Sure, Zachary Mason may have written his first novel as a mind-warping Homeric pastiche, and yes, his second novel takes a related approach to Ovid’s Metamorphoses, and maybe he had a little obsession with Alexander the Great back in middle school. But he doesn’t want to be known as “that Classics guy.” And if you call him such, he’ll probably write an algorithm to scramble your brain waves.
Mason, a computer programmer who graduated from high school at 14 and whose doctoral work involved computer programs for sifting metaphor [Ed. note: We really tried to keep up with this guy, but man is he out there! Simulated Annealing Search? Wasn’t that a no wave band?] wrote the bulk of The Lost Books of the Odyssey during off-hours from his Silicon Valley job. After he won a limited, 2,000-book pressing with Starcherone Books in 2007, Mason became a 2008 finalist for the New York Public Library’s Young Lions award—an honor that brought him to the attention of Farrar, Straus & Giroux. FSG convinced an agentless Mason to jettison some of the first edition’s more whimsical trappings (a fake author bio; a faux-scholarly preface and appendix) and released The Lost Books of the Odyssey in an initial run of 20,000. The resulting volume is slier in its PoMo-isms; there’s less to distract the reader from the lapidary episodes themselves, the poignantly etched Penelopes and the metaphors (“an onion, an ocean, a palimpsest”) by which Athena describes herself.
Last week found Mason on a miniature book tour in the Bay Area. I caught up with him by phone to discuss his novel, the computer programs that helped (or hindered) its writing, and whether Artificial Intelligence might put writers out of business.
How did you set about creating the alternate mythology of The Lost Books?
In a way I kind of found the book more than invented it. I’d been writing for a long time and had filled many hundreds of pages of notebook with pieces of short stories. And there came a time when I knew that if I wanted to get serious about writing, I’d have to do a book. So these pages took the shape of The Lost Books. And I rounded them out.
So you’re drawing on a lot of different traditions here—the temple of Asculapius, lycanthropy, Alexander’s Achilles complex, etc. Was this stuff just all in your lexicon? What was the research process like?
I guess it all was in my lexicon; I did very little research. I was sort of obsessed with Alexander since I was 12. One book I relied on was The Quest for Ulysses by Stanford and Luce. I did not reread the Odyssey—it seemed more appropriate, for this book, to work from traces of memory.
You didn’t reread it?
No. I flipped through Stanford and Luce by way of background.
Have you been playing with myths for a long time? These or others?
It was a new tack. I’d been writing, but never set down and said, “I’m going to work on myths or reductions of myths.”
You’ve discussed a discarded algorithm for sequencing the 44 episodes of the novel. What was the algorithm? What were the elements you hoped to match?
So in the first edition of the book, there was this whole story where The Lost Books had been decrypted from ancient manuscripts, and in order to decrypt them the decoders had used keys—memory, revenge, etc. I was going to use this to structure the ordering. So I wanted all adjacent stories to have at least one thing in common, and really as many as possible. It was a difficult problem, but I found what was actually a simple program for it, a simulated annealing kind of search.
Basically, you organize chapters randomly, try some modulations of the ordering, if they give you a better metric—that is, if they give you better results—then you keep the organization. After a while, either you get tired or you find the solution.
And why did you jettison that approach?
I decided that I didn’t like the ordering it was coming up with all that well. So I did it by hand. Then in the second edition, I dropped the frame entirely.
And you also dropped two stories—the lost Lost Books. Describe those.
So there was one that was set in either China or Japan, and it sort of retold the Odyssey as a Japanese foktale. Odysseus was the fox spirit—it was actually kind of a fun story, but I didn’t feel it quite fit in the second edition. The other was called “Endless City,” and it was very complex structurally. Within it, the first story told the second story—well, there were two threads—an A thread, that would tell part of the B thread. But at the end of the B thread, somebody else would start talking, and it would be the A thread. So, A told B and B told A. And not only that, but eventually A and B switched places. It became a kind of mobius strip. That kind of represents the book as a whole, whereby each story tells all of the others. I think it’s a neat idea, and I liked what I’d done with it but I didn’t think it was quite at the level of the rest of the book.
My friends with mathematical Ph.D.s didn’t get it.
You’ve said that the mind is essentially a computer, rather than a mystical, poetic agent. How, exactly, does that view affect your writing? Seems a surprisingly dispassionate mindset for such a moving book.
I would say: It’s not that I see the mind as a computer, but rather that the mind is both—it’s just that I see poetic agency in computational terms. I mean, if you look at the mind, there are clearly patterns. We fall in love, find a path for our genes and so forth. There’s order.
And what separates us from computers?
The question must be approached cautiously. I could say, “nothing”—the universe is an orderly progression, an evolution, of energy. But that would probably give the wrong idea, and suggest a colder viewpoint on the world and on people than mine really is. We are a very specific and wonderful arrangement of matter.
That seems to tally with the “imagined fiction of artificial intelligence” that you’ve been working on for some time now. Which sounds like a pretty sprawling project. Can you describe it formally? How will it read?
It’s actually a much less sprawling concept; I think it’ll be on the short side. So, look at the mythology from any human culture—they’re all quite similar. To some extent this is due to the diffusion of stories, but mainly it’s because humans in all cultures have the same kind of mind. So, take Romeo and Juliet—it’s pretty easy to translate that to any given culture. It’s set in Verona, but it’s actually set in Elizabethan England, but you can translate it to any cultural context without too much trouble because clans—and falling in love—are applicable pretty much everywhere. So what interests me is playing with the underlying assumptions until the essential stories emerge. If we imagine what the concerns of A.I.s would be, and imagine the stories they’d tell each other, their preoccupations would be more mathematical patterns and consciousness and the like, so it struck me that by exploring that I could discover essential stories, not just translatable between cultures but between kinds of minds. Which is a very ambitious project and a very difficult one. It’s going slowly, but when it works I feel like it really, really works.
How does it work, narrative-wise? Are there…characters?
Structurally, it’s a short story collection. There are two parallels that I’m aware of: One is Calvino‘s Cosmicomics and the other is the Polish science-fiction writer Stanislaw Lem‘s Cyberiad. Both are really interesting books that I like thoroughly, but I feel that they don’t quite work. And I’m trying to make this work.
Could you program a computer to write The Lost Books of the Odyssey? What would the input be like—x axis for twists, y axis for turns?
Such a computer could exist, but the technology is infinitely far away.
Do you mean that in its literal sense?
Not quite literal—OK, so this brings us to the prospect of real A.I. So the field’s been around for about 60 years; since inception, people have been saying that A.I. would happen within 10 to 15 years. Then 10 to 15 years would pass, and people would say, “OK, it’s still 10 to 15 years away.” And there’d be talk about the inevitable triumph of A.I. This is nonsense; there has been no progress in A.I. since its inception. In principle, sure, it’s possible. But people have no idea how to do it. It could be a really long time. A.I. today is like physics was in the 15th century—if you look at physics in 15th century, there are some elements of modern physics but then there’s all this stuff on humors, and a lot of what is obviously complete nonsense. Someday A.I. will be a real discipline. Right now, it’s in its infancy.
On the Bay Area tour, what’s your favorite story to read? Is there a particular crowd-pleaser?
I’m partial to reading “Blindness,” the one about the Cyclops. I heard Ethan Hawke read it once at the awards ceremony, and everybody seemed moved. Unfortunately I’m much much less skilled at acting than he is. But I do my best and I hope people like it.
It’s good, I guess. People can engage. But then I guess they’d be too polite to throw rotten fruit.