Feat/Desired retention warning improvements #3995
Conversation
|
Nothing else in the deck options screen shows an info 'warning' out of the box. This PR increases the amount of text the user sees by default, which I'm not really fond of - those boxes are intended to call the user's attention to things that may not be right. That said, I do appreciate that devs want to make the option more discoverable, and showing something in the 90% case may be the best way to do that. On the wording: it doesn't feel very accessible. If your goal is to appeal to the masses, 'retention', 'intervals' etc is not great. What about something like the following? "To see cards more frequently, increase this above 90%" |
|
Maybe both this message and(?) the
With the exception of if desired retention is set to a "warning" or "danger" value. |
|
Dae
But the setting is literally called "desired retention" Luc
No, please don't hide "A 100 days interval will become X days". Hiding the default message is ok. |
|
The idea of a subtitle/caption is to explain the (by necessity, concise) title in more detail/simpler language. If you're arguing that the description should use complicated language because the title does, I don't understand that logic. If you're arguing that we shouldn't bother as the title itself is too complicated, do you have a suggestion for a better title we could use? |
Why not? As far as I see it it's extra information that's provided to give you perspective while changing your desired retention. If you're not modifying it and there's no problem with it then how would it help? |
|
would it be clear "this value" refers to desired retention and not the params below it? sure, many native speakers might understand "this value" is a single value but there might be some confusion. any thoughts on this?
have said this last year that this info is almost meaningless. when it comes to exact stats, how workload is increasing matters more from a user perspective. ivls are secondary to that. +1 to removing this bit. |
The goal is to make people understand how desired retention affects interval lengths. Removing that message would move us further away from achieving that goal. |
|
@dae @Luc-Mcgrady how about only displaying the "The higher the desired retention, the shorter the intervals between reviews." message if the user has never changed DR from the default value of 90? Add another variable like |
|
There's already a page explaining what DR does inside the settings menu for people who care. I think this would just be unnecessary clutter for people who either don't care or already know. If this is implemented I def agree that it should only be shown when the user enables FSRS for the first time, or if the user has never changed DR from the default. |
|
I don't want to faff about with extra logic to show such a message only if the value has never been changed before, as we have to track that then. I'm not particularly fond of the current wording either, as I don't like how we switch between quantitive and qualitative messages for only 90%. If this is something that is really worth pursuing, and there's consensus that focusing on workload is more intuitive/useful than interval, then perhaps the "a 100 day interval will become 46 days" message could be reworked into something like "Reviews required: 217% of default"? Then the 100 case doesn't look silly. |
|
Dae, with all due respect, you make making progress very hard. Or a sequence of "Good" reviews So we have a problem and the simplest possible solution to that problem, and yet you still don't say, "Nice, let's merge this PR immediately". And don't get me started on Evaluate. Or on the Hard button misuse.
Really? That is your objection? Really? Do you not think that the current solution has a good "amount of work"/"problems solved" ratio?
Sadly, that would require simulations using the FSRS simulator, which is too slow to be practical. I thought of something like this when thinking of the desired retention overhaul, and yeah, unless we speed the simulator up by a factor of 50, it's not viable. |
|
Re: the extra logic
Is it simpler if it's a one-time message with a cross to close it? Just
spitballing for everyone.
Re: the workload message
It would be hard to get quite accurate. But we can show estimates like
"Expect 10 times more reviews", say when the user is setting 99% DR. Maybe
even refer the user to simulator to get more accurate stats?
About the 90% DR message. I think your suggestion was quite good: "To see
cards more frequently, increase this above 90%". Perhaps, this can become
(and what I previously had in mind): "To get more/less daily reviews,
increase/decrease this above 90%".
…On Mon, 19 May 2025, 06:14 Damien Elmes, ***@***.***> wrote:
*dae* left a comment (ankitects/anki#3995)
<#3995 (comment)>
I don't want to faff about with extra logic to show such a message only if
the value has never been changed before, as we have to track that then. I'm
not particularly fond of the current wording either, as I don't like how we
switch between quantitive and qualitative messages for only 90%. If this is
something that is really worth pursuing, and there's consensus that
focusing on workload is more intuitive/useful than interval, then perhaps
the "a 100 day interval will become 46 days" message could be reworked into
something like "Reviews required: 217%"? Then the 100 case doesn't look
silly.
—
Reply to this email directly, view it on GitHub
<#3995 (comment)>, or
unsubscribe
<https://github.com/notifications/unsubscribe-auth/BATHNXR4OU4NFET3KR65J6327ESP7AVCNFSM6AAAAAB435JNPGVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDQOBZGMZDKOBVHE>
.
