HackerBooks Reading Notes: Choose Your Next Book in 2026
A reading decision, not another aspirational shelf: what to save before and after you open a book.
November 02, 2026
HackerBooks reading notes should start with a question you want a book to answer. Use the site to find candidates, open the comments behind a recommendation, then try a sample. Its AI-written recommendation summaries describe other people's reactions. They aren't a substitute for reading the book or deciding whether it fits your task.[1]
On November 2, HackerBooks was second on the Hacker News front page when I checked, with 1,723 points and 282 comments. Korben also published an independent report that day.[2][3] There is enough interest here to build another enormous reading list. I'd use it to make a smaller one.
What the recommendation numbers actually count
HackerBooks says its current dataset covers 15,766 books and 281,363 mentions from Hacker News comments between 2007 and 2026. The launch post still describes roughly 16,500 books. The live catalogue is dated November 2, so the figures in this article follow that version rather than treating the launch total as fixed.[1][3]
The useful change is that the site separates being mentioned from being recommended. A complaint about a book can make its name appear more often without giving you a reason to buy it. HackerBooks labels mentions as recommending, neutral, critical, or unrelated. Its recommendation count measures different people who recommended a title at least once.[1]
That still leaves judgment in the pipeline. The About page says the labels were compared with another Claude model on 400 mentions and agreed 85% of the time. This is agreement between two models, not a human-verified accuracy rate. It doesn't establish that 85% of the recommendations are correct, useful, or appropriate for you.[1]
Korben gets the important distinction right by drawing attention to mentions versus recommendations.[2] But even a clean recommendation count can't tell whether a book will help with the document on your desk. A person praising a reference manual after years of experience may have different needs from someone seeking an introduction.
A summary of praise is not a summary of the book
Open the entry for William Zinsser's On Writing Well. Beneath its recommendation numbers, HackerBooks shows a short explanation of why readers recommend it. The page explicitly labels that explanation as a Claude summary of recommending comments. Further down, you can read the underlying comments and follow links back to their threads.[1]
That label deserves to survive into your notes. If you save the explanation as "what the book says," you've changed the kind of evidence you have. A description of readers praising clear prose doesn't show which exercises the book contains, what a particular chapter argues, or whether an edition covers your problem.
The page also preserves the untidiness of real discussion. One displayed comment attaches the title to Zig Ziglar, while the book entry identifies William Zinsser. That doesn't prove the catalogue's author field is wrong. It shows why a collection of comments shouldn't be copied into a polished bibliography without checking the actual book.[1]
For a writer choosing something to read, this is manageable. Follow a recommendation that sounds relevant. Look at what the person was trying to do, not just the positive label. If the comment gives no reason, keep browsing. You don't owe the highest-ranked title your next weekend.
Avoid asking an assistant to turn the catalogue's praise into a chapter-by-chapter account. Without the book as a source, that request invites an answer beyond the material you supplied. Asking which comments discuss your particular writing problem is a narrower, checkable task.
Choose a book for the document you need to finish
Suppose you're preparing an internal guide for new volunteers. This is a hypothetical example. People keep misunderstanding the handover section, and your next draft needs clearer instructions. A book's reputation matters less here than whether its advice helps you revise that section.
Write the problem before opening another book page: "I need a new volunteer to understand what to do when the next shift hasn't arrived." Then browse Writing & Language, open a promising title, and read the recommendation context. A comment about elegant essays may interest you without answering this immediate need. Keep it for another day.
Next, look for an accessible sample through the publisher, a library, or an authorized preview. Check the contents and read enough prose to judge whether you want to continue. A recommendation shelf can point you toward a sample. It cannot tell you what you found there until you've opened it.
In your HackerBooks reading notes, record the decision in ordinary language. "Try the sample for clearer instructions" is enough. Add the book link so you can return. Don't mark it read, write a review, or collect quotations from a summary of comments.
After sampling, return to the volunteer guide. Attempt one revision using an idea you actually read. If the sample didn't help, say so. You can stop there without deciding that the book is bad. It may simply be the wrong purchase for the work you need to finish.
Speak a reading intention, then update it after reading
You can type that short note anywhere you already keep your reading list. Voice input is useful if choosing words aloud feels easier than opening another elaborate notes template. Keep the book link separate and paste it accurately. Speaking an address or an edition number adds work you don't need.
For example, a spoken note might say: "Unread. Found through HackerBooks. Check whether the sample helps me explain the late-shift handover. I haven't checked the chapter yet." This is an invented example, not a quotation from a reader. The words "unread" and "haven't checked" are important. Review them after transcription or AI cleanup.
DictaFlow provides hold-to-talk dictation at the active cursor on desktop: hold the shortcut, speak, then release to insert text. Its current product facts describe the workflow and its limits.[4] It can help put your own reading intention into an editable note. It doesn't read HackerBooks for you, check the book, or make a catalogue summary reliable.
You don't need another subscription just to try this habit. Use a keyboard or a voice tool you already have. If you try DictaFlow, test a disposable note first and check that the text lands in the intended field. The setup guide covers insertion controls. Keep whichever input method makes this small task easier.
When you return after reading, leave the original intention visible and add what happened. Name the chapter or passage you actually reached. Explain whether it changed the volunteer instructions. A decision not to finish the book belongs in the note too. That gives you something useful to revisit when the same title appears on another recommendation list.
Disclosure: This is a DictaFlow editorial written with AI from published sources. The volunteer-guide examples are hypothetical. We inspected the catalogue and its methodology, not the books themselves.
Sources
[1] HackerBooks methodology and live catalogue, November 2
[2] Korben: independent report, November 2