The Orthogonal Bet: Novelist Robin Sloan’s Love for Books with Maps on the First Page

Samuel Arbesman:
Robin, great to be talking with you. I guess I'll just jump in about ... Okay. So you have this new novel, Moonbound. It is sprawling in scope. First of all, I just want to let you know, I gulped this thing down in several sittings. It was great. It's also, in some ways, very similar to your previous novels and other ways wildly different. The previous ones are very much of the moment, while this one is about the future with emphatically capital F, where it's like-

‍

Robin Sloan:
Towering, a towering letter F.

‍

Samuel Arbesman:
Yeah. It's like 11,000 years in the future prompted you to make this change if you even view it as a change.

‍

Robin Sloan:
Yeah, it is a change or I guess you might say it's an evolution, maybe even a maturation. I think, of course, there's a lot of ways to answer that and a lot of influences that go into it, but I think one thing that anyone who read either of my first two novels, Mr. Penumbra's 24-Hour Bookstore and Sourdough, might observe that they seem to be written by someone who really likes science fiction. There's a lot of homage and reference and, as you say, they're notionally set in our real world. They really are the stuff we know and are capable of today.

‍

Samuel Arbesman:
Right. It's like the world skewed just like a tiny bit to say, "Okay."

‍

Robin Sloan:
15 minutes in the future.

‍

Samuel Arbesman:
Exactly.

‍

Robin Sloan:
Again, that affection and appetite for science fiction, I think, is really clear on the page. So in some ways, what I finally had to do is say, "Robin, if you like science fiction so much," and I do, I mean, I'm just a voracious reader of the genre, "why not go for it? Just write some forthright capital F Future vision and see where it takes you," and then of course, as soon as I started, I was just carried away. I was delighted with what was coming out.

‍

Samuel Arbesman:
That's amazing. When you write this massively futuristic, so it's 11,000 years in the future, in some ways, almost anything is possible within the broad sweep of the laws of physics, and so you can blend almost a fantastical world with a science fiction. How did you go about doing this? Was it an iterative sense of exploration? Did you do tons of world building? The book has very clear dates. There's maps, although the early version I had did not have the maps, and so I'm eagerly awaiting finally seeing the map.

‍

Robin Sloan:
Oh, just wait. Oh, just wait. I have the maps of beauty. It's absolute beauty.

‍

Samuel Arbesman:
Oh, like I have faded breath, I'm waiting for the maps. Yeah, how did you do that?

‍

Robin Sloan:
Yeah, it's funny. I say I like the genre of science fiction and that is true, but really, of course, my home genre is books with maps on the first page. That's where you know you got the good stuff.

‍

Samuel Arbesman:
That is a proper genre. I love that.

‍

Robin Sloan:
Yeah, that's a great question, and the answer is that there's the genre called hard science fiction. As you say, it can be wild, it can be wacky, but it does have to obey the laws of physics as we know them or understand them more or less right now. One of the bright lines in that might be that there's no such thing as a warp drive because that basically is magic. That's saying like, "And then we cast a magic spell and we were in Alpha Centauri." I like that discipline. I don't necessarily always like the coldness of some other of the work of that genre, but any creative work is sort of a game you're playing with yourself or kind of a Sudoku in some sense that you've created to challenge yourself as a writer. I always found that a really interesting and actually energizing constraint.
So this book is hard science fiction. I tried to make it something that could happen in our real physical world at the limits of time and imagination and energy and everything else as part of that more than I ever have before in any of my work. I took time and space very seriously. I actually went on NASA, and I downloaded their far future ephemeris predictions for the locations of the moon and all the planets, and they go out tens of thousands of years. This is not the default. You got to go digging. They're like, "Well, we don't really know who wants this, but here it is."
So on these days, and the year that the book opens by our calendar is the year 13777. So yeah, a little more than 11,000 years in the future, and there's these little asides. The characters, obviously, noticed the moon and the planets, bright Jupiter in the sky, and those are the real positions, at least as far as we can calculate them today.

‍

Samuel Arbesman:
I love that because you mentioned there were these conjunction of several planets, then I think there's a lunar eclipse mentioned. As I'm reading this I'm like, "I assume you've done the ... trying to figure this." I actually tried to do some digging to try to find these kinds of things and I was having some trouble, but I'm glad you did that for me. That's amazing.

‍

Robin Sloan:
It was really fun. I think especially, again, this I think can be applied generally to a lot of creative projects, in some sense, the weirder you're going to go, the wilder you want to push with your imagination, I actually think the more helpful it is to pin some things down to reality in that way. Obviously, I hope some people notice the trick and appreciate it the way it sounds like you did, but even if people don't, I think it helps with an almost subliminal sense of crunchy reality, and I love that.

