Writing · 21 September 2026

The read-later graveyard.

Mozilla switched Pocket off on 8 July 2025 and deleted its data on 12 November. The shutdown showed how little of a read-later app is about the reading.

Three dates

22 May 2025. Mozilla announced Pocket was closing. It came off the app stores that day, and new sign-ups and renewals stopped.
8 July 2025. Pocket stopped working. The export tool stayed open for 90 days.
12 November 2025. Mozilla closed the export tool, switched off the API and deleted the data.

The last date is the one to think about. Mozilla had shipped Pocket inside Firefox since 2015, so a lot of people got it by default, without picking it. The articles they had saved lived on Mozilla's servers, and Mozilla decided to switch those off.

The easy lesson is to distrust free services. A read-later app is a promise about tomorrow, and that promise is the whole product: you save something today on the strength of reading it in a year. After a shutdown the useful question is which survivor you would hand the next ten years of your reading to.

The pile problem

Read-later apps sell the same first half of the loop: save it, get it out of your way. That half is solved, and it is solved for free. Firefox's reader mode, Safari's Reader and the browser's own reading list take a page, strip the noise and hold the text. If you only need "I will read it in a minute", you do not need an app.

The second half goes unsolved because saves are what the category sells. Saves are the number that goes up. No one puts "articles you got to" on a pricing page.

A read-later app with no answer for how the pile shrinks is a storage product with a magazine logo on it. A company that needs to cut costs goes for storage first.

The survivors

Three kinds of app carried on after Pocket. Each one answers the second half of the loop in its own way. The names matter less than the pattern.

The habit machines. Readwise Reader is the clearest example. You highlight as you read, the highlights flow into wherever your notes already live, and a review step puts old highlights back in front of you on purpose. The design assumes you will come back to what you marked, and it plans for that.

The format specialists. Instapaper syncs with a Kobo. Matter does newsletters and audio. Both picked one place where reading happens and got good at it, instead of trying to be everywhere a page might be read.

The ones you run yourself. Wallabag, Karakeep, Linkwarden. They sell insurance rather than features: you hold the server, so no company can switch it off.

The browser's own reading list survives too, along with the folder of tabs you keep meaning to get to. It is unglamorous and hard to beat on the two things that matter most: it opens anywhere, and it costs nothing.

Each survivor gives you a reason to come back, or a reason to trust it with the pile. Pocket managed neither by the end, which is forgivable. Mozilla set the deadline, and Mozilla gave you 90 days to get your reading out, which is harder to forgive.

If you are choosing one now

These are ordered by what it costs you to get one wrong.

Can you take the article text with you? Ask for the article text rather than the list of addresses. An export that hands you links which need the live web to resolve is an address book, and in ten years some of those addresses will be gone.
Does it work when the network does not? Test it on a train, a flight or in a hospital waiting room. An app that needs a connection to show you a saved article has not solved the problem.
Who is paying for it? Free is a good answer as long as you know which kind of free. Someone else pays, someone is investing, or you are the thing being sold. Avoid the middle case, where no one pays and no one invests.
Does it nag you? A streak counter turns a reading list into a guilt meter. Give someone a score for reading and they will read less.
Is the queue still a list you made? Once an app starts ranking your saves for you, the pile has become the business model.

Where Settee fits

The claim is narrow on purpose: the queue should be the list you made, and it should still be there tomorrow. In practice:

Saving is one click, and Settee keeps the article. Settee fetches the piece, strips it back to the text and stores it alongside the item, so what you saved is what you read, even if someone rewrites the page, paywalls it or takes it down.
It works with no signal. Settee keeps the articles in your browser's own storage, so the queue opens on the train.
A Pocket export works as it is. Hand Settee the file Mozilla gave you and it reads the whole thing, so you are not pasting addresses in one at a time.
Your library exports with the article text in it. The file carries the text of each piece, and export is part of the paid tier from day one. It ships on the first day rather than ninety days after we announce we are closing, which is the honest way to promise something we cannot otherwise prove.
No recommendation feed and no streak counter. Nothing in Settee is designed to make a number go up except the pieces you finish.