What do you call a thriving marketplace of robots buying nonexistent books from other robots for millions of dollars?
The online retailer we know and mostly love can be something of a wild west, especially on the frontiers of its business. Consider the experience of Carlos Bueno, a software engineer at Facebook who, late last year, wrote and self-published a children’s book about computer science called Lauren Ipsum.
The book is inspired by Bueno’s self-education in programming as a young man and his belief that everyone in our increasingly computerized society should understand how to program.
“If someone could read, but they couldn’t write, in our society today, it would be kind of weird,” Bueno says. “But we totally accept the idea of someone who can use a computer but can’t program it. There are amazing things you can do when you don’t tell a child it’s too hard.”
Bueno raised money with Kickstarter to publish his book through Amazon’s self-publishing service, making his book available in a variety of electronic formats and also as a print-on-demand book—each time a physical copy is purchased, it’s printed specifically for that order. Bueno set the price of the book at $14.95 and has sold about 1,000 copies.
But in the last few weeks, Bueno has seen his book become the center of a strange phenomenon on Amazon: the bot market. A reseller in Amazon’s used books section was offering the book for $55—even though the book was available for forty dollars less on the same website. Then another one appeared, selling for $14.94—lower than the retail price. Another was for sale for $12.50. The only way these resellers could profit would be through excessive shipping and handling charges.
Even stranger, these resellers are offering “Very Good” or “Like New” used copies of a book that is printed on demand—that is, they’re offering used copies of books that probably don’t even exist.
If you haven’t guessed yet, these resellers aren’t people at all, but bots—software created by someone to try and game Amazon’s marketplace, making a profit by exploiting loopholes and engaging in bidding wars. But when they go wrong, they end up doing crazy things; a biologist at UC-Berkeley documented a bidding fight between two bots over a “classic work in developmental biology” called The Making of a Fly. By the time they were done, copies of the book were advertised for $2,198,177.95
“I’m a programmer, and I’m mystified,” Bueno says. “We’re just going to have to watch and see what these bots do now.”
Amazon, he points out, is getting in on the bot action. In response to the apparently illogical prices offered by the robotic resellers, it cut prices, too, now selling the book for $10.76. That surprised Bueno, who thought he had set the final price, but Amazon’s price matching algorithm works automatically. Bueno was pleasantly surprised to find that the cut doesn’t come out of his royalties—Amazon is eating the cost of matching the prices of these fake-book purveying bots. Bueno estimates that Amazon is losing $4.19 of profit each time it sells the book at the discounted price.
Bueno and his friends have speculated that perhaps the point of these bots isn’t actually to game Amazon’s markets, but to use them for other purposes, perhaps laundering money from stolen credit cards by purchasing goods from fellow conspirators at inflated prices. Some of the resellers are based internationally; perhaps they are gaming exchange rates.
Some of these bots even make print-on-demand books, copying Wikipedia articles into books that they print on demand and sell to, well, whoever would be silly enough to buy something like that. We’ve asked Amazon what they think of the robots in their store and what, if anything they’ll do about it. They're not the only online retailer facing this challenge—Barnes & Noble, among others, appears to have its own entrepreneurial algorithms.
I ask Bueno if these programs-gone-wild offer any lessons, perhaps for his next book.
“The big lesson, no matter how clever you are, if you don’t have the right data, you’re just as helpless as if you didn’t know how to program,” he says. “As a general thing, I think large companies like Google and Facebook, they’ve figured this out, the real secret is data, and not really the algorithms.”