ten thousand takes on tech, culture, criticism, and bait
my podcast conversation with the writer and technologist Jasmine Sun ✦
One of the great pleasures of writing this newsletter is meeting internet strangers who become dear friends and intellectual colleagues.
About 3 or 4 months after I started this Substack, I met
at a weekend co-writing event…and then very aggressively set out to befriend her. Jasmine is a former Substack PM, a cofounding editor of (which publishes an annual print magazine, Kernel), and now an independent writer. She’s also someone I admire enormously: every conversation with her is incredibly invigorating and thought-provoking. Speaking with her always makes me feel like my brain is overclocking—in a good way!
So it was a huge pleasure to record a podcast with her about our favorite topics: tech, tech writing, culture, culture writing, the (supposed) death of the humanities, the future of criticism, what AI can do—and, crucially, what it can’t.
You can find our conversation below (including a transcript!) and you can also subscribe to Jasmine’s podcast on Spotify, Apple, or YouTube. Previous guests include the NYT tech columnist Kevin Roose and the Stanford professor Fred Turner (who wrote the influential From Counterculture to Cyberculture).
What do Jasmine and I discuss? Well—
Silicon Valley ideology (the pros and cons)
Early on, we discuss the culture clash I experienced when I (briefly) left Silicon Valley tech to go to art school in London—
When you’re in the tech world, it just feels like things are always expanding. Things are always getting better and better.…Whereas in the arts and humanities world, there’s this feeling of decline.…That shift from this world that is constantly growing to this world that feels itself to be contracting and under threat was so interesting to me.
I am curious about what it takes to bring that kind of optimism and that feeling of expanding possibility into the humanities. I went to undergrad and studied computer science. I learned to code. And then when I became more interested in literature, history, philosophy, and so on, I just felt like I had this incredibly, intellectually expansive period of my life…I feel that there are a lot of people who are interested in the humanities, and it’s about figuring out, How do you bring more people into the fold?
—and Jasmine’s own experiences going from a much more humanities-oriented environment to the startup-obsessed world of Stanford.
I think our experiences leaping across disciplines—the “two cultures” of STEM and the humanities—is what makes it so interesting to speak to each other! And we do a little thinking out loud about what the intellectual foundations of Silicon Valley are:
The value of just dipping into all these disciplines and worlds that all have their own vocabulary, their own canonical texts—is figuring out how to…not just describe the surface level, “this is the stuff people are saying or doing,” but “Here’s the underlying ideology or philosophy or worldview that is transmitted by this.”
In Silicon Valley, people do think of disruption as inherently good. They think of information as inherently abstracted and decontextualized. And some of this comes from the philosophical origins, maybe, of how data structures work, how algorithms work, how you can abstract away from what is the specific data and just think about how to handle it and how to store it…[which leads to] tech people believing that they can take these tools and try to solve problems in a lot of different fields. It explains why tech people are always trying to get into health-tech and urbanism.
This ability to get beyond the surface-level—and comprehensively explain what’s going on in tech—is something Jasmine is exceptionally good at. I recommend:
…plus Jasmine’s WSJ piece this May about vibe-coded apps; and her Business Insider piece in January about Silicon Valley’s new Red Scare. (If you’re an editor, you should consider commissioning her!)
How the internet can lead to “radicalization for good and for cultural edification”
I know people are very anti-internet (our deteriorating attention spans! our instinctual, id-driven addiction to bad content!) but Jasmine and I heroically defend the internet as a force for good…yes, even for intellectualism!
We discuss my (first) viral newsletter—
—which Jasmine describes as “a celebration, implicitly, of the internet as something that lowers the barriers to doing personal research.” And why I’m ultimately a “contrarian optimist” when it comes to the internet, because:
When I think about my interest in literature, a lot of it was very much facilitated by the internet. I did not grow up in a New Yorker household…And so a lot of the books that I now think of as foundational to how I see the world, to my aesthetic worldview, my ethical worldview, I just found out about because people would post on Reddit or Twitter. I think that is something really special about the internet, that you do not need to be in the right family context, geographic context, social context where these things are automatically accessible to you. You can just discover your interests online.
