eroMark Sign in

Why you never read what you save

Saving costs a tap. Reading costs an evening. The gap is the whole story.

Saving takes one second and reading takes ten minutes. Any system where the input is six hundred times cheaper than the output accumulates without limit. That is not a personal failing and no amount of discipline fixes an arithmetic problem.

This piece is about that ratio and what follows from it: why the pile compounds rather than accumulating, why a better-organised archive makes the situation worse rather than better, and which four changes actually move the number.

THE TAP 1 s one second THE SAME POST, READ 1 2 3 4 5 6 7 8 9 10 ten minutes 600 : 1 The folder is what this price does.

Figure 1. The flame chip is the tap: one second. Each slab is one minute of actually reading the same post. Ten minutes is six hundred seconds. Six hundred to one. No amount of being more selective closes a gap of that shape. Arithmetic: a one-second tap against a ten-minute read.

What a save actually is

When you bookmark something you are not deciding to read it. You are deciding not to decide, and buying the feeling of having acted while the thread scrolls on. The save closes the small anxiety of losing something, at the price of a future obligation you have not scheduled.

The obligation is real but invisible, which is the problem. Nothing in the interface shows you that you have accepted twelve hours of reading, and nothing asks when you plan to do it.

Backlogs are not linear

Backlogs are not linear. Several things compound:

This is why "I will get to it at the weekend" is not a plan. Each week the pile is larger and the average item is worth less.

0200400600 Saved Read THE BACKLOG is this area, not a number Month 06121824 Month 8: the pile gets large enough to avoid, and the reading line bends the wrong way.

Figure 2. Both lines only ever rise, so the backlog is the area between them, and it widens every month whether or not you read anything. Note where the reading line bends: not at the start, but around the point the pile becomes large enough to be a mood rather than a list. Schematic, at thirty saves a month against a reading rate that decays as the pile grows.

Fifteen years of optimising the wrong half

The read-it-later category has spent fifteen years optimising the wrong half. Faster capture, browser extensions, share sheets, one-tap saves, sync, full-text search, tags, nested collections, highlights. Every one of those makes saving cheaper or storage more comfortable.

None of them makes the pile smaller. A beautifully organised archive of two thousand unread articles is still two thousand unread articles, and it is now pleasant enough to live with that you will not do anything about it. Comfort is the enemy here: it removes the pressure that would otherwise force a decision.

Ask what number goes down when you use it. If the answer is none, it is an archive.

A LIBRARY The total only rises. Correct behaviour, if you are building a reference collection you intend to query. A QUEUE The total reaches zero. Correct behaviour, if the pile is the thing that is bothering you. Nothing else fixes that. Ask of any tool: what number goes down when I use this?

Figure 3. The same saves arrive in both. The only difference is whether anything ever leaves, and it is the difference between a collection you are building and a queue you are behind on. Most read-it-later tools have no number that can go down, which tells you which one they are. Schematic.

Why a folder of links produces so much dread

A bookmark folder with two thousand items in it produces a disproportionate amount of dread, and it is worth understanding why, because the explanation points at the fix.

Each save was a small promise: I will come back to this. Individually trivial. In aggregate you are carrying two thousand open commitments you made to yourself and did not keep, and the pile is the visible evidence. That is why "just delete it all" feels like it costs something even though the contents are objectively worthless to you: the cost is admitting the promises were never going to be kept.

It is also why organising is so seductive. Filing feels like partial credit. You have not read anything, but you have handled it, and the anxiety drops for a day or two before the pile resumes being a pile with better labels.

The saving is fine. The missing half is the exit.

Worth saying, because the usual advice is "be more selective" and that is mostly wrong. Saving is cheap, fast, and a decent signal of interest in the moment. The impulse is fine.

What is missing is the other side. Every functional queue in your life has an exit: email gets archived, tasks get closed, laundry gets put away. Bookmarks have a save action, a delete action nobody uses, and nothing in between. There is no equivalent of "handled". So the only two states are unsaved and saved-forever, and everything accumulates in the second one.

Add a way for things to leave that is not "finish reading it" and the arithmetic changes without anyone having to develop more discipline.

Four things that move the number

A finish line

Zero has to be reachable and visible. "Unread: 1,847" is not a target, it is weather. A queue you can empty in fifteen minutes is a task; one you cannot is a condition.

Letting go as a normal move

Most of what you save should be deleted unread, and the interface should make that feel like progress rather than failure. Any system where the only ways out are "read it" or "keep it forever" will fill up, because those are the two most expensive options available.

Deciding at a different time than saving

The save is impulsive and that is fine; impulse is a decent filter for interest. The mistake is expecting the same impulsive moment to also produce a plan. Separate them: capture freely, decide in a block, on a schedule.

Oldest first

Newest-first is how a queue becomes permanent. The recent stuff gets re-read and re-postponed while the bottom of the pile is never seen again. Working oldest first is the only order that terminates. The method for doing that, oldest reachable save first, is in How to clear a bookmark backlog.

Read It Later names the wrong half of the transaction

The category named itself after the promise rather than the outcome, and the name has been doing quiet damage ever since. "Read it later" says the default is reading and the only variable is timing. Both halves are wrong: the default is not reading, and there is no later, there is only a growing set of things competing with whatever is in front of you right now.

A name closer to the truth would be "decide about this later", which is both less appealing and what actually happens. It also suggests the right interface, because a decision queue and a reading list want completely different things. A reading list wants comfort, storage, and beautiful typography. A decision queue wants speed, a single item at a time, and an exit.

Not a collection you are building. A queue you are behind on.

Stop thinking of it as a reading list and start thinking of it as an inbox. Nobody feels guilty about deleting email unread. Nobody organises their inbox into a permanent archive of unanswered messages and calls it a library. The measure of an inbox is whether it empties.

Applied to bookmarks that means: the pile is not a collection you are building. It is a queue you are behind on, and the only healthy state is empty.

Read next