How to find your next book: 5 systems that actually work

Goodreads recommendations are garbage, BookTok is a slot machine, and your last great book came from a friend who knows you — here's how to systematize that

The paradox of book discovery: we have more ways to find books than ever, and most of them don't work. Goodreads serves you algorithmic slop based on star ratings from people who think The Kite Runner and Where the Crawdads Sing are both "life-changing." BookTok gives you 47 seconds of someone crying over A Little Life with zero context about whether you'll also cry or just feel manipulated. Amazon's "customers who bought this also bought" is just collaborative filtering that thinks you're a genre, not a person.

The books you actually love came from somewhere else. A friend who texted you a photo of a cover. A essay that quoted three paragraphs. A used bookstore where you grabbed something because the spine was broken in the right way. Those aren't repeatable systems, which is why you're here.

Here are five methods that work, with trade-offs for each type of reader.

1. The One-Degree Network (for readers who trust people, not platforms)

This is the oldest system and still the best: ask someone whose taste you've pre-vetted. Not "what are you reading," which gets you their current book (might be a hate-read or a book club hostage situation). Ask "what's a book you've recommended three times in the last year?"

That question surfaces the book they're evangelizing, which means it's doing emotional work for them. They want to convert people. You want to be converted.

The network is one degree: friends, coworkers, that person at the bookstore who remembers you bought The Debt to Pleasure and correctly guessed you'd like The Sailor Who Fell from Grace with the Sea. Not Reddit threads where 4,000 strangers upvote Project Hail Mary because it's competent and inoffensive.

Trade-off: This only works if you know readers. If your friends are non-readers or only read Colleen Hoover, you need a different system. Also slow — you can't summon a recommendation at 11 p.m. when you just finished a book and need another one now.

Best for: Re-Reader Loyalists who are careful about letting new books into the rotation, and Library-Card Maximizers who can't afford to burn a hold slot on a dud.

2. The Omnivore's Bibliography (for readers who treat books like a syllabus)

Find a writer or critic whose taste is 80% aligned with yours — not 100%, because then they're redundant — and read their bibliographies. Not their book recommendations, their bibliographies. The books they cite in footnotes. The books they name-drop in essays. The books that taught them how to write.

Examples: If you loved The Empathy Exams by Leslie Jamison, go read her essays in The Believer and pull every title she mentions. If you're into Maggie Nelson, her The Argonauts bibliography is a MFA reading list that costs $0 to access. If you liked How to Do Nothing by Jenny Odell, she gives you 40 books in the back matter.

This works because good writers are good readers, and they're not trying to sell you anything when they cite a source. They're showing you the infrastructure of their thinking.

Trade-off: You have to read a certain kind of book for this to work — books with bibliographies, footnotes, acknowledgments that do intellectual work. If you mostly read thrillers or romance, this method gives you nothing. Also, it's backward-looking. You're reading the books that shaped a writer 10 years ago, not the books published last month.

Best for: Annotator Scholars who are already reading with a pen in hand, and One-Book-a-Night Devourers who need a pipeline that doesn't run dry.

How to actually do this

1. Pick a book you loved that has a bibliography or acknowledgments section.
2. Pull 5-10 titles that the author cites more than once or describes in detail.
3. Cross-reference those titles on your library's catalog. Put 3 on hold.
4. When you finish one, repeat with that book's bibliography.

You've just built a reading chain that compounds. Every book leads to three more, and they're all vetted by someone whose judgment you trust because you already read 300 pages of their work.

3. The Negative-Space Method (for readers who know what they don't want)

Most discovery systems ask what you liked. This one asks what you hated, then finds books that are orthogonal to that thing.

Example: You hated Where the Crawdads Sing because the prose was bad and the plot was a Lifetime movie. What's the opposite? Not "literary fiction" (too broad). The opposite is a book where the prose does real work and the plot isn't manipulative. That gets you to someone like The Great Believers by Rebecca Makkai or Commonwealth by Ann Patchett — books where the writing is the point and the plot earns its emotions.

Or: You hated Atomic Habits because it was 20 pages of ideas stretched to 300 pages of anecdata. The opposite isn't another productivity book. It's a book that's dense and doesn't waste your time. That gets you to Thinking, Fast and Slow by Daniel Kahneman or The Righteous Mind by Jonathan Haidt — books that could've been longer.

Trade-off: This requires you to articulate why you hated something, which is harder than knowing you hated it. Also, "opposite" is subjective. If you hate Crawdads because it's set in a swamp, the opposite is... a book set in a city? That's not a useful vector.

