Y Combinator Startup Podcast - #49 - Microbes, Robots, and Ambition - Robin Sloan on His Novel Sourdough

Episode Date: November 29, 2017

Robin Sloan is a writer and media inventor based in Oakland.He just released his second novel, Sourdough.Kat Manalac is a Partner at YC.The YC podcast is hosted by Craig Cannon.

Hey, this is Craig Cannon, and you're listening to Y Combinators podcast. Today's episode is with Robin Sloan and Katman Yalik. Robin's a writer and media inventor based in Oakland and Katz and partner here at YC. So this October, Robin released his second novel, Sourdough. And a couple years ago, he released his first novel, Mr. Penumbra's 24-hour bookstore. So before we get going, if you could take a minute to review the podcast, that'd be awesome. All right, here we go. So this is kind of a weird jumping off point, but I listen to you.

on, I think it was a Mother Jones podcast, and you very briefly mentioned a machine learning experiment for the audiobook. Yeah. Could you talk about that a little bit longer? Sure. Yeah, of course. Well, okay, so the background is that as I've been working on these books that are in a lot of ways traditionally published, even though I have an interesting, very sort of forward-thinking publisher MCD, they still get printed on paper and, you know, sold mostly in bookstores and online and places like that. As I've been working on all that stuff for a few years now, I've also, like many people in this area, many people that I'm sure you guys know, I've been really interested to the point of sort of preoccupation with machine learning. In particular, the creative applications, like less

the super practical sort of uses and the ways that it might transform the economy and all that. And, I mean, truly more like the ways that we can use some of these systems to mash and mangle and just interact with like words and pictures and sounds in different ways. So that's all preface. The audiobook, first of all, it's actually interesting to know for folks that aren't totally plugged into the publishing industry, audiobooks are huge. It's like they're growing like gangbusters. Every time I go to a bookstore and do a reading, I ask people like, how many of you

also listen to audiobooks? And I mean, truly everybody's hand goes out. It's like just a really, really popular way to consume this media. Yeah. So kind of in step with that, audiobook producers have gotten really serious and frankly a little bit demanding. They're like, okay, Mr. Sloan, it's time to produce the audiobook. What do you got for me? You know, like what can you produce or what will you produce that will make this a little distinct from just the printed book? They don't want to just do like a recitation of what's on the page. So for this book, Sourdough, the story happens to hinge on this sourdough, you know,

this little funny community of microbes that you use to bake this delicious bread. And in the story, there's a starter with some strange properties. And there's also this singing. There's like this music that I don't want to give any spoilers, but it kind of helps the starter grow. And it's all part of this mysterious package. So I described the music over and over in the book. I mean, like at great length.

Like, oh, it's so slow and sad and mysterious. And it's like in a language that no one understands. And so come time to produce. the audio book. I was like, if you go through this whole thing in your earbuds and you never hear even a scrap of this music, like, it's going to be kind of disappointing. Unfortunately, we did not have the budget to like hire the people who invent Dothraki for Game of Thrones or Klingon or whatever. So we need some other way to, to like synthesize this sound that would be truly alien. Like it's not any language on planet Earth. It's something fiction.

something invented. So this is where it loops back around to that obsession with machine learning, as I think you guys probably know, one of the things that these models can do really well is sort of take a corpus of stuff of, you know, training material and extract some patterns, some more general patterns, and then use those to generate something new, but different, you know, not just kind of mimicry of what you put in. So I mean, at that point, it actually, you know, took a lot of kind of learning and tinkering with the code and actually struggling with the code. Is it at this point? TensorFlow or?

This was in, actually, funnily enough, this program was in Lua in Torch, the original Torch, and totally a testament to just the power of this open source ecosystem. I mean, this was a paper written by one group of researchers implemented by this like rogue, mad machine learning genius in the UK, this guy named Richard Asar, who's just like, I like bow down to him and his generosity truly in like making this really wonderful and very usable implementation of this tool. It's called sample RNN. And it takes as many MP3s as you want to feed it, chops them up into bits, churns learning for days and days and days, at least on my deep learning rig. I'm sure Google would be like, got it. For me, it took a few days. And then in the end

spits out this really, to my ear, at least, weird and lovely kind of generalization. You know, it tries its hardest to learn the essence of that music you fet it. Of course, it kind of fails because none of these models are actually that good yet, at least stuff of that level of complexity. But the way in which it fails is really interesting. So it's all to say that now in this audiobook, there is just these little whispers of this fictional music in this fictional language. And to my knowledge, it's the first time that like the creative output of a machine learning

system has been included in an audiobook. Whoa, that's huge. I mean, I actually think it might be a tiny distinction. but I am all about tiny distinctions. I just want to rack up all the tiny steps forward in the state of the art. Have you used the model to create anything else? Oh, yeah. I mean, basically, after you've assembled one of these rigs or a couple of them, in my case,

you feel bad about ever letting them sit idle because, I mean, that's what they're for. They've got these big, beefy GPUs, and it's so hard. Truly, it's so hard to get all the weird little libraries and dependencies all lined up right. So having, like, done that, you never want to let them sit. So, yeah, like I have some models that are kind of always churning on different bodies of text, you know, just to try to see what happens and what emerges. But there's one where I've thrown in all sorts of different kinds of music and sound just to see what it sounds like on the other end. And sure enough, I mean, sometimes what comes out is just kind of a messy garble. And sometimes it's really interesting.

Can we listen to it on SoundCloud? No, no. You can listen to it on the sourdough audiobook, which is available on audible.com. But no, none of the other stuff is public yet. It's not quite, it's the other stuff is still at the level of, you know, and a lot of this stuff is kind of, I think, in this state. It's like that sort of frothy, fermenty, experimental, like, ooh, there's like something there. But I personally don't quite know what that something is. And so a lot of my work with this machine learning stuff is kind of trying to push through that to get to like, you know, a thing, like a thing that's actually worth sharing on the other end.