You are receiving this because you commented.Message ID:
***@***.***>
|
|
Dae, if you want another idea, here: https://forums.ankiweb.net/t/desired-retention-ui-overhaul/57678/59?u=expertium Instead of showing intervals per buttons, we show a sequence of “Good” reviews and their intervals. But this is more complicated and will require more discussions than the simple fix in this PR. If you like this idea - great. If not - this PR offers the simplest possible solution. Actually, on second thought, while this would be an improvement overall, it wouldn't solve the specific problem of users not realizing that DR affects interval lengths if they have never changed DR. So we would still have to somehow give a nudge to people who always keep DR at the default value, which this approach with a sequence of "Good" does not. Completely unrelated, but Dae, please take a look at this and give Luc a green light: https://forums.ankiweb.net/t/estimated-total-knowledge-graph-over-time/57390?u=expertium |
You keep saying it's the "simplest solution" -- but that doesn't mean it's a good solution to that problem. I still don't see any indication that there's a segment of users who will understand this better by having this warning in the Deck Options screen (vs. somewhere else), or that the folks who are looking for solutions will find this more actionable by having it there. In my experience, many users simply ignore anything in a colored box in that screen 😅 -- just like they ignore the ❔'s staring them in the face when they have questions about something. |
|
NNgroup's eye tracking study comes to mind. Users ignore UX elements that
resemble ads, so boxes like these.
It's still better than nothing though.
…On Wed, 21 May 2025, 05:56 Danika-Dakika, ***@***.***> wrote:
*Danika-Dakika* left a comment (ankitects/anki#3995)
<#3995 (comment)>
There is a problem - users don't understand that DR affects intervals
...
So we have a problem and the simplest possible solution to that problem
You keep saying it's the "simplest solution" -- but that doesn't mean it's
a *good* solution to that problem. I still don't see any indication that
there's a segment of users who will understand this *better* by having
this warning *in the Deck Options screen* (vs. somewhere else), or that
the folks who are looking for solutions will find this more actionable by
having it there. In my experience, many users simply ignore anything in a
colored box in that screen 😅 -- just like they ignore the ❔'s staring them
in the face when they have *questions* about something.
—
Reply to this email directly, view it on GitHub
<#3995 (comment)>, or
unsubscribe
<https://github.com/notifications/unsubscribe-auth/BATHNXSX3BDDU5DGVA5M2F327PB4NAVCNFSM6AAAAAB435JNPGVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDQOJWGEZDIMJQHA>
.
You are receiving this because you commented.Message ID:
***@***.***>
|
Then we might as well give up, to be honest. We could rename "Desired retention" to "Desired retention (affects intervals)", but that's just a worse version of what this PR does. Btw, do you mean this one? |
|
@Danika-Dakika I've said this before, and I'll say it again: both here and on the forums, people don't agree on anything and don't understand that having a solution is better than having no solution. With UI there are often many ways to approach a problem, and many different better alternatives to the current implementation. But if nobody agrees on anything, then none of the better alternatives will be implemented. Is showing a blue box with "The higher the desired retention, the shorter the intervals between reviews." better than the current implementation? Yes. Just to clarify, I would be happy with any of the above, or with some other solution, but we have to choose something instead of playing the Internet version of tug-of-war for days and writing a hundred comments. |
Yes, probably. I read some shorter articles from their website about this.
I think dae agreed to add this but alas, then the issue was complicated too much. #3995 (comment) Just agree to dae's suggestion and you have something rather than nothing. |
But having a solution is not necessarily better and some "solutions" are worse than making no change. I know you get frustrated when people won't sign-on to your proposals, because they are obvious improvements as far as you are concerned. But you cast those who disagree with you as people who "don't understand" instead of letting them be people who disagree with you. Both here and on the forums, there are plenty of things that do get done -- often by consensus-building and compromise. They just might not be the things you pick to draw a line in the sand about.
I disagree with all of those "yes"es (some more strongly than others!). It's not because I don't understand, it's because I don't think they are overall benefits to the UI, or that they will do anything to solve the problem you've set your mind on solving. Not every proposal is an improvement, and I, like many, would rather see things stay how they are now, until there's a better solution. |
Do you think maybe it could be beneficial to compare this to the previously set desired retention instead of default. e.g. (just opened config, DR previously set to 80%) (change the value)
or, for the load: |
|
|
Ah, looks like your latest push already handled the first point - thanks Luc. If nobody else has something to add, I'll merge this in ~tomorrow morning. |
Co-authored-by: user1823 <92206575+user1823@users.noreply.github.com>
|
I noticed this message appears after clicking the Optimize button. Was that intended? |
|
Yep 03b42a9 |
|
@Luc-Mcgrady, but why? It seems weird for a popup to appear in a place other than the one I am interacting with. Also, would it make sense to show this popup for any user who has DR equal to the default (90%)? |
That seems annoying for people who genuinely want to set their DR to 90%. |
Keeps a warning active when the users desired retention 90% to try and hint at the correlation between desired retention and intervals for people who have not yet changed the value.
I'm not positive on what the message should be, suggestions welcome.
c.c. @Expertium