Home » Advertising-Promotion-Marketing, Discovery, Non-US » The End of the Human Publisher? Introducing the First Novel to Be Chosen by an Algorithm

The End of the Human Publisher? Introducing the First Novel to Be Chosen by an Algorithm

3 May 2016

From Flavorwire:

[Y]esterday, the Berlin-based company Inkitt announced a partnership with Tor Books that will bring about the first ever book chosen by predictive data.

The novel chosen by Inkitt’s “artificially intelligent” algorithm is Erin Swan’s Bright Star, a young adult fiction submitted to the publisher through a writing contest called “Hidden Gems.” Part of a multi-book “Sky Rider” series, it tells the story of the “fantasyland” Paerolia, “where war and conflict has created strong divides,” and where a a rebel leader named Kael helps a slave named Andra “discover the strength that has always been within her” and “fight to win back what Fate kept beyond her reach” — namely a dragon “that should have been her own.” Bright Star is expected to be released in 2017.

Inkitt, the company responsible for discovering the novel, is an online writing platform where “budding authors” share their work with “inquisitive readers.” It relies on an “artificially intelligent” algorithm to bring the two together with the purpose of uncovering “blockbuster books.” This description calls up a number of questions. Did Inkitt invent artificial intelligence? Should we be surprised that the first artificially intelligent being prefers genre fiction? If you put aside Inkitt’s overheated claims about artificial intelligence, you’ll find a publisher that just wants to do the write thing: “Inkitt’s goal is to remove the middle person so that a blockbuster book is never rejected by a publishing house again.”

. . . .

“This book deal sends a clear signal to the publishing industry that predictive data analysis is the way of the future,” says Albazaz. “Inkitt is at the forefront of the movement to use predictive data in publishing, and this deal shows that our business model works.”

. . . .

Still, it’s hard to say whether Inkitt’s first major deal is a function of its algorithm or its status as a thriving online world, which “stretches from the US to Australia.” By its own account, Inkitt has a community of half a million loyal readers. And its business plan – now seeing its first moments of success — is to bring the “future bestsellers” validated by this community to publishers, like Tor. It also plans to independently publish ebooks of selected novels from its own platform, “with supporting in-house marketing campaigns.”

Link to the rest at Flavorwire and thanks to Dave for the tip.

“Inkitt’s goal is to remove the middle person so that a blockbuster book is never rejected by a publishing house again.”

For PG, Tor is a classic example of a “middle person” which stands between a book and its readers. Is a literary agent a middle person? Or an acquiring editor employed by Tor? If Inkitt is going to “independently publish ebooks,” it’s a middle person as well.

Suspecting that the awkward “middle person” terminology might be a poor translation, PG did some brief Google research on the German term for middleman (he knows it’s politically incorrect, but nothing came up for middleperson) and found Vermittler,  Mittelsmann and  Zwischenhändler. Similar terms appear to be used for the English word, intermediary.

PG also discovered that a person who would be called a real estate agent in the US is a  Grundstücksmakler.

In preparing this comment, PG has approximately tripled his knowledge of the German language.

Advertising-Promotion-Marketing, Discovery, Non-US

20 Comments to “The End of the Human Publisher? Introducing the First Novel to Be Chosen by an Algorithm”

  1. In other news, these newfangled “talkies” have ruined motion pictures

  2. Ah, but who set the ‘requirement/standards’ for the algorithm? I’ll bet my settings would pick a different book than the next person in line …

    • They could crowd source the criteria. It worked so well for Microsoft’s AI project.

      • Smart Debut Author

        Funny. 😀

        Microsoft should sell the Tay AI to the big traditional publishing houses as an “author-platform-building bot” for publishers to clone, name after all their authors, and turn loose on social media. 😉

  3. So if this book was “discovered” on a Wattpad-like platform, presumably real human readers liked it? They’re just taking credit for the algorithm matching up the book with readers who would accurately rate it? I think?
    Clearly, no computer algorithm could pick out a bestseller by itself. There’s a lot of hyperbole in their claim.

  4. Inkitt is full of crap.

    They say that they want to cut out middlemen, but their whole goal is to sell books to publishers. They say that there are no experts, but then they use an expert algorithm to choose which books to publish.

  5. Hah! I thought Tor was just a teensiest bit more ethical than the others … someone didn’t do their homework, or maybe they did.

  6. So, so, so many quotation marks. That second paragraph really left a bad taste in my mouth.

  7. PG, your research made me smile. I love my native language.

    Inkitt sounds as if it wants to be an agent with a computer assisted decision-making system, rather than relying on experience and an understanding of stories and the marketplace (not that this is helping literary agents at the moment).

    Having said all that, the “Sky-Rider” series sounds like rather average, generic fantasy. I wish Erin Swan all the best for it, but it sure isn’t having me wait for the release with bated breath.

    • Felix J. Torres

      That was my reaction.
      I don’t know if the books are any good but the blurb is so generic that if I saw it on a Kindle listing I would move on in a blink. Totally empty.

      Maybe they should be working on a blurb-writing algorithm. Though I suspect that is something beyond even WATSON.

  8. Well, I guess at least TOR found a way to decrease its infamously long/slow response time —

    By outsourcing the process to a social algorithm engine?

    And I’m with Hannah. What is written there of Sky Rider sounds like somebody threw a whole bunch of fantasy words into a separate randomizer and then some other algorithm or other spat out a word-salad plot.

  9. Buy my book, a computer says it’s a bestseller! :::rolls around in piles of filthy lucre:::

    The only real way to get rid of the middleman is to self-publish, and even then the retailer is a step between the author and reader.

  10. Typical clickbait headline. Every trivial “advance” in technology brings up the same dreary topic. Humans will still be around — unfortunately — when the technologies are no longer functional. Don’t worry, kiddies, you can still write and publish your totally irrelevant books when the oceans have inundated the coasts, California has slid into the ocean, the entire southwest of the US has burned up in drought, and the last coral reef has dissolved into the acid sea. Lots to write and argue about. Keep it going.

  11. Wait a minute. Inkitt are a real company, but what about this and this?

    I told them to naff off when they tried to convince me to join their competition on Twitter last week. I thought it was a just another con man in the publishing industry. Oh, wait…

  12. The article doesn’t tell us enough to really understand the process. If I had to guess, it’s a slush pile cruncher designed to eliminate books rather than select them. Those not eliminated are then sent to humans who read that genre.

    They would not be limited to using just one AI engine. They could use ten or ten thousand, each set up in a different way. Run every book through them all and see their composite scores.

    These things are used in screening candidates for a job, it’s reasonable they could be used for books.

    A more interesting approach is to run the books through a series of AI engines, then feed the results into a neural net trained on past best sellers.

    • Any bets Potter and 50Shades would have been ‘cut’?

      You can’t program for what you don’t understand yourself — and nobody can tell ahead of time what ‘will’ sell.

      I wonder how many diamonds in the rough we miss because someone else (or their program) couldn’t see …

      • In a neural net, Potter and 50 would be used as training data to teach the net what a best seller looks like.

        One of the most interesting and frustrating aspects of neural nets is that after they are trained, they have developed decision mechanisms that no human input into them. The human wrote the NN program, but after data from ten thousand books have been fed into it, he doesn’t know the decision process that resulted. He didn’t write it.

Sorry, the comment form is closed at this time.