And that is definitely a work in progress. Man, what were the input files? Yeah. In this case, it was almost like a short circuit straight from my own inspiration to the output because, of course, you know, this all started. This project was only necessary or, you know, I only sort of thought to do it because I had written about this fictional music and this fictional language in my book. And that came from my own experience listening to a kind of music that I've just long enjoyed and sort of thought was beautiful. It's a kind of Croatian folk singing, all a cappella, sort of a chorus of voices in sort of odd. harmonies. It's called Klapa, K-L-A-P-A. And so I just had my like, you know, folder, pre-existing

folder of just all the Kloppa MP3s I'd ever. How did you come across Kloppa in the first place? Oh, who knows? On the internet. I mean, that's like always the answer to that question. Yeah, somewhere on the internet. And it just, like for years, I just, I'd long loved it and thought it sounded lovely. And so of course it was in my brain when I'm when I'm imagining this stuff and putting it into my book. And so then in a way, I mean, this might be reaching a little too far, but I kind of feel like the machine learning system, and I did the same thing, just on different tracks, like input, right, and like the experience of listening to something, followed by some munching, mashing step and kind of abstraction step.

Of course, in mine, I kind of had my response reaction in the music and the neural network had a different thing going on, but then in the end, we both spit out something, actually something new and different and transformed, but still obviously based on. that same input. And did you educate yourself similar around robotic arms? We've done all these like basically technical advisor interviews for TV shows and movies and stuff. And I was researching her the other day, hoping that there was someone there.

And Spike Jones explicitly said we don't want a technical advisor because I don't want to be bounded by reality. Yeah. I just want to go for it. That's awesome. Actually, I wonder, I don't know if I like that or not. I think I don't actually. I think there's probably,

I respect the people who are able to like dive deep and nerd out with the experts and remain unbound. Like you have to sort of take, you have to be sort of greedy and like take everything that they tell you and they be like, cool, got to go, you know, and off into the realm of imagination and strangeness and wonder. I feel like that's, because I feel like that's sort of self-serving because I feel like that's sort of my model.

I do love to read about this stuff. I do not own a robot arm of my own. I've never personally operated one, but I have seen them operated and I think they're pretty amazing. Not just, you know, like mechanically or sort of computationally, aesthetically, I think they're really interesting and lovely. And yeah, and so the inclusion of lots of robotic arms and people talking about robotic arms and thinking about robotic arms in this book was definitely driven just by my own,

yeah, interest. Did you have your own technical advisor? all just. Not direct. I mean, that's the great thing about living the Bay Area and also just, frankly, the level of tech journalism that exists today. Yeah. Boy, you can glean a lot and not just surface level stuff, but like the really deep mechanics, just by reading on the internet and kind of going down those rabbit holes, YouTube, right? Like, anytime you want to go beyond these sort of description, be like, okay, well, what does it really look like? Or like, how does it, how does it swivel in space? I mean, there's a YouTube video to show you. It's amazing.

But then, okay, so to give away a little bit of the book, Lois gets the sourdough starter. She eventually starts making it herself. She ends up in this kind of hidden farmer's market. And part of her store is like making the sourdough with a robotic arm, right? And core to that is cracking the eggs. Is that a real technical problem that exists? That's a good question.

I don't know for sure. I suspect, yeah, I call it like the egg problem. capital E, capital P. It's like, you know, like, and all these roboticists are like, yes, all that stands between us and domination of all the world's economies is AGI, then the egg problem. Yeah, exactly. And so I don't know. I kind of suspect that it is a problem.

It has been my perception. I could be wrong. This is probably where it fuzzes into fiction a little bit. But it has been my perception that as a character in the book says, you know, there's something really appealing about these arms. working in kitchens. Like, I mean, you're like, we will have done something when an arm can like make you an omelet or do some of those kitchen tasks. But in fact, it's really challenging because that kitchen space is in fact really chaotic and unbounded. It's not one of these spaces

that's sort of at least a normal kitchen, like a restaurant kitchen or a home kitchen. McDonald's obviously is kind of kind of designed around sort of modularity and it's already pretty robotic to start with. But those other kind of kitchens, there's more organic kitchens. if you actually kind of do that thing of like defamiliarizing it and like looking trying to look at everything in that space and the way it all works together with fresh eyes you're like that's fucking impossible it is just the angles and the slopes and everything's a weird shape and and you can't recognize it all and um and I actually think that's wonderful I think that's lovely that it's like oh man what a what a magic show actually for a human cook to manage all those things

at once and then reach over and just with one hand go and crack an egg yeah so Yeah, well, because you kind of, you spoke to it in the book around having two hands, basically. Like, I can't do that either. I've tried it today. I think like, oh, yeah, this will be easy. Just crack an egg with on hand. No, not so much. I will tell you that you can learn.

It is in YouTube, once again, YouTube is your guide. I think in the book, if I remember right, I actually have, I have lowest learned by watching YouTube clips, including some that are so great. They're just, because, again, truly, everything, everything you need to know. And every task, no matter how small, is somewhere documented on YouTube. And often it'll be like these disembodied hands. Just the video is like a minute long. And they just kind of show you and demonstrate and do a test run and then, you know, tap it on the edge and it comes right open.

And I was skeptical at first. But in fact, it is not that hard. Right on. Yeah. What was the inspiration to do this book? Inspiration to do this book was, you know, it's actually a lot. Like the other books I've done, it's really. really living here in the Bay Area, which I, I mean, truly think it sounds a little dorky,

almost a little Pollyannaish, but, but truly, I just, for all, for all of its complexity and problems and everything else, I think it's just a really, really inspiring place. I think it's a place where people do interesting things and end up leading these really interesting lives. And so, yeah, over the last few years, I personally, just as a human, had gotten, have gotten pulled more into the world of food. First as an observer, kind of just like someone curious about it. And slowly as more of a participant, although I'm still much more of an observer than a participant. And it just, I mean, this is what happens with everything.

