Hey all, hope you’re doing well. I’ve encountered a (potential) problem in Anki 25.09, when studying sub-decks with different desired retentions under a single parent deck.
So I have set 4 decks under different desired retentions based on the “High Yield” tag on Anki, all with the same preset (given that in 25.09, you can now change deck’s desired retentions). So, my retention looks like 90% for one, 85% for the other, 82% for another, and 80% for the last. It looks like this
However, when studying over the parent deck that I’ve put them, instead of before (I had them under different presets each), where it would go from cards with under 90%, then under 85%, etc., now it will for some reason have cards that are 78%, 76%, etc. come before the ones that are 89%, 87%
Is this a feature or a bug? Thanks so much for any help
It is a feature. Anki calculates how much R deviates from the desired retention and then sorts them accordingly.
I know that the name and description of the option are slightly misleading. But, it is difficult to describe what Anki actually does in the name of the sort order.
Why did we decide to do that? Personally, I’d much rather have it do what it says, sort by descending retrievability straight up, not some convoluted reverse version of the old “Relative Overdueness” sort.
If I have a subdeck set to a DR of 0.95, that’s probably because I want to prioritize those, but now those cards lose their Retrievability the fastest and get buried faster if I don’t study them on the day they’re due.
Funnily enough, it was you who requested it first.
However, I get your point.
The “Ascending R” sort is basically an “urgency sort”. Here, Anki is sorting by (Elapsed Time / Interval) (the actual formula is different but they are equivalent) in descending order. This prioritizes cards that are the most “overdue” relative to their scheduled interval. This doesn’t have the problem you mentioned because it prioritizes cards that have fallen furthest from their Desired Retention (DR). So, it naturally prioritizes high-DR cards.
The “Descending R” sort is basically a “freshness sort”. The goal here is to prioritize the cards that are “fresher” — have their current R closer to the DR.
Simply reversing the Urgency Sort produces the problem that you mentioned. However, sorting purely by R would cause Anki to ignore low DR cards for too long — until all the high DR cards are reviewed.
So, I am not sure what’s the best way to solve the issue.
PS:
Personally, I use the ascending R sort order when faced with a backlog.
My understanding at that time was that relative overdueness was just a straight ascending retrievability sort, so the reverse would be a direct descending retrievability, and that’s why I worded this post as a “convoluted reverse version of the old [SM-2] “Relative Overdueness” sort.” I never would have wanted it to work the way it is now, and I didn’t anticipate being able to give different subdecks different DRs.
I mean, and I know I don’t speak for everyone, but that’s exactly what I would want. If I give one subdeck a higher DR than another, that means I wanna prioritize those cards. Also, cards lose their R faster at the top of the curve, so if I miss a day, those high DR cards are going to get buried.
It’s also just bad design/confusing. The sort is labeled “Descending Retrievability” and that’s not what it’s doing.
