"Save for Later" Has No Later

hyungyunlim shared this post · 3h ago

"Save for Later" Has No Later

Join "Interface Studies" to help make more videos. Your support keeps new content coming. https://www.youtube.com/channel/UCqv7gk4p_rB4nRz0j7B5yFA/join

Almost every app today has a bookmark button. Almost none of them has a later.

This video follows the save-for-later pattern from a 1983 study of office desks, through the browser's bookmarks bar and the hidden menu that outlived it, to the Watch Later lists we never open.

Subscribe on Substack: https://interfacestudies.substack.com/

Happy to jump on a call if it'd be useful: design strategy, experience design, critique, feedback. https://intro.co/salehkayyali

Elsewhere:
X: https://x.com/mskayyali
BlueSky: https://bsky.app/profile/mskayyali.com

CHAPTERS

[0:00] The ribbon and the promise
[1:32] The desk: finding and reminding
[2:20] The browser had both answers
[4:32] The failure, measured
[5:19] Cause one: nothing arrives
[6:39] Cause two: the save already paid you
[7:50] Cause three: the wanting decays
[11:00] Why nothing gets fixed
[12:27] Designing for later
[16:05] A note on AI
[16:46] The second job

Transcript

0:00 welcome to Interface Studies my name is Saleh

0:05 most save for later

0:06 buttons are stylised drawings of a ribbon

0:09 look at the icons on X Instagram

0:13 Reddit there's a small pennant

0:15 with a notched tail

0:17 now the real ribbon

0:19 or the real bookmark lived inside the book

0:22 it held the exact page where you stopped

0:25 and it was there

0:26 at the moment of return

0:29 pick the book up

0:30 and it's the first thing your thumb finds

0:33 on the other hand that button in an interface kept the

0:37 shape and dropped the properties

0:40 tap it or click it

0:41 and the thing you're saving gets filed somewhere

0:44 out of sight waiting for a return

0:47 the button does nothing to arrange

0:50 I know this from my own watch

0:52 later list on YouTube is just one example

0:55 I stopped scrolling to the bottom of that list

0:58 years ago

0:59 every video in it was saved with a sincere intent

1:03 but I rarely went back

1:06 the button is now on almost every service we use

1:09 and the same list sits behind it

1:12 whether the service is a social platform

1:14 or a paid bookmarking app

1:16 all of them work on this pattern

1:19 a button and a list somewhere

1:22 and it's always about this later part

1:25 this video is about it

1:27 we save constantly and return rarely

1:30 and I want to work out why

1:32 now to understand that let's start

1:34 with 1983 a paper that I mentioned in a previous video

1:38 stuck with me Thomas Malone

1:41 interviewed office workers

1:42 about the state of their desks

1:44 and the study how do people

1:46 organize their desks found that a desk does two jobs

1:51 one is finding

1:53 you file a document so you can locate it again

1:57 the other is reminding a pile of paper sits on the desk

2:01 and its sheer presence

2:03 tells you it still needs dealing with

2:06 Malone warned that electronic systems

2:09 built to support finding while neglecting reminding

2:13 would undermine their own usefulness

2:16 he wrote that before the web existed

2:20 the web's first feature took the warning half-seriously

2:24 Netscape shipped a bookmarks menu in 1994

2:28 a hidden list it's Malone's filing cabinet

2:32 but within a few years

2:34 the browser grew the other half too

2:36 Netscape 4 added the personal toolbar in 1997

2:41 Internet Explorer added its links bar

2:44 around the same time suddenly

2:46 the same window held both of Malone's surfaces at once

2:50 the bar is the pile on the desk

2:53 a handful of links permanently visible

2:56 their presence doing the nagging

2:59 the menu is the cabinet unbounded and out of sight

3:04 but that bar that we see in browsers has a cost

3:08 there's a very limited number of bookmarks

3:10 that can be displayed all the time

3:12 in our web browser's interface

3:14 and the case is harder on mobile

3:17 presence is rationed by pixels and saving compounds

3:22 a 1998 study of over 300 bookmark users

3:26 information archiving with bookmarks

3:28 found collections growing steadily

3:31 while the organizing never kept up

3:34 so the bar overflows

3:36 and what was done to handle this problem

3:38 was relatively simple