I think it happens to journalists too. You dip a toe and do something and you realize that it's just full of story. I mean, there's so much stuff there, so many little dramas and mysteries and mechanics that you can pretty easily turn into. into plots. And so that's one of my questions is that your stories from penumbra to sourdough kind of explore the intersection between, you know, tech and startups and old crafts, like breadmaking or, you know, book selling. And so what about that interaction is particularly interesting to you? Well, you know, I think it's some, it's, it's, it's at the conjunction between them. And it, the honest answer might be annoyance. I'm so routinely annoyed when the old and the

you get framed as an adversarial relationship, basically when people say or, or versus. And, you know, of course this happens with books. And it was so much the sort of conversation, especially around like e-books and print books, like the internet and old school publishing, like, well, which will it be? You know, which will triumph? Or like, what is the road forward? And I don't know why. I don't know if it's just temperamental or I've had good mentors and good advisors or, you know,

I've had the opportunity to read smart thinkers for a long time. That has always seemed so nonsensical to me. Like it seems to me like it is always and. Instead of the new thing, replacing the old thing, it all just piles up in this like multi-car crash, nonviolent crash, glorious crash, everything is just kind of like and boom, blah, and the mountain is getting higher and higher and higher. And I don't know, I find that totally exciting because it means we get new things all the time, but we, also get the benefit of crafts and ideas and obsessions that have been sort of compelling people for a really long time. So I think mostly, not mostly, but at least in large part, my books are intended as a sort of rebuttal to all the people who want to make us choose between the old

and the new. I'm like, nope, I want both. Option C. Yeah. Do you have a name for the genre that you exist in? Because I was talking to Kat before about this. And there seems to be, you know, increasing amounts, possibly infinite amounts of dystopian, like black mirror type content. But you're kind of into this like positive satire genre. What do you call that? Oh, man. I don't know is the short answer. Someone did suggest a genre tag that I quite liked.

They said, I think your books are sort of like puzzle fiction. And I just like the sound of that. I'm not sure that it's an actual genre or even if my books are that puzzling. but I was like, if I ever saw my books on a shelf labeled puzzle fiction, I would be very happy. I mean, in my heart, they are sci-fi-inflected literary fiction. And all I mean by that is I try to take care with the sentences and just, you know, produce language and, you know, scenes that are like interesting, somewhat interesting to read and not just like glop on the page. at the same time, there's no denying that my core genre as a reader is science fiction. And so that stuff just gets kind of metabolized and regurgitated, I guess, or transformed.

I think it's refreshing to see sort of like a non-dispopic. Like there are funny, you know, elements like the slurry drinking that I definitely personally recognize. But I think it's there, I get an overall like positive sense about the world you're building. Yeah. Yeah. I would say so. I mean, part of that for me is root. And this is not necessarily a sort of dystopian or not distinction,

but part of that for me is rooted in the desire to have like narrators and other characters that you actually want to spend time with. And that's only because my most memorable experiences of reading from, you know, early childhood onward have been ones where I finished the book, you know, closed the back cover. And I missed the mind that I was spending time with. And there's a lot of good books that are, you know, compelling, engrossing and rewarding that aren't that. We're like, ooh, what a bunch of creeps, you know, like what a bunch of broken humans.

I mean, there's a lot of stuff like that. Books and TV alike, probably more that than anything else. Maybe it's actually for exactly that reason. And I'm like, okay, cool. Other writers have got that on lock. They've got the like, everyone is broken beat, nailed down. and so maybe I'm going to do that thing that I like so much, which is, you know, voices that you miss.

No, I love it because I was talking once to a YC founder and he was saying, you know, if you talk to the vast majority of founders in Silicon Valley, they were influenced by things like Star Trek or Hitchhiker's Guide, and these are all like pretty rosy visions of the future. Whereas, you know, right now we have black mirror and all this really dark stuff. It's like, what is going to be built 20 years from now? and the young people reading this and watching this. That's a good question.

Boy, you know, and I feel like that should almost be like a challenge. That's a great sort of like challenge to the science fiction writers working today. And I think some of them, by the way, are meeting that challenge. I think people like Corey Doctor Rao, Annalie Newitz here in San Francisco is an amazing science fiction writer. They don't write dystopias. They don't write those sort of, you know, naive Panglossian features either. Like, and guess what? It all worked out great.

Like, obviously that's just. It's like that's not actually useful either. But yeah, to find a path through the thorniness of real life and power and, you know, the way that people inflict pain and other people, but still remain kind of like, whoa, excited about what's coming next. I mean, I do. I think that's, I mean, it's what I like to read. And I hope that's what people are, I hope they're still producing it. Well, I mean, even in a couple interviews I read with you and listened to, you were influenced by children's books. When you're ready,

you're cloudy with a chance of meatballs. You specifically mentioned. Well, yeah, for sourdough in particular, there is a great genre of, I mean, this is even, the genre is very specific.

Talk about specific, specific little slices of books. This is the sort of like runaway mutant food genre. You've got your clad with a chance of meatballs. You've got your streganona. Stinky cheese man. Yeah,

absolutely. Lesser known book called the Duchessus Bakes a Cake. Also great. And basically it's like all like food that, you know, grows out of control, becomes giant, and there's clearly something

that, like, resonates deeply with the child's mind because it's like, you could totally populate a shelf with those books. Oh, man. And were you reading these beforehand or you're just like,

I think this needs to go a little further. I can't say that those were the direct influence for this book. That was more of a kind of hook shot where after writing this book, I was like, oh, wait, you know, I think this has a lineage.

The more direct influence for that kind of stuff in sourdough, that kind of like runaway food vibe was truly it's all the scholarship that's been done in the last few years, you know, five years maybe going on a decade about microbes and the microbiome. And just the daily, near daily wonder of like, oh, let's see what's in the news today. Microbes do what? Like, they what? Huh?

