slog, so I kept procrastinating by sketching up ideas for the piece about not dating.
It should have been obvious that this was a good piece. When something forces you to write even though you can’t justify it and even feel a little ashamed, it must have some inherent energy to be able to push past your defenses. Pieces you don’t feel awkward about (when you have an idea and go, “Ah, this is great, people will like this!”)—those ideas don’t need to be as good to get themselves written. My resistance is a filter, and the ideas that pass through it are my best pieces.
By midnight (when I really should have been asleep for several hours), the draft looked like this footnote1. It is not good—”The first draft of anything is shit”, as Hemingway had it—but if you’ve read the essay you recognize that a few of the things that made it into the finished piece have already appeared on the first night.
29th of December
The evening after, I again felt drawn to the idea and ended up collecting notes and sketching for four hours. At this stage, I still had no intention to write the essay—it’s just that ideas kept popping into my mind, and I wrote them down.
I searched my email since I have had a decade-plus-long conversation with my ex, P, and we’ve talked a lot about dating, so maybe there is something there I could use? I emailed her asking if there was anything I had said that had been particularly not-stupid.
Looking over the material this haphazard notetaking surfaced, I realized I was looking at a listicle: 5 things to do if you want to find a good match. And listicles are good, readers like that, and having every part of a piece answer the same question gives it a strong coherence.
But.
Also.
Lists are tacky.
So I figured I should embellish the list—turn it into something that wasn’t obviously as a list. I could turn it into a story. If I sorted the advice in the order it was relevant in the story of how I met Johanna then I could string a few biographical details through the list, and it would feel like a narrative—while the fact that it was also a list would give it a deeper coherence and pull. I shuffled the ideas around a bit and wrote down stories and details from my life that I figured could craft a sort of narrative.
I liked the energy of having a narrative overlay a list. But still: I wasn’t going to write it.
I figured it should be a bit of an obstacle course in the middle, to give it a narrative pulse, so I put the part where it’s not working in the middle, and I made a mental note to make that part longer than necessary from a strictly intellectual point of view so that the story would have an arc.
Everything just came together of itself, against my will. I usually find it hard to figure out the right way to organize the ideas I want to cover in a sequence that makes sense to the reader.
Now I had a structure that looked like this—a series of ideas, all answering the same question, and a narrative flowing through all of it.
Some intro hook—what?
learn to pattern match (about before we met, and the first time we met, and the gertrude stein anecdote)
Don’t let go if you find someone cool (get to the point where she doesn’t want to hang out, and then lean into those 18 months as a vehicle to talk about the ideas)
Speedrun authentic (first I was in her apartment and talked like an idiot, maybe I can fit in the Herzog story here??)
Don’t talk to other people about your dating life ([redacted])
Outro
I made these points into titles in the document and sorted the notes I had written under the appropriate titles while adding new notes about what I wanted to do in each part. Anything that came to mind: maybe a part about Bayaki pygmies singing? an analysis of Pierre’s engagement in War and Peace? a commentary on Bachtin’s comment on Dostoevsky?
About two hours into that session, I had a great insight. I realized that the underlying idea in all the advice was that you should be open-minded about what a relationship is and not assume too much, and that anecdote about Gertrude Stein saying that she isn’t lesbian encapsulates that! I moved it to the opening line, as a hook:
Someone once asked Gertrude Stein if she was a lesbian. Stein answered no, I just like Alice.
That felt like a bold opening. Could I write like that? I wasn’t sure. But I liked the energy.
It felt like a concept handle, yes, a concept handle. That is a Scott Alexander phrase: a concept handle is a memorable phrase that makes it easy to remember an idea, like a handle that makes the concept easy to grasp. I should lean into the concept-handle-yness of the Gertrude Stein anecdote, I thought. Maybe the piece should be called something something Alice?
I played around with that idea. Maybe an Alice could be the concept handle for the idea that I was trying to communicate? “You are not looking for a girlfriend or a boyfriend, you are looking for an Alice.” Yes, that is good: it will make it easy to remember the way I frame the ideas. “Looking for Alice: that’s what you should do. Not dating.”
In fact, let’s make that the title. Looking for Alice. And in case that is too poetic for people, let’s make the subtitle Not dating.
I could now, after 8 hours of unintentionally poking at the idea, see what the piece wanted to be. I had a progression of ideas, I had a narrative pull, and I had a framing device that brought it together. From that point on, I just needed to type it out.
Even better, I had already written most of it. When I tidied up the notes, I noticed that I had good material for all parts except the last two.
Before moving on, let me make a general comment about outlining.
The problem with outlining
Many of my favorite writers—Knausgaard, Kapuschinski, Dawkins, Sebald—do not outline (like I do; not in the piece you are reading right now, though). And the reason for this is that writing from an outline tends to produce dead writing. Good writing needs to be a happening, as Kapuschinski says.
When you have a set of bullet points you want to cover and a lot of notes, it is common that the piece either gets stiff and list-like, or that it feels like a messy collage that doesn’t form a coherent whole. To avoid that, I don’t pay too much attention to the outline and the notes when I draft the actual piece. It is just a scaffolding that keeps me focused. When I get lost, I can peek at it to see where I’m supposed to be going. But I let the logic of the sentences, the rhythm, decide what needs to happen next. I will often drop a few bullet points and only use half of the sketches I've made because they don't fit with the flow of the piece.
Everything must be subservient to the form, as Knausgaard says. When a writer is so eager to say something that she says it at the expense of the structural logic of the piece—the writing loses energy.