3:40 a small Chevron at the end of the bar

3:43 that opens a hidden list you can access the full list

3:46 of course but that is hidden behind the browser's menus

3:51 the overflow control

3:52 for the visible pile is the invisible pile again

3:56 presence does not scale

3:59 when the platforms arrived

4:01 and needed a save pattern to copy

4:04 both designs were sitting there in the browser

4:07 and they took the menu Watch Later saved posts

4:11 favorites bookmarks

4:13 every one of them is the unbounded hidden list

4:16 and it was even shipped without the bar

4:18 that we see in browsers it was out of sight

4:22 and out of sight was never an accident

4:24 of the save button it was one of two available designs

4:28 and it is the one everybody chose

4:32 that hidden list to access things that we saved

4:35 fails in a measurable way

4:38 in 2021

4:39 Ofer Bergman and his colleagues published a study

4:42 whose title does half of the work for me

4:44 out of sight and out of mind

4:47 bookmarks are created but not used

4:50 they asked 50 people to retrieve pages

4:53 they'd personally bookmarked

4:55 and watched how they did it

4:57 people get back to pages by other routes

5:00 and the bookmarks themselves just sit there

5:03 now we don't know much about the numbers

5:05 because nobody publishes them

5:07 we have some signals

5:08 from specialized apps that we will go through later

5:11 but the question is what exactly stops the return

5:15 the evidence seems to point at three things

5:19 the first cause is forgetting

5:21 and it is the solved one

5:23 a saved item has no presence

5:26 so nothing brings it back to mind

5:29 interestingly

5:30 paper offices solved this for most of a century

5:33 with the tickler file a drawer

5:36 of 43 folders one for each day of the month

5:39 and one for each month of the year

5:42 drop a document in the folder for the 9th

5:44 and on the morning of the 9th

5:46 it lands on your desk in software

5:50 email solved it

5:52 Gmail added

5:53 snooze in 2018 after it was introduced in a brief

5:56 experimental app called inbox

5:59 dismiss a message name a time

6:02 and at that time

6:03 it returns to your inbox as though newly arrived

6:08 Slack went further in March 2023

6:11 and merged saved items with reminders

6:14 into a feature named of all things later it is where

6:18 a save has states and can carry a return time

6:23 now when it comes to email

6:24 what makes it particularly useful

6:27 is that it has something most software lacks

6:30 an arrival surface

6:31 and a habit attached to it it is a place where things land

6:35 and where your attention already goes every day

6:39 the second cause is quieter

6:42 in 2015

6:43 Benjamin Storm and Sean Stone ran a set of experiments

6:47 exploring the saving enhanced memory effect

6:50 which is a cognitive phenomenon

6:51 where saving information digitally or physically

6:55 improves your ability to learn and remember

6:57 new information

6:59 in their paper

7:00 people studied a list of words in a computer file

7:04 some could save the file before studying a second list

7:08 and the people who saved

7:10 remembered the new material better

7:13 saving the first list let their minds release it

7:17 the effect vanished though

7:18 when the saving seemed unreliable

7:22 your mind only lets go of what it trusts the system

7:25 to hold

7:27 that's a lab result about word lists

7:30 so let's carry it as a mechanism

7:32 rather than a measurement

7:33 but the mechanism itself explains something real

7:37 part of the reason you never go back to your

7:39 saved items list is that

7:42 the save closed the loop at the moment of capture

7:46 that itch was scratched

7:47 the moment you pressed that button

7:50 the third cause is the one worth the deepest look

7:52 because no reminder can beat it

7:54 and most saves die of it

7:58 the cleanest evidence comes from a system

8:00 where forgetting was nearly impossible

8:03 DVDs by mail

8:05 now before streaming

8:07 this is how Netflix and services like it worked

8:10 you kept a

8:11 ranked wishlist of movies

8:12 the company posted you the disc at the top

8:15 and returning it summoned the next one

8:18 the system remembered for you and delivered for you

8:21 and the disc

8:22 then sat in your house until you dealt with it