They talk to each other and they're controlling my and there's how many and we don't even know. I mean, it's like, I truly, it's amazing. That's got to be such a fun discipline to be in now and for the last little while. And yeah, so I was like, yes, I need to put some characters in this book that are not people, but microbes or microbial communities. And try to give them some personality on the page. And did you think at any point you wanted to make an app as well? What for this book? For Sowerdow?

Yeah, well, you know, you're talking about staying around with these characters. And I use fish a long time ago because when I actually sat next. to Patrick Mowberg while he was building it. I was working out of Beto work. Oh my gosh, that's amazing. You're going to get out of my life, man. But I just been wondering, like, if you would consider building some kind of extension.

I had not considered it for this book. I, you know, there's a thing you hear where people say, like, okay, you have to come up with a story and then you find the right vessel for that story. Like, does this story want to be a book or a video game or an HBO series? Whatever.

I think that is wrong. I do not think that's actually how good things get produced. So if you ever hear someone say that, you can tell them. Robinstone thinks you're full of shit. I think it's at least for me. And I'll speak. I'll speak actually just for myself. It's almost exactly the opposite.

You get interested first in a format. Either just, you know, out of pure kind of like childlike interest. Like, oh, that comic book was so good. Or like, wow, an amazing movie. Or like, even like that YouTube video was so weird and, like, creepy. Whatever it is. And then at some point, you start. to sort of feel that itch of like, oh, maybe I could make one too.

You know, maybe I could play in that sandbox. You can trace everything I've ever made, whether it's a book or a weird digital experiment to that impulse of truly starting with some form that I admired and thought was awesome and then trying to figure out like what kind of fits in that box. So that's all to say that this book came out of a sort of a book-shaped impulse. And yeah, I think the next app is going to come out of an app-shaped impulse. Have you had any impulses since? Like, what's the next form?

No, I definitely have. I mean, too many. The problem is you're like, there's so many cool things. I will tell you that I'm kind of still in, I feel this like tension. It's totally unresolved. I don't have like a plan, a theory, a solution or anything like that. But the app thing right now, it's super, well, you know, I'll make it really practical. This is what happened.

I made this app called Fish several years ago, totally on a lark. I mean, because I kind of admired the iPhone screen. In particular, the lack of a row of browser tabs across the top or like other windows tiled in the background, you know? Like it is the case that often, more often than on laptops and other things, people will just do one thing on an iPhone at a time. And that's actually really lovely. Anyways, thought I'd take advantage of that, made this app, it was great, well received.

I had a lot of fun making it. Years go by. The newest version of iOS, something is, changed and now that, like the binary that I uploaded to Apple all those years ago will not run anymore. It was getting a little funky, like it looked sort of weird on the newer iPhones, but you could still run it. Now I am alerted by Apple via an email. It's not going to run anymore. Because I am a writer and a tinkerer and not a professional software developer, I could not immediately find the source code to this app. And I rooted it around and I dug out some old

laptops and I booted them up. I just searched everything. I checked my email. I was like, surely I must emailed it to someone, I can't find it anywhere. So I think it's gone and that bums me out. It is sad. It is sad. So I fear, as of this moment, my current thought is that Fish is lost to the world or at least users of iOS 11 and above. But there's actually something, I mean, that's just my sort of carelessness, which is

a bummer. But there's actually something embedded in that, which is you make these digital things. And unlike a book where once you bind it up and put it on a show, even if it gets forgotten, even if it gets damaged or, you know, whatever, it's like still it boots somehow. You can, you can like compile and run that printed book basically forever. And boy, that is just not the case with digital stuff. Like the ground shifts under your feet and suddenly the platform you built it on goes out

of business or, you know, the OS gets upgraded and you're like the APIs are different. And that's unsettling. And to sort of figure out how to navigate that, how to still embrace the fun of those digital experiments, but like not sign up for infinite ongoing maintenance for like an ever increasing number of projects. I just don't know. So that's, it's one of the things I'm really thinking about right now. I think it's a, it's very distinct from the way that frankly people in like companies would

think about making and maintaining apps. But there's plenty of people out there who aren't people like me who are kind of just tinkurers, experimenters, writers and artists. And I think it's kind of a funny moment for that kind of work right now. Well, if you actually look around at most people who work in the tech industry, a lot of the work does fall away. Like if you spend two years working at Facebook or Twitter or Google, a lot of your work is gone six months after you leave. Totally. So maybe the question is how do you give something like an everlasting life online?

Right. Right. I think it is such a big question. I mean, there's stability comes, it seems at least partially with time. I mean, if I was going to make a web page and not a fancy one, like a real simple one. And then if I took that webpage and I made sure to kind of have it hosted in a few different places, but it's a experiment. Say it's a story and it's presented in some cool way. And there's a few kind of interactive bits in JavaScript, say, you know, as the page proceeds.

And I host one on my website and maybe I host one somewhere else. And I also make sure that it's indexed by the Internet Archive. I still don't feel as good about that as I do about a printed book on a shelf, but I feel pretty good about it. I mean, I think there's like, you're like, okay, webpages have been around for a one. But like 30 years now, you're like, finally we can kind of count on this technology continuing to be legible and accessible. But at the same time, you're like, that's not the most exciting thing. Like you don't, I can't make a cool mobile, I don't know, like AR enabled experience

with a web page, really not in the way that I would want to. And so there's basically, I think there is this tradeoff. And maybe it's really natural between that sort of like bleeding edge coolness and interestingness and that sort of stability in. I don't know. I don't know if it's just like the archive or I don't know. You want to be able to share with people. Well, a lot of now is just falling into the hands of basically archivists in museums.

Yeah. So you see it. It's happening. I think rhizomes doing a project to preserve digital art. Yeah. But you basically are like buying, you know, a Dell from 96 and like booting it. Yeah. And like air gaping it and making sure that the automatic like OS updates never run.