‍

Samuel Arbesman:
Within world building, there's two views of making a rich world, either have a ton of stuff hidden underneath or give enough hints that it feels like there's that. It's sort of like a Potemkin village versus there's really something behind there, and both of them can work. Which one are you partial to?

‍

Robin Sloan:
Well, obviously, I think most people would answer somewhere in the middle, and I have to say I was ... That's my answer as you know.

‍

Samuel Arbesman:
I know.

‍

Robin Sloan:
Obviously, there was planning, and I've got notions about what exists in this world that I didn't burden the reader with all of that, but I have to say, I got a little radicalized on that question recently. It wasn't more than a couple years ago, I read these tremendous volumes. They're so weird. It's a series of books called The History of Middle Earth. They're not by JRR Tolkien or not totally. They're actually compiled and edited and presented by his son and longtime literary executor, Christopher Tolkien. What they do is they actually show you all these iterative drafts of all his work. The book I really focused on were part of this long series that focuses on the Lord of the Rings, which, of course, is the one that I know best. The reason it was such a actually powerful, profound study was that Tolkien, of course, is considered the great world builder.

‍

Samuel Arbesman:
It's like he creates a language and then, by the way, makes a story around it kind of thing.

‍

Robin Sloan:
Exactly. Language, script, mythology, whole deep history and it's like, "Oh, man, well, nobody else wants to read this. Okay. Fine. I'll write a story to get you interested." That, of course, is all true. It turns out though that the composition of The Lord of the Rings was as much improvisational as it was sketched out and planned. He did have this material, huge swaths of that story, ideas of how things fit together, the function of the freaking one ring, the iconic centerpiece of this saga, he didn't figure out what that thing even was or what it did until the fifth draft.

‍

Samuel Arbesman:
Oh, that's wild.

‍

Robin Sloan:
I like that it complicated that notion we often have of what the great world builders do and, I don't know, it also gave me ... I was like, "If Tolkien can figure out the one ring of Lord of the Rings deep into the process, then I can figure out stuff."

‍

Samuel Arbesman:
Even mere mortals can build these kind of world.

‍

Robin Sloan:
Yeah, exactly, an energizing thing to realize that we're all kind of doing the same thing. There's not these different species out there.

‍

Samuel Arbesman:
That's amazing. Oh, I love that. The beauty of the novel is the idea that world gets slowly revealed. The closest animal I can think of in terms of a hard science fiction or somewhat harder science fiction world that is also in this quest exploration of our future is the novel ring world where it's like because it's these people landing, being dropped on this weird giant artifact and having to figure out what is going on there. Are there other kinds of science fictiony, more science fictiony novels that maybe also have maps that were kind of inspirations for this kind of approach?

‍

Robin Sloan:
I think often to kind of the mode, the style of a writer like Isaac Asimov or Arthur C. Clarke or frankly even CS Lewis, who, of course, is better known for the Chronicles of Narnia, but he wrote these pretty wild science fiction stories and they're cool. I mean, they have the same level of imagination. What I liked about them is I guess I would call it a sense of adventurousness, and part of it is it's from a different time. Different styles of writing were cool back then, and all media was different and everything. It just is a different era without wanting to, I mean, it's not like I'm like, "Oh, science fiction has lost its way." No, I think it's great. I think we've got better science fiction we've ever had before, and I do wonder if maybe we've lost some of that adventure and just real straightforward sense of, as you say, "Let's go on a journey and let's really just have fun exploring and seeing the vistas unfold."
I thought of those often. I thought of CS Lewis, I thought of Asimov and Foundation, and they were great inspirations. Ursula K. Le Guin maybe was probably if I had to put one book or series at the apex of the aspirational influences, it would be Ursula K. Le Guin and her Chronicles of Earthsea, starting with a Wizard of Earthsea.

‍

Samuel Arbesman:
All of those books that you mentioned, they're all series, and my sense is Moonbound is intended to be a series or I believe a trilogy, and I was reading some of yourself, you have very strong opinions about the benefits and the power of series and trilogies and things like that. Maybe you can kind of share a little bit more of why they're so special in terms of the reader experience.