The sociologist Zeynep Tufekci has this TED Talk about how YouTube encourages political extremism…[but] this push into the extremes happens in so many parts of the internet…You can develop this intellectual radicalization where you’re kind of like, “Ooh, I wanna know what book I should read,” and then two years later you’re a brodernist.
We also discuss my resolution to use social media more, and how to use it a “vehicle for aspiration,” in Jasmine’s elegant turn of phrase.
Writing advice, why we love Substack, and why we still love being edited
Writing this newsletter has forced me to be very intentional about when I lean into self-promotion and growth hacking…and when I seclude myself from all those harrowing statistics in order to pursue my pure, unfettered artistic vision…
So when Jasmine asked me, “As a writer, what role do metrics play for you?” I was delighted to opine:
You, as a creative person, are trying to defend yourself against this encroachment of numbers that measure whether your work is worth it or not. The number of likes is not actually how worthy your writing is, but it really feels like that’s the case. And obviously you want to be receptive to feedback, and it’s valuable to know what lands more with people and what lands less, but when all reactions are filtered through the number of views, the number of likes...
There’s this funny balance where some writers get way too purist about it and they’re like, I don’t want to have to package my work up for the masses and for the hordes. I think if you really believe in the quality of your work, you owe it to your work to package it in a way where it can reach the most people. There’s obviously a way to do it that is clickbait-y and growth-hacky and kind of trashy. But then there’s also a way to do it that’s just like, are you backing your own work? Do you believe that it’s meaningful to people? There’s this awkward balance where you want to market your work, but then you also want to retain some purity.
We also discuss why we believe in the democratic potential of writing on the internet:
The internet…has created this publishing environment and this media environment where if you are a random person, you can write a review of an album, you can write a review of a book, you can disseminate it around. And I don’t mean just in the vulgar sense of, oh, everyone thinks they’re a critic even if they’re writing a two-star Goodreads one-liner. But everyone can be a critic in the sense that everyone has a chance to try to achieve a certain level of analytical excellence, literary excellence, intellectual excellence.
Democratization leads to a lot of slop, but it also means that a lot of people get a chance to refine their slop into something that’s really special and innovative…
Sometimes the difference between someone who’s a really excellent writer and someone who seems clearly amateurish is just that the excellent writer was praised at the right periods in their journey, received the right mentorship and encouragement to keep on going. I really find it precious that the internet can offer more of that encouragement to people…You don’t have to be picked by a gatekeeper to be brought into an institution before you’re allowed to make your work. You get to make your work and see—do people care about this? Is this resonating with people? Can other people see something in it that I didn’t see initially?
I also offer some Substack tips—and specifically how I get people with fractured attention spans to read my 5,000-word newsletters:
I have this Proust newsletter that went viral earlier this year. It’s something like 5,000 words…a lot of people read to the end. I have comments and emails that reference things deep into the post. I feel very strongly that people can read something long-form on their phones, but it’s very easy to get bored. And so a lot of the formatting I do is to [make the post] feel as visually varied as possible. Lists are a nice visual break; block quotes are a nice visual break; having generous images throughout; having a few short paragraphs and a few long ones. Sometimes I see newsletters where I’m like, the content in this is so good, but there are no images. There are no subheadings. There are no little lists…Visual variation is how you get people to read long-form on a screen.
But Jasmine and I are both people invested in media institutions and newspapers, and especially working with editors!—so we get into why editors are irreplaceable, despite Substack and self-publishing. And I get into some stories of great experiences being edited:
All of the editors I’ve worked with have taught me something about writing. The first book review I published was with the Cleveland Review of Books, which I love as a publication. And my editor there [Philip Harris] would just make these little directional edits that were so incredibly helpful and shaped every other book review I’ve done. One of the reasons I’m most sad about the economic model for cultural criticism and newspapers collapsing is just that being edited is how you learn as a writer. For a nonfiction writer, if you get edited by a bunch of really good people, that’s kind of equivalent to a creative writing MFA. But if there aren’t enough people who can be full-time editors making money, then how are you going to get that training?