Like do not connect to the internet. Yeah. Totally. Totally. It will. Somebody. somebody I mean yeah the museums have those collections I would love to imagine there's someone somewhere that's just like yes I have like the um the menagerie every computer every OS like what's that you need a mac a C running OS 7.2 I've got one of those boop powered up well it's it's a big thing in TV too we did I am a podcast with Coridana from Mr. Robot and it's so funny because their show is only set two years ago but they're still trying to

to keep like a consistent date line. So they're actually having to get apps from like two years ago and then to do the screenshots and the simulations and they're like, we can't even, you can't find like Facebook from two years ago. Right, right. That's admirable, first of all. That is some, that is some like 21st century Kubrick attention to nerdy detail. I really appreciate that.

But it is. It's right. It's a weird time. I would not bet against everything that we're pouring into kind of the vast box, you know, like everything. The Facebook updates, the Instagram photos, all of it, basically being gone in 2060, 2050. Like I hope that's not the case. And I mean, it seems hard to believe because it's like everybody's everything is in these

systems. But based on just what we know and the way the patterns we've seen already, I like, if we were like, okay, place your bets, I would be like, and that will all be deleted. Speaking of deleting, Craig was telling me something that I hadn't heard before that you have written a script that deletes your tweets. Oh, yeah. And can you, how did that come about and why? Yeah. Oh, no, no.

So this is very, this is very clear cut to me at least. I still love Twitter. Even though Twitter is so fraught and so complicated, I started using it like a lot of people kind of around 2009, 2010. I actually worked at Twitter for a couple years, too. and it was very at that time like everyone who worked at Twitter was like on Twitter deeply and it was really pure fun

and I think it was because it was so much more casual it was the era of like grabbing a sandwich with at Craig and at cat you know it was just so you're like oh we were children then so things have changed a lot but I but for me at least I think

there are some ways to kind of anchor it in that other feeling I like to think of Twitter as not like sort of an ongoing transcript of a congressional hearing, you know, to which we can always return and be like, well, actually, cats, four years ago, you said this. And that was bullshit. But rather as like a huge, weird, interlinked conversation in a cafe somewhere, you know, people making bold claims and referring to things and saying nice things complimenting each other, but then all of it just kind of like fading into the night air and it's gone. Like you wouldn't,

You wouldn't record your cafe conversation for posterity. I mean, if you would, maybe some people do, but that's weird. That's clearly like sort of strangely antisocial. I record everything and transcribe it and index my whole life. I don't know what you're talking about. Yeah, that's a whole different conversation. But it is, it is like that's actually quite strange. And yet it's the default for Twitter of like recording everything and sort of keeping it forever.

So that's all to say, my very tiny kind of lever against that, that system. and my way of kind of keeping it something like a cafe conversation with the words just fading away is to delete my tweets on the regular. And oh, I can report that it feels great. That does sound quite refreshing. Yeah. Yeah. Were there any like greatest hits? You're like, sad to see it go.

It was, you know what? Yes. And it is a good. It's a healthy thing. You can contend for a remote with your own vanity. And you're like, oh, but there's just so many faves. And but then it's gone. And you're like, I didn't need those faves.

It truly is healthy. It's like a scalding sort of cleanse. So this brings me to one of my questions. I think I first came across you when I was reading Snark Market. And so I loved it because, you know, you would, you know, post an essay and then it would kick off a sort of discussion or debate below it. And sometimes they were sort of contentious. And you came out reading the essay feeling one way.

And then after reading the comments, you were like, oh, no, but I see these other sides of it as well. and that died in what 2015? I mean, it was like a lot of blogs. It was just a sort of slow tail off. I mean, I'm sure you, I've never done this. It would actually make me sad to do it. I'm sure you could graph the post frequency. And it would be a real kind of actually a slow ramp into this sort of bulging sort of heyday.

And then a slow tail off. And furthermore, I'll bet if you took that and compared it to like average hours employed per week by all snark market authors, it would be a perfect sort of. They would fit together like puzzle pieces. That makes sense. I was going to ask if Twitter killed it. I mean, it did. It did in part.

Boy, I mean, that's just a whole thing, a whole melancholy story to kind of unspool, the story of blogs, the rise and fall of blogs. Because I think people who kind of came up during that time, not only as bloggers, but kind of as writers more broadly, because it was. It was this amazing way to kind of cross-train and just force yourself to like, write something in public every day, even if you didn't have it all perfectly locked down or even know exactly what you thought. You're like, well, I'll find out what I think by like starting to write this little blog post

in my movable type installation. It was truly lovely. But yeah, there's no question. It was killed by Twitter and Facebook. And just really the sense that people moved on. I mean, the reason you wrote a post is because you knew people would read it. In some cases, you knew very specifically who would read it. Like, I remember in some ways the greatest pleasure of writing for that blog,

as I understood our position exactly in like the food chain of blogs. We were not at the bottom. And in fact, some of the blogs we were reading were like closer to the bottom, just in terms of number of readers. They were like super weird, specific, nichey, like strange little blogs. And when you found one that you liked, you'd add it to your Google Reader, long lamented. But, I mean, the loss of Google Reader, it was. was another true I mean truly it was one of the nails in the coffin of blogging um but you

add as your google reader and that became part of your like secret stockpile and i remember talking to people about feeds and they would be cagey they'd be like oh whatever it's just a just a scientist feed i found somewhere it doesn't matter don't wait about it i'll blog the good stuff anyway just you'd find interesting things you'd write about it snark market was not top tier i mean there's many sort of tiers above us but there were a couple much bigger blogs in particular and just you know through the experience of linking something and writing a little about it and then maybe eight hours pass or a day passes and then you see it at that next blog up I began to understand very clear I was like I knew our role in that system in that sort of economy of

ideas and it was really cool it just felt awesome to like have a place but um as those readers go away like as basically it's like there started to be these holes in that mesh and then I think the parts kind of shear off and then disintegrate entirely and now I mean you open up the it's WordPress now we did it we did upgrade sometime between 2008 and 2017 and you open up the little composition window and you're like no one is going to read this or if they do it'll be only because I essentially recreate it or give it a new home on Facebook or Twitter do you have strong opinions on 280 characters I do I do now I I have to just always remember this my first experience

with like user rage being an enraged user that is to say was in college I was my first experience with high speed internet and I loved the New York Times website which was I mean it was probably designed to be like 500 pixels wide at that time it was just this tiny little thing

very proto you know web experience but I loved it and in the early days one of the things they did is they actually you know to keep their timesness they rendered all the headlines as images in like the Times font because there weren't like web fonts and they couldn't do it justice, they thought,

and so they rendered this as images. And I thought that was amazing. I was like, of course, their images. It's so beautiful. It looks like a newspaper. So at one point, very reasonably, they decided, this is a little weird and kind of rigid and inflexible and probably, you know, wasting people's bandwidth.