8:26 it's Malone's reminding pile

8:27 but it's rebuilt by a DVD company

8:30 in 2009 Katherine Milkman and her colleagues

8:34 got hold of the rental records of

8:36 one of these services

8:38 and they measured how long people kept each disc

8:41 the films people simply wanted to watch

8:44 came and went quickly

8:46 the films they felt they must watch or ought to watch

8:50 like the documentaries and the serious dramas

8:53 sat for longer unwatched nobody had forgotten anything

8:57 the disc was right there what had faded was the wanting

9:03 it was somewhere between the day they queued the film

9:05 and the day it hit the doormat

9:08 the study is called Highbrow Films Gather Dust

9:13 economists have a name for this present bias

9:17 that disproportionate weight

9:19 we give to whatever is immediate

9:22 in doing it now or later

9:24 which is a 1999 paper by Ted O'donoghue

9:28 and Matthew Rabin formalized it

9:30 and it means that every save is a plan

9:33 made by your present self for a future one

9:36 who will probably not agree to it

9:39 and that decay in interest is Fast. Pocket

9:42 which was an app for saving for later

9:45 its internal data

9:46 put the active life of and even successful

9:49 long form article at about 37 days

9:52 with most saves cooling quicker

9:55 the read later founders knew this

9:58 when they were asked in 2013

10:00 about the fraction of pocket saves that got read

10:03 they offered 10 to 70% averaging about 50

10:07 another unexpected

10:09 thing that finished this argument for me

10:11 was the fact that saved objects decay as well

10:16 Pew Research measured what they call digital decay

10:19 in 2024 and they found that 38% of web

10:24 pages that existed in 2013 were gone

10:27 a decade later

10:29 a quarter of all pages from the decade had vanished

10:32 even pages from 2021 were dead within two years

10:36 at a rate of roughly one to five

10:39 the pile seems to rot from both ends

10:42 your interest has a half life measured in weeks

10:45 and the page itself

10:47 has meaningful odds of dying before you're back

10:50 every saved link is a race between two decays

10:54 and a reminder doesn't stop either

10:57 it can only arrive before they finish

11:00 all of this leaves the question of why

11:03 the solved cause stayed unsolved on every platform

11:08 as I said earlier

11:10 platforms don't publish their save-to-return figures

11:13 but there's something that we can see

11:15 in the deployment of these features

11:17 it seems that at the moment of capture

11:20 a save is one of the strongest signals you can send

11:24 it tells the recommendation system what you value

11:28 and it deepens your investment in the account

11:31 holding the list

11:33 that value is banked the instant you press

11:36 the return produces nothing

11:38 the platform wasn't already collecting

11:41 and in their interest

11:42 it's better to show newer content than an old one

11:45 an active reminder would cost something real as well

11:49 it's a slot in the feed or even a notification

11:53 and each is priced against the recommendation or worse

11:57 an ad

11:58 notifications themselves are a challenge

12:00 as we saw in the previous video

12:02 and you cannot just send notifications on everything

12:05 on the other hand we see some

12:07 thoughtfulness

12:07 in the way that Slack built its later feature

12:10 the feature that I spoke about earlier

12:12 because Slack sells completed work

12:15 a returning message getting handled

12:17 is the product doing its job

12:20 YouTube sells you watch time

12:22 and the feed already supplies it

12:24 without honoring old promises

12:27 now this does not mean that designing

12:30 for later is an impossible task

12:32 and if we suppose that we're building a save feature

12:35 and we want later to mean something

12:37 it seems that we have good examples

12:39 from paid bookmarking apps

12:41 especially the modern ones

12:43 they are scattered and everywhere

12:45 and no single tool holds them all

12:47 but at least they exist

12:49 so how can we do this

12:51 the first one is resurfacing early

12:53 because decay sets the clock

12:56 a reminder after a week

12:58 can still catch the motive that produced the save

13:01 but if it was four or five weeks later

13:03 it arrives addressed to someone else

