Why do I never read my highlights again?

15 September 2026

Almost everybody who highlights seriously has the same complaint: hundreds of saved passages, and no memory of reading any of them twice. It is worth being precise about why, because the usual advice treats it as a willpower problem and it is not.

You never read your highlights again because reading them requires you to decide to, and that decision competes with everything else you could open instead. Saving takes a second and happens while you are already interested. Returning takes a deliberate trip to an app you have no other reason to open, at a moment when nothing is prompting you.

Every highlighting tool is excellent at the first half and leaves the second half entirely to you. The archive grows, the reading does not, and the gap gets wider the more you save, because a larger library is more intimidating to open rather than less.

That is a structural problem, and it has a structural fix: the passages have to arrive somewhere you already are, rather than waiting somewhere you have to go.

Does it help to organise highlights into folders and tags?

Organising highlights into folders and tags reliably makes the collection tidier and almost never makes it more read. Tagging is retrieval infrastructure, and retrieval only helps when you already know you want something and are going to look for it.

The problem with unread highlights is not that you searched and could not find them. It is that you never searched. Adding a taxonomy adds work at save time and solves a problem you do not have.

Tags earn their keep for active project work, where you genuinely will go looking for everything you collected about one subject. For the general pile of things that struck you while reading, they are effort spent on the wrong half of the problem.

Should you review your highlights on a schedule?

A scheduled review of your highlights helps far less than it sounds, because the thing that makes review work is being asked a question you have to answer, and a paragraph of prose asks you nothing. What you get instead is a session you have to remember to attend, which is the same problem in a new shape.

Grading is where it breaks down in practice. Any review system wants to know whether you remembered an item so it can schedule the next one, and that question is close to meaningless for a passage that simply made you think. Systems that ask it anyway produce people clicking through reviews without reading them.

What does transfer is the spacing. Encountering something again weeks later does far more than reading it twice in one sitting, and you get that benefit from resurfacing alone, with no schedule to keep and nothing to grade.

What is the best way to actually revisit what you save?

The most effective approach is to put the passages somewhere you already look, so that revisiting requires no decision at all. The strong version of this is the new tab page, which the average person opens between twenty and fifty times a day without ever intending to. Second Read is built around exactly this: one saved passage on every new tab, with nothing to open and no review to attend.

The weaker but still useful versions are an email digest on a schedule and a message to a chat app you check anyway. Both work because they arrive rather than wait. The one that does not work is a dedicated app with a review screen, no matter how good the review screen is.

One passage at a time is better than a list. A list is a task; a single passage is just something to read, and you read it before you have decided whether to.

How many highlights is too many?

There is no number at which highlights become too many, as long as you are never asked to face them all at once. A library of five thousand passages shown one at a time is exactly as easy to read as a library of fifty.

The number that matters is how many you are shown per sitting, and the right answer is one. The dread people describe when they open a highlighting app comes from seeing the scale of the backlog, not from the content of any particular item.

So the advice to highlight less is aimed at the wrong thing. Highlight freely. Just never build an interface that shows you the pile.