Reading for the Seams

I instantly know a Zadie Smith piece, the second I start reading. She is there on the page. She has some combination of literary movement and self-referential interruption in her sentences that is distinctively her. Ideas are followed a bit, until another idea appears out of nowhere and turns out to be at least as interesting. Which makes her words recognizable to me. There are certain writers like this, unmistakable after a few paragraphs: Joan Didion, David Foster Wallace, etc.

Sometimes a person is so present in their writing that you could recognize them without a byline or author picture; you know they're there before you've even consciously gone looking for them. I used to think occasionally about how strange that was, but it was a rare thought while staring out a window. Whatever reading is, some part of it has always involved making this hard to understand inference from marks on a page or a screen toward some mind elsewhere in the world, or backward in time.

That window stare, where I inferred another mind, used to be automatic. It was natural and so I didn't interrogate it too hard. Now I'm interrogating it like I'm running a CIA black site, as I struggle with two different types of distrust. The first: I know enough about whatever is being written about to immediately suspect I'm staring at some Borg-like amalgamation where the words are actually pretty close to a bullseye, but maybe off in one or two directions. The second: I know next-to-nothing about xyz subject, so I start staring suspiciously at all the surface-level clues: polish, frictionless prose, hey I've seen that color coding before, etc.

Here's what feels strange: I'm either looking for evidence that someone truly grappled with whatever they're writing about, or if I don't know the subject (i.e., I can't look for that grapple evidence because I don't know how), I start to inspect the language itself, looking for whatever tells we've all decided this month counts as machine-like. A ha! Load-bearing: caught you!

Neither is generous in the way that once went into reading somebody. In either case I'm not just riding the wave, reading the words.

More than we'd like to admit, I think most of us spend a lot of our time running into subjects we barely know (i.e., Path 2 above), which is now creating a real, if not devastating, problem for expertise: fluency and compression of thought now look machine-generated or AI-suspicious.^[1] Easy to read work, work that is logically tight, more and more resembles an LLM's output because it isn't tripping all over itself.

So will creators have to now intentionally "rough it up" or unlearn years of here-is-how-you-write-polished-text because the models are so well trained on how to shoot out easy, flowing, well-constructed arguments, that those are now a priori suspicious?

Hmm, okay - never, ever use load bearing or delve or an em dash! But the problem with intentionally roughing things up is that the machines will learn the roughness trick. It's an evolutionary arms race. It's hard not to worry that readers will continually lose whatever it is that is fun and reassuring about reading, or at least about reading other humans.

There's another consequence of this in that evidence "a person was here" is now moving out of whatever thing that person actually made. Where is it moving? I don't really know. But I imagine going to the cinema ten years from now and asking myself "did Wes Anderson think up this movie, write it out, film it up?" Provenance as we used to know it is now a lot muddier.

Or take bylines, which used to do real work; the writing underneath a given byline more or less carried all the proof of authorship you needed. There was no other possibility, outside of a ghostwriter, and ghostwriters never really seemed to bother anyone all that much because...human. Those words could only have been chosen by the person behind that byline, or by the person approving that the words are in fact good enough for the byline, publishing house, whatever.

Now I find myself ignoring all the old signals. And I see other people doing it too. Humans would have drafts! They could provide drafts! I cannot imagine a world where I ask someone for their drafts. So instead I'm like Luke Skywalker, in the Death Star trench, letting go and feeling the Force. Do I sense... a person here?

None of which has much to do with the sentences in front of me. We're in the midst of an upheaval where writing itself isn't the evidence it once was, and does that matter anymore? Will it? I couldn't say.

To break down and move past all this hand wringing, we have to ask what we get from reading in the first place. Sometimes I care zero about the writer. The value is solely in the words because I just want my coffeemaker to work or I need to take a train. I want the right information, no frills, and that's it.

But sometimes I want the frills. I want to see the peculiar connections another person could make that didn't or wouldn't ever occur to me. And I don't want an approximation. I want to know what Joan Didion or Zadie Smith or Taylor Sheridan or insert-your-favorite-writer thought about that thing, after spending some real part of their life thinking about it. I walked out of the theater after seeing Sicario and couldn't believe it; I pick up Slouching Toward Bethlehem once a year.

And those two "worlds" are now colliding. Which makes “good writing” feel like a less stable category than it did a few years ago. Sometimes the value is just in the words, and sometimes the value is tied in part to who chose these words.

I again search for an analogy, and the closest thing in nature I can think of is the orchid mantis.

I watch a lot of National Geographic. A lot. Usually if I wake up way too early on a weekend, or can't sleep on a weeknight, I'll decide to learn about bees or hyenas or orcas or wherever NatGeo wants to take me. My true favorites, though, have always been the tricksters, the mimics. The octopus who is now a rock, now sand. The mantis that's somehow a better flower.

The orchid mantis is a mimic, but not in the usual way you or I would think before we turn on NatGeo, bleary-eyed. Hiding? No. Disappearing into the flowers around it? Also no. This little scary bug becomes the best possible flower, derived by some evolutionary approach beyond my understanding where its forebears figured out what makes a flower good. This scary bug is brighter, more symmetrical, and so good at being flower-like that all kinds of little pollinators just walk right up to their deaths. There is something wonderful and horrible about it. This predator succeeds because it's an unusually convincing version of the thing its prey needs.

That's roughly where my National Geographic-addled brain has landed on what reading is kind of like today.

The flowers have always been everywhere. Our version of the mantises, well, now there are suddenly a spectacular number of them.

So we look a little harder at the flower. We are wary pollinators. We land on a leaf. Hmmmm. That flower is... too perfect. Those petals look aligned, too aligned, and over there those petals are uneven but exactly uneven.

I'm not a bug on a leaf. I need to read stuff is what I actually need to do. So I think to myself, as I read, that it's a weird place to be where something slightly off-kilter now feels deeply reassuring.

I may not be hopping from leaf to leaf, but my mantises keep getting better, noticing what I notice.