‍

Robin Sloan:
Yeah. Well, I think broadly, when I'm not writing and trying to be practical and get good words down on the page, in the screen, I like to just sort of think broadly about media and forms of media and how it all works together. The thing is that I feel like that's really criticism of media or the way we review a movie or TV show or book or whatever. It really focuses on the content, the words, the story, the characters, all that kind of stuff. Those are obviously very important. I think the kind of deployment of media, the way it unfolds in time is actually also really important for the feeling, for the public perception, for the way it fits into our lives. I think that's just a cool thing to think about.
That's all to say that the reason to write a series or a trilogy is not just like, "Oh, well, my story needed more pages, so I guess I need four books." It's to do something in time and create those beats of anticipation and excitement and the midnight release party of the book or thinking of TV shows and the huge difference between these bingeable drops where it's like, "Okay, 10 episodes, watch them tonight," or one every week or a movie and it sequels every Christmas or whatever it is. I just think that stuff is awesome to think about across all media. I want to be able to play in that sandbox.

‍

Samuel Arbesman:
Actually, I mean, when you were saying with the bingeable drop an entire season, it's great to watch, but you're right, you miss that kind of living with the characters throughout a year, whatever it is. So my daughter's really into fantasy and she's reading a lot of older fantasy series that are already finished, and then occasionally when she bumps up against ones that are still in process or the final book has just come out and she actually has to wait, it's a very weird thing because she's not used to it. I'm like, "No, no, you don't understand. This is the way it normally is. You wait years in between some of these things. There is anticipation."

‍

Robin Sloan:
Well, that's the dark side. I think about this often how it's ... As with all things, it's a balancing point. A little bit of anticipation of waiting can be so delicious and actually fun for everyone. The flip side of the coin is where you're like, "Oh, are they ever going to finish this one? I don't know. It seems like the delay between books has been increasing exponentially."

‍

Samuel Arbesman:
So is there an ideal? Is it like two years, five years? Because Lord of the Rings was also, it was very fast.

‍

Robin Sloan:
It was very fast, and Lord of the Rings, actually, again, I mean, I could talk about Lord of the Rings for an hour, the meta publication story of it. It actually, in a way, and he would insist, Tolkien would slap anybody down who called it a trilogy. It is one book. In his mind it's one book. To the earlier point, it just is big so it happened to have been published in multiple parts, and it did. It came out over the course of not even two years, its actual publication.
I think the perfect is a book a year. I think that's freaking fun for everyone, especially when it can become almost like a seasonal ritual. That can be difficult to achieve, obviously. I will say a special mention though, there is this iconic series published I think almost 10 years ago now by my same publisher, MCD, and this is from the great science fiction writer Jeff Vandermeer, and it was a trilogy of books, The Southern Reach Trilogy, and they came out over the course of 18 months. So it was almost like a hybrid of book publishing speed and TV show speed, and it was really cool. You'd finish one, and they were great and mysterious and you want to know what happened next and you only had to wait four months. I think that's pretty admirable.

‍

Samuel Arbesman:
Did he write them all at the same time?

‍

Robin Sloan:
He wrote them all at the same time. They were already-

‍

Samuel Arbesman:
Okay. So that was kind of the secret.

‍

Robin Sloan:
That's a heavy lift.

‍

Samuel Arbesman:
Have you written the remaining books all yet?

‍

Robin Sloan:
I know. Sam, no. Yeah, I got a good head of steam on the second one.

‍

Samuel Arbesman:
Okay. That's great.

‍

Robin Sloan:
Total vision for where I want to go and it's exciting. The other great pleasure of both quest stories in general and then especially I think quests that can unfold across a series of books is that widening aperture. I love stories in all. I'm talking books, movies, video games included, books that start really, really zoomed in one little town or a little village or whatever, and then slowly but very steadily and then maybe going faster and faster, they kind of pop out to these larger scales, and by the end of it, the whole world is somehow involved. I just think it's thrilling. You're seeing more all the time and you're seeing how the scene that you thought was everything is actually just a couple pixels in a larger picture. I love to do that, and that's what I'm going to try to do with this series.

‍

Samuel Arbesman:
Shifting gears sort of, I want to talk about AI. So there's AI as part of the story itself, but even before that, I mean, you've experimented a lot with using AI in writing and things like that. I get the sense, maybe I'm wrong, that you didn't actually use it that much for this novel, even though you had thought about it before and played with it. You were very early adopters to just a lot of these kinds things. What has been your journey with all these kinds of tools?

‍

Robin Sloan:
Yeah. Thank you for asking. It's a great question, and it's fun to talk about and fun to think about too. You're right. I did. I got started early like language model stuff before even GPT-2 was the cool, the hot, cool, state-of-the-art thing, and I got really, really excited about it. Specifically, I got excited about the potential to, as you say, use it as a sort of writing assistant. At that time, it was pretty wonky. So the flavor, I mean, it's wonky now too but in kind of different ways. Back then, its output would read sometimes very experimental poetry, which was cool. It would write things that you would never think of and often they'd be beautiful and surprising.