13:05 Raindrop

13:06 which is a bookmark manager funded by subscriptions

13:09 sells exactly this as a paid feature

13:12 pick a bookmark and a time

13:14 and it comes back as an email or a notification

13:17 it would be interesting if this was automated

13:19 but at least within Reindrop you have the option

13:23 the second one is bounding the arrivals

13:26 because the bookmark bar already taught us that

13:28 presence doesn't scale

13:30 Readwise which is another subscription tool

13:33 sends a daily review of your saved highlights

13:37 a few items a day never the pile

13:40 the daily budget matters

13:41 more than the algorithm behind it

13:43 and the budget must carry no debt

13:46 an item ignored today should just reschedule

13:49 silently show people their saves as a liability

13:53 and you've built a guilt pile with a counter on it

13:56 Number three is letting items retire

13:59 because the save already paid

14:00 at the moment

14:01 we press. a save resurfaced twice without action

14:05 should archive itself

14:07 and the interface should frame that as completion

14:10 mymind a bookmarking tool with no free tier

14:14 ships this as Serendipity

14:16 which is a feature that lives inside the app

14:18 a short session of about 20 saves resurfaced at random

14:22 each met with one of two actions keep or forget

14:28 this session is bounded so it never becomes work

14:31 the selection is random so there's no debt

14:34 forgetting is a first class action here

14:38 the one thing serendipity

14:39 what doesn't do is arrive

14:41 you have to open it

14:43 it might make an occasional appearance

14:45 in their regular marketing communication

14:49 finally the hard one

14:51 finding an arrival surface

14:53 because the forgetting half itself

14:55 is only solved when something lands

14:58 for a standalone tool

15:00 it seems very difficult to find another surface

15:03 other than email

15:04 which is why the Readwise review and Raindrop reminder

15:07 both arrive there

15:09 that is not to say that email

15:10 doesn't come with its problems

15:12 because a lot of things compete

15:13 for our attention in email

15:15 but it is the most stable one we have

15:18 and even for some of the modern

15:19 standalone bookmarking services

15:21 they are now developing what can be only called a feed

15:24 and that takes us to the second point

15:26 for a platform like X, BlueSky, Instagram, TikTok

15:31 the only honest surface is the feed itself

15:35 there a resurfaced save

15:38 competes with everything the feed would rather show you

15:41 and from what we have seen so far

15:43 the algorithm only prefers new things

15:46 my hunch is that a resurfaced save

15:49 could pay for its slot

15:51 since the content that we resurface

15:53 there

15:53 is something that has been declared

15:55 as valuable to the user

15:57 the platform's behavior now says that they disagree

16:00 part of the answer as I said earlier

16:02 is probably because they only care about the signal

16:05 also note

16:06 that none of what I spoke about so far involves AI

16:09 because I think this is not an intelligence problem

16:12 what we see now

16:13 even in modern tools is using AI to summarize links

16:17 but a model that summarizes your saved link

16:20 answers a question and most saves were never a question

16:24 summarizing might spoil the article even

16:27 what the return needs is a surface to land on

16:30 and a reason to build that surface

16:33 no model supplies either

16:36 it would be interesting if certain machine

16:37 learning algorithms were used in the feed

16:40 to combine all the previous best practices

16:43 and deliver bookmark items regularly

16:46 at the end of all this

16:47 and by all the evidence that we went through

16:50 most of those saves that we do on all platforms

16:54 had served their purpose in the moment they were made

16:57 and the rest were racing decays

16:59 they were always going to lose

17:02 fixing the return stays

17:04 open

17:05 some of the ideas for it are old

17:07 it works on paper it works in email

17:10 when it comes to platforms

17:12 they come and go

17:13 what matters is a surface

17:15 that lets the bookmarking button

17:17 earn its place

17:18 the same way the physical bookmarking ribbon

17:21 earned its place in the book