So they switched to just normal text links. And it was like Times New Roman. It wasn't a pretty font because, again, no web fonts yet. So they switched it, and I wrote an email. I was probably a junior in college. I was like, to whom it may concern, to whom it may concern, you clearly don't understand your own brand. You've made a terrible mistake with these ugly text links. So dumb.

Literally 48 hours later, I was like, oh, man, this page loads a lot faster now. This is great. It was just a good grounding early experience with like, come on. Things change and you're going to be okay. And so I have retained that consciousness to this day. And I was not super upset about 280 characters, even though I, of course, like many of you, I appreciated the sort of the sort of haiku constraints of 140.

But I, oh, man, I love it. I just keep typing. It's great. And someone made the point. I thought this was actually very super, super sharp. There's, you know, a lot of different ways to think about and talk about it. Someone made the point that with 280 characters, it actually gives you enough space.

It's not that much space, but it gives you enough space to do something you almost never could with 140 characters on their own, which is to present one idea, kind of like set up the thing, and then turn it around, react to it. You could basically say, well, a lot of people say X, but I think Y. Or you could say, like, I used to think X and now I think Y. Like, there's enough space for, like, two ideas to be in dialogue with each other. Whereas before, and after this, it was like an academic who kind of pointed this out. I was like, oh, yeah, that's right.

Before all tweets were like, blah. here's the thing here's the thing I mean even if it was a nice thing or a funny thing they were just like blah but now a tweet there is enough space for it to be kind of a like

hmm or like what about and that might be healthy I think hmm I don't know if any fewer people would be taken out of context but it makes sense I do appreciate the little circle like progress thing rather than the countdown

I thought that was really slick yeah yeah totally yeah we've been and it is it's one of the things we're like oh right I've been playing a weird game this whole time. And I didn't even realize. And as soon as the numbers go away, you're like, right.

Didn't matter. Yeah. That's really funny. So what are you excited about right now? Are you thinking like, all right, I mean, maybe you've already sold a book and I don't even know it. What are you into? Boy, I'm into a lot of things.

I will say that the machine learning stuff continues to preoccupy me. I think, in part because I think there's an art. There continues to be an arbitrage opportunity for artists or, you know, kind of art adjacent people. So much of the energy is focused on super practical and like economically valuable applications, which is fine. And I totally, you know, I truly like I cannot wait for self-driving cars to come fully online and for robot arms to be doing, you know, all sorts of tasks and all that.

But the creative applications are really, really interesting. And I think for people who have even like just the fingernails grip on the technology and can kind of hack their way through the code and also have, again, maybe the temperament to think it's interesting and not like, like rise of the machines. It's like totally wide open, like just blue sky, interesting stuff. So I would definitely like to try to write a novel. And I'm in the process of trying to write a novel that has as kind of part of its text, the product of some of these machine learning systems. But it's, you know, it's tricky because you want it to be good. You don't want it to just be a sort of parlor trick. Like, did you read the machine learning novel?

No, but I'm aware of it. And it's very interesting. Like, you don't want that. You actually want to produce something that's good and like worth people's time and like interesting and worthwhile to read. But you want to do it, or at least in part, using these tools that kind of no one has ever used before. So it's a cool challenge. Have you tried magenta yet? Well, magenta is not really like a thing thing. It's like the suite of different. tools. Right. They are really focused on sound, like mostly music stuff. It's really cool. I actually think, I think they're one of the sort of the few people slash groups that are doing it right. People on the podcast, people on the podcast who know I'm making air quotes, doing it right.

And that's only to say they're super competent technically and interested in kind of pushing the state of the art in that sort of very codey, sort of academicy sense. But they're also very clearly just interested in like aesthetics. Like they measure, you can tell that they're measuring their own, judging their own output, not on like, well, we've achieved a new state of the art of 2.7 bits on the loss. You're like, cool, is that better? They're like looking at the, they're looking at the sort of the qualitative, like, well, is it, does it sound interesting?

Is it lovely? Could it be lovely? And they're also, this is actually the best part. They're building tools, right? It's not just raw code. They're like building these things for people to use. and very clearly like learning how to use them themselves kind of in public in real time. And I just admire the hell out of that whenever I see it.

I think it ends up offending a lot of artists because so many of these image recreation things in particular are like style content transfer. See? Right. Style doesn't matter. We can just throw it on anything. Right, right, right. Exactly. Yeah.

Yeah, it is. And there's something about that stuff that is. I mean, it's exciting in its own way, but it is at the same time a little tasteless and a little bit like, oh, really, that's what you think it is? You think it's just like, you think you think. You think Van Gogh is swirly trees? Okay. Come to the museum with me. There's some things we need to talk about.

But it is like I think making something that you can learn, that you can use is kind of the key. There is a lot of this code, including some of the style transfer stuff where you kind of like the only way to use it is to kind of say run, like do your thing. Tell me when you're done. Which is, that's a good starting point. I do think that the challenge or that kind of the next step that gets really. really interesting is learning to make tools or even instruments that like a person can get better at,

which is not easy to do. I mean, people try to invent instruments and often, like musical instruments, and often you're like, there's just not enough depth there. You're like, yeah, I can play hot cross buns and like, and that's all. You're like, this is no viola or like this is no, you know, cool MOOG synthesizer or whatever. And I think that's coming for the machine learning stuff, or it should be. What do you think is missing right now? Well, I mean, this is a little self-aggrandizing, but I'll describe one of the projects that I did,

which I think is a tiny, tiny step in that direction. So there's these neural networks that operate on text, and they can essentially learn these statistical models of text such that given the beginning of a sentence, they can just keep writing it for you, right? And they end up, of course, of course, being sort of wacky and nonsensical, but they really do learn something.

