Bot's book club: Giller finalists reprogram algorithmic reading recommendations

Adina Bresge
·4 min read

TORONTO — Readers used to rely on a cadre of cultural tastemakers for book recommendations. But nowadays, platforms like Goodreads can generate individualized reading lists in a matter of milliseconds.

Much like the Netflix model, these literary social media sites help bookworms find titles that are similar to stories they've already enjoyed based on algorithmic data about their reading habits, user reviews and hyper-specific subgenres.

The Canadian Press asked the five finalists for the Scotiabank Giller Prize to put themselves in the programmer's seat to determine where their books should fall in the code. The writers also weighed in on how computer-customized canons compare to a personal endorsement.

The $100,000 Giller Prize will be awarded at a virtual ceremony on Monday. The nominees' written responses have been edited and condensed for clarity.

Gil Adamson, author of "Ridgerunner"

CP: What niche subgenre would you categorize your book under?

Adamson: The literary western and the child's adventure tale. Also, the historical novel, the war novel, the bildungsroman, the mystery, and the crime novel. That’s a lot, but you asked.

CP: What should fans of your book read next?

Adamson: "A Prayer for the Dying" by Stewart O'Nan. Written in the second person (which is hard to pull off), it is quite brilliant, beautiful, scary, and very relevant to today.


David Bergen, author of "Here the Dark"

CP: If you were in charge of the recommendation algorithm, how would fans find your book?

Bergen, booting up his mental algorithm: On a recent glut of reading novellas, we take up "The Death of Ivan Ilyich" (by Leo Tolstoy) and then "Death in Venice" (by Thomas Mann).

And now, by death association, we read (Franz Kafka's "The Metamorphosis," which — this being a strange thread of circumstances — leads us to a jewel of a novella, "Lucy," by Jamaica Kincaid.

And out of longing for other stories about young women who are going through a sexual and intellectual metamorphosis, we find "Here the Dark," a novella (by Bergen) nesting at the end of a collection of stories.

And so we go to the front end, to the stories, which are about men, and then move on to the story collection "To be a Man" by Nicole Krauss.

Now we have fallen into the short story hole, and into Isaac Bashevis Singer's "Gimpel the Fool," which captures so brilliantly, and in a few pages, the themes of those previous novellas and short stories: desire, faith and resignation.


Emily St. John Mandel, author of "The Glass Hotel"

CP: If you were in charge of the recommendation algorithm, how would fans find your book?

Mandel: I find "The Glass Hotel" algorithm difficult to imagine, because the book is difficult to summarize. Fans of Dan Chaon's "Ill Will" might like the book because it defies easy genre categorization and there's an emphasis on character development, but "The Glass Hotel" has very little else in common with "Ill Will."

CP: How do you think algorithmic book recommendations compare to personal endorsements?

Mandel: This morning I signed up for a year's worth of books from Shakespeare and Company. Three times over the next year, I’ll receive a package in the mail from Paris, with a selection of books that someone else chose... I’m excited about it precisely because those recommendations are not algorithmic; the box will contain books that other human beings at a faraway bookstore are excited about.


Shani Mootoo, author of "Polar Vortex"

CP: How do you think algorithmic book recommendations compare to personal endorsements?

Mootoo: That my reading choices might be culled and filtered by type and habit, and then fed back to me as such, suggests that I can be put and kept in a box, one where there'd be little chance of discovering the new and unexpected. I absolutely delight in being introduced to new ideas, ways of being and thinking, and cultures that are foreign to me.

Even if algorithms are able to recognize this about me, a reduction must take place. I prefer to take more charge of the constant process of creating myself. My reason is that predetermined recommendations seem to be a spoke in the wheel of a very large machine that is attempting to create its ideal customer by stripping us of our autonomy and giving us not what we really want — after all, that machine has for a long time been robbing us of knowledge of ourselves — but what it needs us to want.


Souvankham Thammavongsa, author of "How to Pronounce Knife"

CP: What are the titles your book shares literary DNA with?

"Lost in the City" by Edward P. Jones, "One Arm" by Tennessee Williams, and "Rock Springs" by Richard Ford.

CP: Fans of (blank) will like your book because...

Thammavongsa: Fans of food will like my book because food, in particular Lao food, is as important as character.

CP: How do you think algorithmic book recommendations compare to personal endorsements?

Thammavongsa: You can pay for an algorithmic book recommendation. The book doesn't even have to be any good. Whereas a personal endorsement comes from lived and loved experience, and you can’t buy that.

This report by The Canadian Press was first published Nov. 5, 2020.

Adina Bresge, The Canadian Press