How We Pick

We choose books by matching them to a reader, verify every fact we can verify against a public catalog, leave blank anything we cannot verify, and say plainly what we did not do. That is the whole method, and the rest of this page is the detail.

Every book on this site has a record in a canonical registry, and the factual half of that record is written by a script rather than by a person. It queries the Open Library search API for the title and author, confirms the match, and stores what comes back.

The point of that table is the third column. A field we cannot verify is left blank, and a page that would need it says so, rather than being quietly filled with a plausible number.

The factual layer is automated. The choosing is not, and it is the part where an honest account matters most.

1. Start from the reader, not the book. Every list begins with a specific person — a heavy reader who owns everything, a teenager who will not be told what to read, someone rebuilding a habit after years off.
2. Write the match line first. Each pick carries a "Give this if…" line, and it is written before the description. If we cannot state who a book is for, it does not go on the list. That rule is enforced in the code rather than by discipline: a pick without a match line does not compile.
3. Check length against the reader. Print length is the single biggest predictor of whether a gift gets finished, and it is the factor gift guides most consistently ignore.
4. Name an exclusion. Every roundup says what it left out and why. A list where everything is wonderful reads as a paid placement, and this is the cheapest available signal that it is not one.
5. Say what we do not know. Where a book is contested, where research has not replicated, where an edition varies — it goes on the page rather than being smoothed over.

Best of the year lists usually arrive with no stated method at all, which makes them impossible to argue with and impossible to check. Ours has one and it is deliberately boring.

We query the public Open Library catalog for books first published in that year, sorted by reading-log count — the number of catalog users who have put the book on a shelf. Then we remove textbooks, exam workbooks, duplicate editions and titles not sold in English, which is a hand edit we would rather admit than disguise.

That measures popularity, not merit, and the two are different. A shelving count tells you what people picked up; a prize list tells you what a jury admired. For a gift the first is usually more useful. The data also has known biases — catalog users are not a random sample of readers, and the coverage is stronger in romance and thriller than in literary fiction or translated work — which we state on the pages themselves.

• Invent a fact. No made-up page counts, publication dates, prices, ratings or review counts, for any reason, including to fill a table.
• Claim reading we have not done. No "in our testing", no "our favorite of the ones we tried", no implied lab.
• Publish a score. We assign no ratings and emit no rating markup, because a score implies a measurement we have not made.
• Quote a price we have not verified. There is no live price feed here, so there are no prices — the buy control links the listing instead.
• Fabricate a testimonial. There are no reader quotes on this site, because we have none.
• Take payment for placement. No publisher has paid for a position on any list, and none will.

If something here is wrong, tell us and we will fix it. Factual errors get corrected first and quickly; a page that has been materially corrected says so. Contact us — corrections go to the top of the pile, ahead of everything else.

The catalog data behind the book facts is refreshed periodically, and subscription pages are reviewed on a monthly cadence because plan terms change. Our editorial policy sets out the update schedule for each page type.