Like if you train this neural network on all of Harry, Potter and you start a sentence like the boy walked into the it'll say castle where he picked up his wand and like you know cast the following spell like it really and like and like in the style of jk. Raleigh like in really actually an impeccable sort of imitation of the style of J.K. Rowling or whoever. So you've been able to kind of do that on the command line for a while and there's you know people have written blog posts about like oh look I wrote some Moby Dick or here's some weird fake Shakespeare was all really cool. I took that style. stuff kind of changed the way it works a little bit.

I thought hard about the corpus I wanted to use, like what I wanted it, what style I wanted to be learning. But then I also built a little plug-in for the Adam text editor. It was really easy to extend. You just like kind of designed to have things modded onto it. And so now, instead of it being this kind of command line thing, you're typing in a window just as you would, as you'd be like composing a manuscript and you hit tab the way you would to like auto-complete something on the command line.

That's great. And a little wheel spins for just a second, and then it goes pop, and it shows you the completion. But it's not just, you don't have to just take what the computer gives you, just as you can with auto-complete on the command line. You can arrow up and arrow down through other alternatives. But I think this is really important. It's like the core, the core interesting thing is the weird sort of wonderful output of this machine. Right.

But at the same time, you acknowledge that having a human kind of curate it and shape it and form it and just be in the loop and be learning how to use it is just as important. And it turns out, I mean, artistic value aside, it's really fun. It's actually like a really fun thing to play with. Well, you see it on Twitter with the auto-complete keyboard all the time. Exactly. Just tap the next thing. Yeah, totally.

To take a Harry Potter tangent for a second, early on in the book, you talk about our generation wanting to be sorted. Is that like a strong opinion or a throwaway line? No, it is. That's funny. That's a funny thing for you to mention. Yeah, so the whole line is the character says, you know, I'm a child of Hogwarts and like, like everyone else in my generation, I'm a child of Hogwarts. And more than anything, I think we just want to be sorted.

When I read that, I was like, that is very true of me. At least in my early 20s, I was like, just someone tell me what to do. Totally. Exactly. Tell me what club I belong to. Give me my shawl, my cardigan in the right colors. And let's just get this over with. Yeah, so I do believe it.

I think there's something to it. However, more than anything else, that line has been a window for me into like what reading is in the 21st century. Because in part because it's near the beginning of the book, in part because it's like very hashtag relatable. You know, it's not about the book. You can pull it out very easily and it kind of stands on its own. And because apparently it resonates with a lot of people, people have taken so many pictures

of that line and that page on Twitter, on. Instagram. I could compile you a gallery of like hundreds of snapshots from around the country in the world of people. And like, and then with a caption, just like, you know, quote of the day or like, oh man, Robin Sloan really gets me. Or like, I got to admit this would hit really close to home or whatever. They're kind of putting their little spin on it. But just to see the way like a book becomes a social object like that. And it's this thing for them to both, of course, just like everything else on social media. They're like announcing that they're reading a book, which is like pretty cool. They're sharing.

something that resonated with them. They're promoting the book a little bit, which is cool. I definitely know. I'm like, yes. And be sure to say what book that is from. Sourdough by Robert Sloan from MCD. I'm published on, you know, October or whatever. So it's cool.

It's actually been like almost like, you know how they put like radioactive dye in your blood to like kind of like trace, whatever? That's been my like dye tracer through. Is there one later in the book? There are a couple things that people like on Kindle highlights. you can see what people liked. But honestly, there's no other line that had that, like, that became, like, truly like a social object that leapt off the page and then just

kind of went off on its own. And it's so fascinating because it doesn't really tie back into the rest of the book for the most part. Yeah. I mean, it's just, it stands on its own. Yeah. And it's really, yeah, it's kind of, it's the starting point. It's how this character feels before she even gets to San Francisco or embroiled in this

world of food or learns any of this weird stuff. But, yeah, apparently there's a lot of. people in the world who feel the same way that this character does as the story opens. Have you seen long-form fiction change as social media influences it? Kat and I were talking about this in the context of museums, right? You know, you have these like Instagram museums, basically. Are there like Instagram books with all these like zingers intentionally? Yeah, that's a good question. So I don't know for sure. That's a, it's a big question actually and a really interesting one. I don't really think so, not yet. And I think that's for a couple reasons of high tech kind of just, just guess.

One is the truth is that the readership for books, I mean, it's obviously diverse and like there's a lot of young people who read books. There's also a lot of older people who read books. And I think there's a certain kind of point at which, and it's course not like these people aren't on Facebook. They are. Like everyone's on Facebook. But I think they're also kind of not. They're like people who just relate to the world and get information and kind of select their media in a different way. I don't know if it's a more old-fashioned way exactly, but it's just different.

they go to the library and get a stack of books, which is, by the way, one of the reasons I love publishing books and reaching this audience this way. I think it's, like, really special to be able to talk to them in a way that, like, no one on Twitter ever will because it's just kind of, you know, sort of separated worlds with not a lot of bridges between them.

So I think that's part of it. And then I do think that people read books for different reasons than they use social media still. That could change. That could change every time. But, yeah, I think books, the good books really do, they're that sort of sense of engrossment. Is that even a word? Yeah.

Engrocing, they're engrossing. I'll count it. So, Epic 2014. What's your explanation? Boy, that's, well, it's kind of the original media experiment for me. I mean, there's been so many along the way, like apps and web projects and, you know, weird books and other things like that. Epic 2014.