I have turned in some remarkably overwrought drafts to Philip, and he’s helped me get them in excellent argumentative shape:
Bad writing and worse takes
I also loved our very stimulating conversation on bad internet culture writing and manufactured trends, which draws from an excellent newsletter Jasmine published recently on vice signaling and bait:
I read her newsletter shortly before we recorded the podcast, and it was a huge pleasure to think out loud with her about what makes for lazy internet trend journalism—and how it encourages manipulative marketing and fake news:
: If you are not in the culture beyond the internet, you will not have good enough antibodies to know whether it’s a one-off viral thing that says nothing deeper, whether or not it’s bait. So many tweets are bait. There are people tweeting about American politics who do not live in America and they are distorting our view of what our fellow citizens believe. Good culture writing has to come from being both an observer and a participant, someone who can vibe-check: how much does this random viral video actually say about how we live together?
What AI can and can’t do…and what this means for artists and writers!
Jasmine and I also discuss our attempts to use AI for writing—something she tweets about and writes about regularly—
—and how AI has been surprisingly good at helping us fact-check our own writing, and point out human-generated hallucinations.
I will say, though, AI hasn’t been as useful for research as I’d hoped:
I have tried to use it for writing with mixed results. I’m a control freak, so I never use it to write the final sentences. I was initially like, oh, can I use AI for researching sources? And I found that it only works for a certain type of source because if you’re a writer, even the sources you’re referencing are a way of demonstrating your taste. It’s the particular reference points, the juxtaposition. I really enjoy the pieces where I’m like, I’m going to take this philosophy book and I’m going to take this sociology book and I’m going to take this history of technology book. It’s fun to pull things from unexpected worlds into the same world of a piece. And so when that becomes a really important part of taste, then you have to prompt AI so excessively.
We also get into the economic anxieties around increasing AI usage, inspired by a conversation Jasmine recently had with
:
And we discuss some situations where using AI can be artistically interesting and even encourage sociability. I loved this story about an unexpected LLM bug:
Jasmine Sun: One funny story I heard about an AI lab was that during the RLHF process (getting human feedback on which responses are better), they asked raters to pick “Which response sounds more human?” And they kept picking the ones with all the typos, because humans have typos and models don’t. But then when the researchers retrained the model based on this feedback, they ended up just producing a model that couldn’t spell anything correctly. All it spit out was prose that otherwise made sense, but every single word was misspelled. It took them ages to figure out that this particular human feedback was the reason why.
Celine Nguyen: This really makes me think of that classic “the flaws of the medium become a signature” thing. One of the reasons I am, on a personal level excited about AI in art, is that it is just a funny, weird technology. It’s very complex. It produces all these strange, idiosyncratic, whimsical outcomes. I think there’s something about artists being able to work in this very dynamic system and deliberately pull out, what are the weird data biases? What are the weird amplification effects? Treating model collapse as something that can become artistically interesting and something that you can work with.
If you’re interested in artistic uses of AI, I highly recommend this episode of
’s podcast, where Jasmine interviews the technologist on what is (in my opinion) one of the most artistically interesting uses of LLMs: a poetry camera that generates text describing what it sees.
If you’ve missed the link to our conversation, here it is again:
And I’d love to hear your takes on our takes—here, on
’s newsletter, on Substack Notes (tag us!) or elsewhere on the internet!
LET THE RECORD SHOW, HOW FICTION WORKS & possibly FOWLER'S ENGLISH?
I always appreciate you appreciating others, your gratitude, your effort to see people and find their goodness. The other learning, books, how-tos, will come, but it comes faster and is more enduring and joyous when the process is kindness and gratitude. Thanks for being you.