Does pressing forget on a card delete its entire review history and treat it as a completely new card? If so, why does the stats page still show the old reviews?
(I pressed forget 5 days ago and started from scratch).
So how can I remove this? Would this affect my FSRS optimization? Because this is the very reason why I pressed forget in the first place, as I did a very dumb thing before which made FSRS think the cards are much, much easier than they are.
And if it is not removable without deleting the cards entirely, how can I limit the reviews used by FSRS for optimization (i.e. ignore the bad reviews)?
Yes, reviews of forgotten cards are used for optimization. Don’t ask why.
You can suspend them, but I assume that’s not what you want. You could tag them and then add -tag:yourtag to the optimizer field or something like that, perhaps.
Yeah, but the problem is every card in the deck (along with its subdecks) has this problem. The deck has its own preset. Should I just the preset of a different normal non-damaged deck?
If I am not mistaken, the feature where you can select the timeframe from which the FSRS could use the reviews for optimization is already a suggestion that has been made. It would help a lot in my case!!
No, they are neither used for optimization nor for scheduling. They are just shown in Stats so that the user can see how many reviews they did in the past, how was their retention, etc.
1 Like
If this is true, this would save me a big headache! Thank you!
They are used for optimization. Select all cards in a specific deck, click “Forget” on all of them, then run the optimizer and see the number of reviews for yourself. Please do not spread misinformation.
Then, it is a bug where the revlogs before Forget are not being ignored if Forget is the last revlog entry. @L.M.Sherlock
@DerIshmaelite, as a workaround, you can add -is:new to the search field when optimizing. For example, preset:"Your Preset" -is:suspended -is:new
1 Like
Does this combo make Anki ignore everything before having pressed Forget?
What about making an entirely new preset for that deck?
Yes
It won’t help.
1 Like
Thank you! I tried it and I suspected, it was taking forgotten reviews into the optimsation. It now says, less than 1000 reviews meaning the optimisation cannot proceed. It did something.
1 Like
As I think I mentioned recently, the count shown there is the count before any filtering is applied. It is only provided to give you a rough idea of how many revlogs your search is matching.
So they aren’t used for optimization and @vaibhav was correct? Then the number shown during optimization is very misleading. We definitely do not want the user to believe that he has thousands of reviews when he has 0 (after using Forget).
The problem is not only about the count. Even if you Forget all the cards, the optimizer does generate new parameters.
The issue is that the filtering of the reviews before the Forget rating is currently based on the code that looks for the first Learn entry in the last group of Learn entries.
But, if Forget is the last entry, the above code doesn’t work.
Simply excluding new cards from optimization will also not work in the case where users introduce new cards using Set Due Date.
So, we need to explicitly filter out the entries up to the Forget entry.
2 Likes
If Forget is the last entry, the card will be new, and memory state is not assigned to new cards. I’ve logged the other issues on the following tickets. Please feel free to add your own if you notice other problems.
I still do not understand. Sorry for being slow 
. Are the old reviews before the forget entry still taken into account for the FSRS optimizer?
It will be fixed in the next release.
1 Like