‍

Samuel Arbesman:
There was a clear complementarity.

‍

Robin Sloan:
Absolutely. I thought so. I thought so. However, after literally many years of building tools for myself and training custom models, again, at that time everything was a little more homebrewed DIY and that kind of stuff was really something you could do. I built a little machine learning rig, this little black box with a couple used GPUs stuffed inside, and I was training models overnight. It was an exciting and certainly interesting thing to tinker with. At the end of that arc for me, and this was even as the really hot rod stuff was coming online, the early days of GPT-3 and the sort of modern, really, really next level systems, I had basically concluded that, well, the truth is that writing with them was not actually fun, at least not in the way that writing is fun for me, and even as their capabilities took off exponentially, I found that they couldn't get what I was doing to the level of specificity that would actually be helpful.
At the same time, and this is where you're like, "Of course, things never work out the way you think they will and the benefits are never what you imagine they will be," at the same time, I had learned that thinking about these systems and tinkering, literally programming them and looking at the math and the code and trying to understand with some level of insight or accuracy what was going on in the guts of these things, that had been really, really rewarding. So ironically, what happened with this story is that I did not use a single lick of AI tools to produce it. You can put the label on the book, "No AI language in here." This is me writing it the old-fashioned way.

‍

Samuel Arbesman:
Totally organic.

‍

Robin Sloan:
Totally. Yeah, exactly, organic.

‍

Samuel Arbesman:
In a very different sort of way.

‍

Robin Sloan:
Pasture-raised fiction, but the ideas, these compelling questions and naughty, philosophical like whoa stuff, that's what went into the story. So there are AIs in the story. There's kind of characters who are part of that lineage. I mean, I don't want to overstate it or sell it too hard, but I think some of the depiction and discussion is on a new level in fiction, a new level in fiction, and it comes from just the experience, the real up-close personal experience of working with these machines and this material.

‍

Samuel Arbesman:
I want to get back to the actual story AI stuff, but first, I mean, you did a speedrun playing with AI in some ways. Do you think a lot of people are going to come to a similar sort of conclusion, and you were just the hipster equivalent when it comes to language models or whatever like, "Oh, I did it already, I'm over it," kind of thing or is there going to be some happy medium kind of thing?

‍

Robin Sloan:
That's a good question. I do, actually. It's funny you said that, the hipster thing. I actually sometimes worry that I have some of that in my head. I'm like, "Oh, I was totally into that band back in 2017," and I'm like, "That's actually not the healthiest thing. Maybe I should just give it a rest." I think people are going to use it more and more, and you're like, "What people? For what?" Let's just restrict our conversation here to people who want to write, and we could say nonfiction and fiction for publication in the normal marketplace of trade books in the US and around the world because, obviously, there's a whole vast worlds of other kinds of writing that operate outside the market or in different markets or whatever.
So will people use these tools in that domain? For sure. I'm quite certain that people are right now, and I think they're not advertising it. They're just like, "Yeah, this is just something I'm going to do behind the scenes and it helps me flesh out a chapter and get started where I was stuck," and that's great. Anything, boy, writers will give anything for an unsticker, something to just get you off over the heavy lift of blank page to pile of garbage. It turns out having a pile of garbage is awesome because then-

‍

Samuel Arbesman:
Right. Then you can edit it. You have something to work with.

‍

Robin Sloan:
Exactly. It's great. Yeah, that's all you hope for. You were like, "Please, please, I just need a pile of garbage." Now, what that means? I guess, I don't know, and I don't know, I'd have to think more about what I think writing is for. It makes me think of certain kinds of genre books, which I enjoy. This is not really a criticism, it's just an observation. Certain kind of genre books are very bag of chips, shall we say? It's like, "Okay. I know what you're looking for, and I'm going to give it to you, and you're going to like it, and then go on to the next one." Talk about writing a book every year or two books every year. That's usually the pace of writers like this.
It seems to me that AI could fit very neatly into systems like that and expectations like that. When it comes to somebody, and this is not just in the world of literary fiction, I think this is anybody in any genre who just is trying to do something a little bit different when you're trying to, I mean, really makes something that is new, that is new in novel. The word novel is so funny because it's novel. That's the point. It's like, "This is a truly new thing and there's at least some elements of it." Even if some are familiar and comfortable, there are others that are going to be challenging, and you'll be like, "Wow, I never had that thought in my brain before." If that's your aspiration as a writer, I'm not sure if AI language models are going to be that useful in part because the way these things work is it's a probability distribution.