Boy, it was an artifact of its time. It was 2003, in 2004, I guess, is when it finally got published. I was working in a place called the Pointer Institute in St. Petersburg, Florida. It was kind of like journalism school and think tank. Really, really cool institution. My colleague there was another young journalist or sort of journalist in training named Matt Thompson, who has gone on to become the executive editor of the Atlantic Monthly. So, wow.

little did I realize. No, that's not true. Actually, I did suspect that that was probably going to be his path even back then. So we're down in the computer lab with one. I mean, it wasn't quite a Mac SC with whatever, System 7, but they were some funky, you know, translucent IMAX down on the computer lab. And we were frustrated because we at that time, it's like, this is the rise of the blogs.

It's really like the early blog boom. And we were part of it. We thought it was so exciting. We thought all this stuff was so exciting. And it did not seem to us like the people we were talking to in the, the newspaper industry primarily then, like, saw what we saw. And we tried to explain it. We tried to, like, give presentations with charts and things like that. And still just, like, we were getting these, like, glazed over looks. And we're like, come on. Like, first of all,

these people need to understand this. And second, it's exciting. Like, there's like, this is, if not good news, it's like very interesting news. So, this is where it's almost a little cliche. We're like, how can we communicate this vividly to people by telling a story? In flash. Yeah. Well, that's why. It was an artifact of its time. We did. We're like, okay, we're going to make this video. It's going to be this sort of faux future documentary that tells the future history of media from 2004 to 2014.

With all this weird stuff that happened, you know, consolidation and the New York Times goes out of business. And there's just like all this customized news. And it's, it's all feels very antique now, in fact. But, um, it was radical, a radical vision in 2004. But, but the great constraint, of course, is that in 2004, you could not post video on the internet. there was like there was no place to do it and if by chance you did find some server space the resulting bandwidth bill would like crush you and there were these stories almost like urban legends of like oh yeah Craig posted a video one time some people watched it he had to sell his

house so the bandwidth efficient way of sharing moving pictures was flash so we authored this whole funky thing in flash it was this like strange stuttering animation and it was actually actually, the web we have lost, like the lamented early web. Even the flash video was kind of like a bit much, because it became very popular. It was like a weird early viral web hit, especially kind of among news people. And so we were kind of feeling the burn and the server would go down sometimes. So some friendly people with web servers around the world would send these emails like, hello, my name is Ivo.

I'm in the Netherlands. I would be happy to mirror your flash movie. and we were like, yes, here, here is the link, please mirror it. So on the main page, it was, it was just beautiful. It was like this United Nations of hosting. And there's this list of like, here are some other mirrors, you know, and you click and you'd go to this sort of other copy of the page on some server in the Netherlands. Oh, man.

It was amazing. That's so cool. Have you thought about doing another one? You know, every so often someone asks, most recently it was 2014, which was kind of the crux of the movie that we made more than a decade ago. And we got some, you know, feelers from people, including, like, media companies that were like, it'll be a big thing. It'll be like our cool. We'll, like, produce it.

It'll be rad. And, you know, we'll pay you for it and all this stuff. And both of us had the same instinct, which was like, it was so, that project was so guileless. It was so, I mean, it was pure. We were just, like, these two 20-something people who were like, people seem, people don't seem to see this as clearly as we do. Perhaps if we tell a story. And we had no other expectations for it beyond that. And you just can't.

If you would do something like that and it works and it's successful, that is 100% your signal to close that box, put it on the shelf, say what luck and move on to something else. Fair enough. But so Pinnomber came out, how many five? Yeah, 2012. And so I think, you know, working, you know, I was in, you know, at Wired. And then, yeah. And then in New York. And I think around the 2012 timeframe, like there was so much positivity about tech, right?

Yeah, yeah, yeah. Especially in the press. And so it's changed a lot. So how did that like change in tone in the coverage of tech, like impact the way that you talk like right about it now? That's a great question. I will let you in on something. That shift was happening. Of course, it's ongoing.

Things always in flux. That particular shift, that sort of darkening almost of the tint was happening even as I was kind of wrapping up the manuscript for Penumbra. And I remember very vividly the last. few passes in the summer of 2012 before it came out that fall going through and changing a few lines that, I mean, the book is very, it is very kind of like magical Silicon Valley Wonderland. There were a few lines in that original manuscript. There were even a bit much then in 2012. And I was like, people are going to think I'm lame and just like, and again, and naive if I

include this sort of line. And so I, I even at that time and even at that point in the process, I was like, hmm, I need to kind of calibrate this a little differently, which is, and that's the risk. It's the fun part, but also kind of the risk and the burden of trying to write fiction that's set now in our world today. I mean, there's definitely like, you understand why people write historical novels because there's like a refuge and like it's not going to, I mean, there's other challenges, but at least it won't kind of shift underneath my feet. I definitely had to change the tone for the new one for sourdough. Part of it was just kind of natural, like the way I think about it, of course, has changed over time like everyone else. were just, yeah, it's just all about kind of power and the way these things have their role in our lives have changed.

So I think any like thinking human has a different relationship to it now. And then of course I always wanted to just like not do another version of the same postcard from Silicon Valley, but like send a different one. So yeah, the character in this new one, she starts the book in a much darker place. Like she's working at this robot factory that wants to change the world and like transform the conditions of human labor. but their labor that they are undertaking to do all that is like pretty intense and she's frustrated and strung out and not eating well definitely not drinking enough water and and that's yeah that's so in some ways I wanted to I wanted to make it feel like a dispatch from Silicon Valley today without taking the I think easy route of just like burn it all down that you see elsewhere

because I don't think that's well first of all other people have got that covered they've Other people are fully on the burn it all downbeat, and that's good. I mean, it's, we need them out there. So maybe I could do something different. That's great. All right. Thanks for coming in, Robin. Thanks for the invitation.

It's a real treat. Cool. All right, thanks for listening. So as always, the video and transcript are at blog. dot ycombinator.com. And if you have a second, please subscribe and review the show. All right. See you next week.