I have raised this before in GitHub. I think there are two issues with how the retrievability graph works now:
Lost functionality. If I search is:suspended the graph doesn’t display anything at all. The only way to see retrievability graph of suspended cards is to maybe first flag + unsuspended. Then undo these actions.
It’s not consistent within the stats screen. This is not how everything else is handled.
I get it’s more consistent with how this is handled in add-on, but add-on users are a tiny minority of Anki userbase. And for a whole year, no one really complained about this.
@L.M.Sherlock changed this because someone noticed the difference in their total knowledge in the add-on versus in Anki. I want to ask, if it’s not more practical to just remove this from the add-on as Anki already has this as native functionality.
Well, retreivability doesn’t include suspended cards anymore, so probably the stats related to that. Do they even serve any function now once available as native functionality?
If dae wants things be consistent with the helper add-on, perhaps it’s better done by removing the functionality from add-on so we are able to also have consistency within the built-in stats.
OK, I will remove “Average stability”, “Average predicted retrievability” and “Estimated total knowledge” from FSRS Stats because they have been provided by the built-in Stats.
Jarrett had removed the add-on stats so there is no worry about compatibility.
First things first, let’s make it consistent with the add-on. If you’re proposing further changes to both, that’s perhaps better discussed separately.
Also, my reasoning is this would allow everyone to get the full benefit of that screen as you can just put -is:suspended in the search bar if you don’t want that included in stats.
But – should future-looking graphs include suspended cards? As long as we’re considering it – let’s consider it! [I’m not trying to propose a radical change, just some careful reflection.]
It makes sense for anything retrospective/historical to include suspended cards [Calendar, Reviews, Card Counts (explicitly), Hourly Breakdown, Answer Buttons, Added] – but once an active Review card is suspended, it seems reasonable to expect it to drop off of the other graphs. Especially for something like Retrievability, which isn’t static.
I go to the trouble to move most of my suspended Reviews out of my main study deck, to avoid them dragging my stats down like the dead weight they are.
It seems like the most typical use case for a suspended Review card is – I was studying that card, but I’m not going to study it anymore / for the foreseeable future. What’s the purpose in continuing to track those dead-ended cards in prospective graphs that are about the health and outlook of a collection?
I’ll note down that “Review Intervals”, “Card Stability”, “Card difficulty” would also be future-looking graph but they include is:suspended.
Eh, but you can just put -is:suspended in the search instead of doing this, right?
I’m thinking it probably doesn’t matter that if you’re studying for fun or you’re studying something with an open-ended syllabus. But in some of my decks where I’m studying for a particular exam it becomes important to look at stats for things I’m not actively learning.
And also, it’s just weird even for my word mining deck that it doesn’t include stats for the suspended cards, especially in estimated total knowledge as that makes the stat completely wrong. The reason I don’t delete any of the cards is because I want to know the number of words I know in a language and the stats don’t include it.
I have the option of unsuspending the suspended cards to look at their stat but it makes things complicated for me.
I don’t have strong feelings here. Given that the change was only made for consistency reasons, if consistency is no longer a concern and nobody has complained before or after it was changed, reverting to the old behaviour seems like it’s not going to be very disruptive.