‍

Samuel Arbesman:
Right, and they're also interpolating between that which has come before it.

‍

Robin Sloan:
Exactly. Exactly. So it's in a way, it's almost like a challenge to artists of any media. It's like, "Yeah, if you're doing something that could be very comfortably and ably produced by an AI model, well, maybe that tells you something about what you're doing," and you actually want to head out to the edge of the distribution and find what's waiting beyond it.

‍

Samuel Arbesman:
Creativity is fundamentally recombination, but if it's recombination of the most used things, then it doesn't feel that creative, and so you want to find the things that are less used.

‍

Robin Sloan:
That's right. That's right. I mean, I take that to heart, but Moonbound, this book, is packed with allusion and homage and winks, all sorts of winks. Most people won't get all of them. Only me. I'm-

‍

Samuel Arbesman:
I enjoyed the Mrs. Frisby and the Rats of NIMH one. I really enjoyed that.

‍

Robin Sloan:
Yeah, exactly, exactly. This is not really a spoiler. The first creature that our protagonist meets in the wider world after we leave sort of the cradle of the early chapters, the first creature that he meets in this strange new world is a beaver, a talking beaver, and of course, in The Lion, The Witch and The Wardrobe, the great launch of the Chronicles of Narnia, the kids who go to Narnia, the first creatures they meet are a couple of talking beavers, and you're like ... and that's just great. I mean, that kind of resonance and frankly repetition, I mean, that's what makes literature go. That's the scaffolding you're building on and in and with and everything else, but as you say, the point is to maybe put together a very idiosyncratic set of those references and allusions and homage rather than just the five dorkiest, most obvious ones from all of human output.

‍

Samuel Arbesman:
Right or subverting some of these kinds of ... You want to subvert the expectations while at the same time having certain things that kind of makes sense.

‍

Robin Sloan:
Exactly.

‍

Samuel Arbesman:
I don't know how much detail you want to share in terms of how AI is thought about in this book, and I would say even outside of the book. There's the sense of, okay, there's the real world and the world of physics and nature and things like that, and then there's the world of stories, which are the myths and plot points and things that we have in the world, and language models, they're not learning from the real world, they're learning from text, which means they're learning story world. They're not learning real world, and that's a fundamentally unfortunate thing if you're trying to train robots, maybe, for example, but it's really interesting, and you see this with the, what is it, the Waluigi effect that people have talked about. Are you familiar with this idea?

‍

Robin Sloan:
I don't know. I don't think I know the Waluigi. Of course, I know Waluigi, a classic part of the cultural canon.

‍

Samuel Arbesman:
Yeah. So in the same way though it's like Mario and Wario while Waluigi is this agent of chaos version of Luigi. I think the idea behind it is that when you train these AIs to be very rule-following, very easily you can tip them over into becoming Waluigis, and this is the way you jailbreak them or do all these kind of weird things because they've imbibed all of these ideas of our stories and narrative tropes and things like that.
In the same way, there's another idea that people talk about like semiotic physics. There's physics of the real world, and then there's the semiotic physics of like, okay, when you choose a random number, most people choose seven, that has nothing to do with actual randomness. When you ask AIs to choose random numbers, they're going to choose a certain type of randomness that has nothing to do with actual randomness. Maybe you can talk about a vague and non-spoilery ways or how do you think about AIs and how they embody or live within this world of story and narrative tropes or whatever it is.

‍

Robin Sloan:
No, absolutely. Again, only at our peril as this stuff advances and takes on more tasks and gets more deeply embedded in society and the economy, only at our peril will we forget or ignore the training data and the stuff that goes into it. It's stuff with history. That's not true for everything, everything they're trained on. More and more, I think as close watchers of the AI world right now understand that some of these models are being trained on synthetic data. It's being produced by other computer programs or even so, so freaking weird by other language models, by other AI systems just like, "I need a million examples of math problems," and it spits them out, and then you say, "Okay. Eat this, next level language model." So that's one thing, obviously.

‍

Samuel Arbesman:
It's like the AI equivalent of chewing your cud or whatever. I don't know what they call it.

