FSRS - Remedy improper anki use

Anki FSRS
1

Hi

I am using ANKI for postgrad medical exams. I have a deck downloaded, which is basically taken the core textbook, and turned the cards into cloze’s, and I am covering the book for the first time this way (about 10000 cards to get through). This means that often when I see a new card, I don’t know the content. I know this is not the ideal use of anki - however, I have been using this deck for a few months and it has been ok for me.

I have just downloaded the update and I have finally made the switch to FSRS yesterday. My interval after learning a new card yesterday is 10m again or 8 days for good. 8 days feels like way too long. I have abused HARD cards before, so I have used the FSRS helper to remedy my misuse and I have changed the ignore cards reviewed before to yesterday so I am basically starting fresh (and I will make sure to use hard properly now having read into this).

My question is - given I use ANKI to learn content from scratch (ie, rather than going through the book and making notes which would be too time consuming as I work full time and the syllabus is extremely vast) - should I be hitting again for a new card I have not seen before or good? The interval given for a new card currently varies between 1d or 4d. Which means I either see the card again for the first time in 4 days unless I hit again, in which case it is going to be seen in 1d. I want FSRS to work for me because if it can reduce my workload in the long run that is great but I fear my improper use of anki (learning things from scratch rather than creating cards myself) may mean I have to revert to the old algorithm.

Screenshot 2025-05-24 at 15.18.47
Screenshot 2025-05-24 at 15.18.472762×1572 462 KB

2
1 Like
3

I think my main question is optimisation of parameters given my ANKI deck style is one which I have not created, and one I am learning from scratch. Hitting again for 75-100% of new unlearnt cards feels like a lot - is this going to screw the algorithm? Previously, I would hit good for the first time seeing a new card, review the card in 1d afterwards, then 4 days etc etc based on the older algorithm. Now it’s see new card > either hit again or good > see in 4-8 days during the first review (I don’t understand why this varies so much for a cards in the learning phase?).

4

You can only mess up the FSRS by using the wrong buttons. You don’t have to worry about the rest.

This is based on the data that you gave the FSRS. Either there is not enough data and FSRS has yet to adjust to you.

5

Are you sure that’s happening with these FSRS parameters and Desired Retention (DR) that you posted? That doesn’t sound right. [I assume your learning step is 10m?]

If you had a Hard mis-use problem [are you sure that you did?], and you used the Helper to fix that – then it’s fixed. You don’t want to also have FSRS ignore every review for every card you’ve already started studying. FSRS needs data to map your memory curve, and you’re taking that away. If you click Evaluate (or Optimize), FSRS will tell you how many reviews it’s counting – I suspect there’s not a lot left.

It doesn’t matter whether you’ve seen the card before – it only matters whether you get the answer right or wrong.

1 Like