Oh, it does look like the new stability is getting clamped to the old stability, so it is staying the same rather than increasing. (I thought it was increasing because I was accidentally looking at a change in interval rather than a change in stabilty, and because the formula said it should increase.)
This still feels like a regression compared to older optimizer versions, with post-lapse stability always at 100% of the prior stability, where previously it was 50–60%. (But maybe my memory really works like that?)
Note: My original numbers were based on a preset containing three decks. The graph above is based on the largest of those decks, which represents about 90% of the reviews for that preset. (I couldn’t just use my entire collection, since it contains decks from other presets that aren’t relevant here. However, I could create a temporary collection with just the single preset, if it matters.)
The blue scatter points show that your PLS is near the prior stability when the prior stability is small. But the orange line doesn’t fit well, too. I haven’t any idea to solve it now. As a workaround, I recommend decrease 0.8895 to ~0.5.
@dae I have a suggestion: don’t hide the “Optimize” button. Optimizing parameters is one of the most important aspects of FSRS, and we need to ensure that it’s impossible to miss. “Compute optimal retention” is optional, so it’s fine if it’s hidden. The same is true for “Custom scheduling”. But optimizing parameters should be in plain sight.
I added a field to one of my note types, getting the usual warning that a full one-way sync would be required. The Sync menu item turned red.
When I click on the red Sync menu item:
The “Added/modified 0 0” box briefly flashes on the screen and disappears.
When I click on Sync again (and again):
A “Checking…” box with horizontal progress bar briefly flashes on the screen along with the “Your decks here and on AnkiWeb differ…” dialog box underneath it, and both disappear.
After exiting from Anki and restarting (note: in Tools / Preferences / Syncing I have “Automatically sync” turned off), the same behavior repeats. The first attempt to sync shows one box, all subsequent attempts to sync show the other pair of boxes.
The same problem happened in 24.04beta1 = bfaa28f
After downgrading to 23.12.1, the one-way sync succeeds normally.
As usual, we have a tradeoff here between optimizing for new users, and for regulars. Presumably, regular users will prefer optimizing all at once from the top, so if this section is permanently expanded, it clutters the other option somewhat. But it’s not something I feel strongly about.
I actually think it would be good for regulars as well, at least for those who don’t have a lot of presets. If a user often uses Optimize/Evaluate, he might become annoyed that he has to unfold this section every time.
In the next rc “Compute optimal retention” will be reworked. Previously, it maximized the sum of all retrievability values. Now it will minimize workload (in minutes of studying)/that sum. More details here: The Optimal Retention · open-spaced-repetition/fsrs4anki Wiki · GitHub
I’m worried that this feature is unintuitive and hard to understand for many users. So I want to ask people for suggestions on how to make it more clear. This is what I proposed:
Another user said that if we add a text annotation for everything, the UI will be very cluttered. I agree, but I also think that “Compute optimal retention” needs special treatment because it’s a rather hard-to-understand feature even among FSRS-related features, which already cause lots of confusion among users.
Anyway, any suggestions on how to make it clear what “Compute optimal retention” is doing (and ideally in a short and concise way) are welcome.
What will it do? The previous algorithms didn’t consider if you introduce enough new cards to look at all of them. It can’t be so with just “Days to simulate”.
Custom study extends the limits by adjusting the record of how many cards have been studied, so the changes will continue to apply after you make changes to the limits.
@dae Is the new build on Mac coming anytime soon? I have issues with shapes on IO with new zooming feature. This feature makes my workflow easier but this issue is pulling me down. So if new RC would take time then I’ll downgrade for the time being. Also is this feature also coming on iOS beta as well?
[Edited to give a simple one-paragraph description]
Using 24.04rc2 (b84fce4) with Qt6 on Windows 11 with system clock using a North American time zone that observes Daylight Saving Time (and the bug also occurred in version 23.12.1)
Whenever the current time of day falls precisely within the one-hour time interval immediately after Anki’s “next day starts at” changeover hour (by default 4:00 am to 5:00 am), then when clicking Stats to open the Statistics window for any deck, in the Calendar section the little square for March 10, 2024 (which was the Daylight Saving Time changeover date in North America, namely the second Sunday in March) displays as white and a mouseover over it shows 0 reviews. At any other time of day, that square is blue as expected and mouseover shows the actual (non-zero) number of reviews.
The little squares for the Daylight Saving Time changeover dates of previous years (2023 and earlier) do not show this behavior either for the March switch to daylight time or the November switch back to standard time.
PS,
The color change from very dark blue squares to lighter blue squares between the two images is associated with switching between “last 12 months” and “all history” radio buttons.