Best for: DNF Queens who have strong opinions and a low tolerance for bullshit, and readers who've taken the readertype quiz and know their non-negotiables.

4. The Bookseller Handoff (for readers who want curation, not algorithms)

Go to an independent bookstore. Not Barnes & Noble, where the staff are rotating teenagers who don't read. An indie where the booksellers are adults with opinions.

Tell them three books you loved and one book you hated. Be specific: not "I like literary fiction" but "I loved The Overstory by Richard Powers, Lincoln in the Bardo by George Saunders, and The Fifth Season by N.K. Jemisin. I hated Normal People because nothing happened."

A good bookseller will walk you to a shelf and hand you two books. One will be obvious (they heard "experimental structure" and grabbed Cloud Cuckoo Land). One will be a curveball (they heard "hated Normal People" and grabbed My Year of Rest and Relaxation, because if you hate quiet books, maybe you want a loud book about doing nothing).

The curveball is why this works. You're not getting algorithmic averages. You're getting a human who's guessing about you based on incomplete information, which is how all good recommendations work.

Trade-off: You have to leave your house. You have to talk to a stranger. If you live somewhere without an indie bookstore, this is a non-starter. Also, this costs money — you're expected to buy one of the books they recommend, and you can't return it if you hate it.

Best for: Multi-Book Jugglers who need variety and aren't loyal to a single genre, and Library-Card Maximizers who are willing to buy one book a quarter to support the store that helps them find the other eleven.

5. The Reverse-Engineering Method (for readers who want to understand their own taste)

This is the most labor-intensive method and the most effective long-term. You're going to build a database of your own taste.

Step one: List the last 10 books you loved. Not liked. Loved. The ones you stayed up until 2 a.m. to finish or reread within a year.

Step two: For each book, write down three attributes that aren't genre. Not "it's a thriller" but "the pacing was relentless," "the protagonist was unlikable but compelling," "the setting was claustrophobic." You're looking for texture, not taxonomy.

Step three: Look for overlaps. If seven of your ten books have "unreliable narrator," you now know that's load-bearing for you. If six have "prose that's smarter than the plot," you know you're not a plot-driven reader. If five are set in small towns, maybe you're into geographic claustrophobia.

Step four: Search those attributes on literary blogs, Bookshop.org, or even Google. "Unreliable narrator literary fiction 2023" gets you better results than "books like Gone Girl."

Trade-off: This takes time. You're doing the work that an algorithm is supposed to do, except you're doing it better because you know what "compelling" means to you and Goodreads doesn't. Also, this only works if you've read enough books to see patterns. If you've only loved three books, you don't have a dataset.

Best for: Annotator Scholars who already track their reading, and One-Book-a-Night Devourers who read enough volume to see their own patterns.

Why these work better than algorithms

Every algorithm recommends based on similarity: people who liked X also liked Y. That's fine if you're buying dish soap, where one brand is mostly like another. Books aren't like that. The thing that makes a book great for you is often not the thing that's legible to an algorithm.

You loved The Goldfinch not because it's "literary fiction set in New York" but because Donna Tartt's sentences have a specific kind of headlong momentum that makes you forget you're reading 700 pages. An algorithm can't see momentum. It can only see metadata: genre, setting, comp titles, star ratings from people who aren't you.

These five systems work because they route around metadata. The One-Degree Network routes through humans who know you. The Omnivore's Bibliography routes through writers who've already done the filtering. The Negative-Space Method routes through your strong opinions, which are more predictive than your mild likes. The Bookseller Handoff routes through curation. The Reverse-Engineering Method routes through your own data, which is the only data that matters.

Which system to use when

If you just finished a book and need another one tonight: One-Degree Network (text three friends) or Bookseller Handoff (if the store is open).

If you're building a reading list for the next three months: Omnivore's Bibliography or Reverse-Engineering Method.

If you're in a reading slump and everything sounds boring: Negative-Space Method. Figure out what you're actively avoiding, then find its opposite.

If you don't know what kind of reader you are yet: Start by figuring that out, because your discovery system should match your reading habits. DNF Queens need low-commitment systems (library holds, bookseller recs you can return). One-Book-a-Night Devourers need high-volume pipelines (bibliographies, reverse-engineering). Re-Reader Loyalists need high-conviction systems (one-degree network, negative-space).

None of these systems are fast. That's the point. If book discovery were easy, Goodreads would've solved it. The reason you're still looking for your next great book is because "great" is subjective and algorithms are not.