‍

Robin Sloan:
Oh, man. Again, you got to be careful with the metaphors. I think of Gavage where they're like force-feeding these animals, fattening them up. It's gross. You're like, "I don't know." You're like, "Oh, robot uprising. What are they going to do? What are they going to feed us?" Underneath all that stuff, and really at the root of the discipline too, this is also part of the social human part of AI, not just the technical machine part. There are books. There are so many books, and there's documents. A document is not just actually a stream of characters. Documents have shape, they have drama, they anticipate questions, they do. They represent the shape of a certain kind of human thought.
One of the things that is most totally just mesmerizing and is amazing to think about to me is the speculation. I've just heard this from some AI researchers and practitioners that a big part of the leap forward we saw with ChatGPT and GPT-3 in that era was that they started adding huge amounts of code to their training Corpora. What is code? I mean, among other things, it's incredibly structured. It's kind of like never wrong. If this is a code that compiles, if it's good code, then by studying code, you're not just learning how to write Python or JavaScript. You're kind of learning about a structured way of thinking.
So the theory goes that these models were then able to learn patterns and ways of associating things that were super valuable back in the realm of normal language. Again, it's there in the training data, and there's a sense in which what language models are doing is they're almost like you think about the early days of electricity. When people thought electricity was also the soul, and so they'd zap a dead animal and it would shiver and they were like, "Look, it's back. It's back to life."
I sometimes think of language models as just giving dead text of all kinds the ability to rise from the table and start walking around and doing spooky things. If that's the case, then the story is right themselves, and as you know, having read Moonbound, there's some of that in this book, some unintended consequences of some of the material that was deep, deep, deep in these-

‍

Samuel Arbesman:
The structured code stuff and kind of how that allows that next ... That's something that I hadn't thought about and it's like-

‍

Robin Sloan:
I don't know if anybody knows for sure if that's the case, but I've heard the speculation.

‍

Samuel Arbesman:
It's a delightful speculation. It's like, "Yeah, we now understand how to write things with like if then else," because of the structure of code. Oh, that's fascinating.

‍

Robin Sloan:
Again, though, because the loop, the loop is complete because what is code? Code is a halfway point, and I'm talking about the higher level languages, again, Python, JavaScript. It's a halfway point between our brains and between our brains and computer brains. It's not computer ease. These languages were designed for humans to be able to be like, "Okay. All right. I can do this. I can hold this in my head."

‍

Samuel Arbesman:
I believe you're a fan of the Long Now Foundation. I'm actually affiliated in a very low level way with them. That's the argument of the clock of the Long Now, for example, which is in the process of building this clock designed to last for 10,000 years, you'll have thought about geopolitics and language evolution and material science and all these different kinds of things. I guess I struggle with to what degree, and I'm a big proponent of thinking very long term, expanding our aperture of how we both spatially as well as temporally and understanding these the vast long term because right now, I'm focused on quarterly returns or whatever, those shorter term kinds of things as a portfolio strategy. How much of our time should be spent thinking on the very long term and how long?

‍

Robin Sloan:
That's a good question.

‍

Samuel Arbesman:
I love thinking about this, but I'm just trying to struggle with how much.

‍

Robin Sloan:
That's a good question. That's a good question because, really, it's obviously not practical to live in a society in which everyone is sort of like, "Hmm, in 12,000 years ..." You're like, "Cool, man. Can you finish making my bagel sandwich? Really, I got to go."

‍

Samuel Arbesman:
Exactly.

‍

Robin Sloan:
That's a great question. I think the answer is in a portfolio, a personal or social portfolio, if you gave it a percentage, I mean, it's certainly not more than 10% and probably less than that. I guess I would argue that personally and socially right now, it's probably at like 0.04% or something like that. I think it's a huge, huge missing piece. Again, it's not like you have to have a plan or you could have a plan for periods of time that long. It really is just entertaining the question.

‍

Samuel Arbesman:
How do we as a society begin to just grapple with these long stretches of time and civilizational kinds of temporal range?

‍

Robin Sloan:
I think it's hugely challenging. It's hugely challenging, and maybe even in some sense impossible. For me, actually, I get so jazzed by that. I just think it's like what a cool thing to try. I often, for years now, I've conceived of imagination as almost like a muscle, like a physical muscle, and in the same way that if you're just a wimpy, regular person who doesn't ever work out or lift weights, if you go and try to lift a sack of beans, whatever, a rock that's 300 pounds, you're not going to even budge it. You could train yourself to lift it.
I mean, people do this. They become these incredibly strong athletes. I really believe that imagination, certainly the kind of science fictional imagination that I'm invested in myself, is the same way. I think you develop it through practice, by pushing yourself further and further to dream and elucidate and come up with details and make things plausible, all that kind of work. So if you start at nothing and then maybe you get to the point where you can make a really good 22nd century, I think you can keep going. I think you can in some sort of quasi objective sense get better at it. Then maybe if you're lucky, you'll reach the greats. You'll be like the great Ian M. Banks, who I think was one of the ... He's the Arnold Schwarzenegger of future imagination. He wrote this linked series of books about a far future civilization called The Culture. It's thrilling to read both because the stories are cool and weird, but also because there's just a sense of this Scottish guy doing this saying like, "All right. Let's go." I mean, his stories unfold across. You talk about the scale of space, time, energy, everything. You can't even believe it, and it's cool.

