Is it the user response behaviour
- on the single-card level
- on the deck level
- on the preset level
- on the level of the note type
- on the collection level?
1 Like
If you’re asking what FSRS uses to schedule an individual card for its next due date? [Is that the “retention time suggestion”? It might be easier to answer your question if you were using standard FSRS/Anki terminology.]
This makes the most sense – the grades given to each study of each card.
But if you’re asking what FSRS uses to optimize parameters for you?
This comes into play. The optimizer looks at all grades given to all cards in all decks assigned to the preset.
I am asking what FSRS uses to schedule an individual card for it next due date. Basically, what statistics does it keep? One for each card? On the preset level, does it save more information anywhere than what is in the parameters, which are defined on the preset level?
I asked for collection level because I sometimes hear that FSRS “learns how my memory works”. That can be interpreted as something on the highest possible Anki level.
I am particularly interested if it stores any information on the deck or preset level to set the due dates, because this is relevant for my strategy how I organize my decks (and presets).
FSRS uses (1) the parameters for the preset and (2) the card’s review history to determine the interval. [If you want to know how it does that, see: A technical explanation of FSRS.] Then it adjusts that interval using the Desired Retention (set at the preset or deck level). The next due date is today’s date plus that adjusted interval (+/- fuzz).
Outside of the review history that Anki has always kept – D, S, and R, for each card.
It learns by optimizing the parameters based on the information you give it. If you give it your whole collection – by putting all of your decks in a single preset – then it will use your whole collection.
1 Like
So i conclude
The actual answering history is considered on the card level only, no aggregates taken on higher level, so the history of the other cards in one deck/preset/card type does not influence the due date calculation.
The parameters and the desired percentage live on the preset level, and that of course goes into the due date calculation. But this is all that happens on the preset level.
So If a card is moved from one deck to the other with the same preset (or a preset with the same settings), the due date will remain the same, regardless of the behaviour of any other cards in those two presets affected.
Moving a card into a new deck and cloning the preset will not make a difference before the presets are optimized,
Those parameters are optimized based on the review history of all the (unsuspended) cards in the preset. So yes, the history of those other cards does influence the due date calculation.
When you move a card, the due data date remains the same – whether you move the card into a different preset or not – because that action doesn’t reschedule the card. The card will be scheduled based on the preset it is in when you study it next.
1 Like
When you move a card, the due data remains the same – whether you move the card into a different preset or not – because that action doesn’t reschedule the card.
Well, my experience is different.
Here I have a card in a 83%-preset deck.
This are the suggested due delays:
Now I moved it to a 70%-preset deck. I got this
And when I moved it back, the original suggested due days came back.
1 Like
Yes, but that’s not the current due date. That’s the result of the calculations for what the next interval/due date will be.
The current due date – before you complete this review – is shown at the top of the Card Info you posted. That didn’t change when you moved the card, did it?
1 Like
This topic was automatically closed 30 days after the last reply. New replies are no longer allowed.