‍

Samuel Arbesman:
I really enjoy The Culture novels. I have to say, I think my favorite piece of culture writing is actually his essay about the world building.

‍

Robin Sloan:
Absolutely.

‍

Samuel Arbesman:
It's so amazing and you're like, "Oh."

‍

Robin Sloan:
It's thrilling. Yeah, it's really thrilling.

‍

Samuel Arbesman:
There's just this amazing scope and scale.

‍

Robin Sloan:
Yeah, this came from a human mind.

‍

Samuel Arbesman:
What would be your exercise regimen for the imagination?

‍

Robin Sloan:
Well, this is true for fiction writing in general. There's no substitute for really doing it. You'd have to write some stories. They could be pretty short. The imagination part of it could get expressed very concisely. It certainly doesn't have to be a novel. I think of a terrific, really, really wonderful thinker and actually practitioner named Jack Clark who works at Anthropic, and he writes a newsletter. It comes out every Monday. It's called Import AI, and it's sort of a research roundup. Really great, super useful. I mean, anyone who's interested in this field ought to subscribe and follow along, but each one ends, concludes with a little micro science fiction story.
I mean, some of them are not more than three paragraphs long, so they're just little flash fictions, but there is a demonstration of someone who is a weekly workout regimen, and you can see the ideas getting deeper and weirder. Honestly, even if a person wasn't interested in writing fiction, if they're like, "I just want to have a really killer imagination so I can, I don't know, make better investments or dream up different products," they just write a really, really short story a couple times a week, just sit down and think of something, and I guarantee the ideas will start getting better and better and better.

‍

Samuel Arbesman:
Changing gears quite a bit, and maybe it's not changing gears, and so one of the other things you're involved in is producing olive oil. Is that just another thing you do? Is that complimentary? How do you build that into the rest of your working life?

‍

Robin Sloan:
That's a good question. I should make up some story where it's super, I'm like, "Ah, yes. So you see ..."

‍

Samuel Arbesman:
You don't have to.

‍

Robin Sloan:
No, I should, I should because so many people ask and it would be fun to be like, "I have learned through the olive the true nature of storytelling." That's not the case. That's not the case. It is complimentary in a couple of respects. One is, and this is the biggest one, and really, I recommend this to anybody who spends most of their time writing in any medium, you got to do something that gets you up away from the computer, away from the desk, out of the office, just out in the world. So obviously, all the olive stuff is standing and driving and stalking through groves and running huge machines. I operate this enormous olive mill. I mean, it almost is kind of a sci-fi machine by itself.
I sometimes squint. I'm like, "This could be the engine room of a great starship," and I use my muscles. I get tired. At the end of the day of doing that, I go to sleep because my body is completely wrung out, and that does not happen after a day of writing or doing computer stuff. I go to sleep and I'm kind of like, "Well, I guess it's time to end this day." So that's awesome. That's just super, I have found that so healthy and energizing everything else.
The other thing though, and this is connected to imagination and where these ideas really come from, is there are a few parts of that process, mostly having to do with driving. I do a lot of driving, just crisscrossing the state of California, and I just daydream. Not all of my habits as a writer are great or impeccable, but one that is really, really good and I'm proud of is I'm a great and very diligent note taker, and that can either be something I type in or a voice memo I record and transcribe later. Indirectly, Moonbound exists because I had a lot of this time to just drive and think and just let my mind roam and wander. I recorded some of those thoughts and they're on the pages of the book you read. I'm grateful for that.

Samuel Arbesman:
Related to note-taking, do you use any AI and the sifting through or reconstituting or recombining the notes or is it much more just going through it, finding things, magpie-like? I don't know what the right way to think about is.

‍

Robin Sloan:
Yeah, it's totally the latter for now, and it is. I mean, actually, I hardly could have more fun than to spend an afternoon just marinating in my own notes, just to open them up and kind of random access because by definition, a big, big, big collection of notes is like reading the most specifically awesome blog for you because by definition, everything is something you thought was really cool or interesting at some point, so you're like, "Oh, yeah. Oh, molly." Not everything is a winner or makes its way into some finished piece of work, but it's great and the way things can connect in unexpected ways.
I mean, again, I'll answer the AI part in a second, but just to underscore, I mean, I think a lot of writers are like this. I'm a freak. I will hear a name in the grocery store and be like, "Okay. Writing that down," or I'll see a sign that says something interesting or, obviously, when I'm reading other books, I'm constantly taking things down or capturing whole passages, and I think it's important to do that basically to train yourself to notice when language or ideas gleam when they have that.
Yeah, magpie-like, when you're like, "Ooh, shiny. I don't know why, but it's shiny and I want it," and to keep those things and then have them in a space where you can just explore them again is, I would never have written a thing without that. So I don't use AI tools, but I'm not opposed to it, and it's interesting. The truth is I haven't really considered what I might build for myself or what I might experiment with, but I think it's interesting to think about.
Personally, this is my opinion, I could be wrong, but I think the best use of these language models as they exist right now is as a helpful reader to basically be like, "Read this and do something with it," and that's a little different from writing stuff from nothing. I think the mode of maybe you have a bunch of files and it's unstructured stuff and you're like, "I want this in a table," or, "I want you to extract a certain kind of information from this and then give it to me in a JSON object," or something like that. I think that's such a sweet spot for what they can do and what they can offer. So that sounds like something really interesting you could do with a bunch of weird notes, so I'll think about it.

‍

Samuel Arbesman:
I guess, I don't mean to keep on mentioning AI again, but you used it a little bit for your audiobook for Sourdough. Are there any kind of, AI-related or not, are there any fun treats for the audiobook of Moonbound?

‍

Robin Sloan:
Well, there are the audiobook, but you write books just as this is a little slightly more general comment, but you write books and you just start to understand the real market out there, what people are looking for. I've listened to an audiobook here and there, but I'm not a huge listener. Boy, people love audiobooks.

‍

Samuel Arbesman:
It's huge.

‍

Robin Sloan:
I mean, it is the part of the publishing market, at least in the US, that is the rest of the parts are like doing okay. They're sort of holding steady print in digital. Audiobooks continue to grow, grow, grow, grow, grow. The appetite is just getting bigger, and the upshot of that is that they do an awesome job with the productions. I mean, they're almost like quasi-theatrical productions with music, and obviously, some audiobooks actually have a full cast. My audiobook is just performed by a single narrator, but she did an amazing job. So no AI, but the absolute best that humanity has to offer.

‍

Samuel Arbesman:
Well, it wasn't AI speaking. The previous one was like AI-generated music, so kind of create this other worldy like the other worldness kind of thing.

‍

Robin Sloan:
That was interesting, and you run into this problem, actually. Actually, so it's not AI, but it was a similar problem in both cases. Books, of course, you describe things and you sometimes can also put images, little images on the page. To go back again to Tolkien, I love his little bits of script, Elvish script on the page so much. I was like, "I got to have that." So I invented a fictional script that's contextual and makes sense in my world and in sort of this story, and it appears there on the page in the printed book.
For the audiobook, what are you going to do? You have to represent that somehow. You have to have a way of sort of vocalizing it or, I don't know, making it sonic. So we came up with some solutions for that, and likewise in Sourdough, a huge part of the story was this sort of mysterious music and it came up again and again and again.
At a certain point we realized, we were like, "It's pretty weird for a person to be listening to this and they just keep hearing the music describe." They're like, "Well, okay, why don't you play some?" So I did. I used a little homebrew. I commissioned a really terrific programmer to help me because it was just beyond my grasp, and together we worked on this little homebrew like music model years ago and produced just the weirdest music you have ever heard, and it's great. It's a really cool part of that audiobook.
The goal always is ... I think it's interesting. You want to, in this kind of stuff, you want to obviously to have that plausibility, that suspension of disbelief and fit contextually into the world. You do also just want it to be super cool. This is a completely base objective of any kind of creative work. You're like, "Yeah, yeah, yeah, okay. It makes sense. It fits with the story. Also, it's awesome." So I tried to make something that I thought looked and just seemed cool and, I don't know, made you feel like you're reading the right book.

Samuel Arbesman:
That might be a great place to end. This is fantastic.

‍

Robin Sloan:
So fun.

‍

Samuel Arbesman:
Thank you, Robin Sloan, author of the new book, Moonbound. Yeah, this is amazing. Thank you. This is fantastic.

‍

Robin Sloan:
Thank you. This is wonderful, Sam. This is wonderful conversation and I really appreciate it.

‍