Using FSRS, I have seen that some mature cards have an unusually high interval, that I am almost certain that I would get wrong once I see them in the future due to how long the interval is. This situation only happens with a small amount of my cards. It goes from something like a 3 month interval for hitting good to a 1 year or 2 year interval for hitting good. I don’t have the bad habit of marking cards as hard when I forget them, that the docs mentioned. Anyone know of a way to fix this situation, or am I just massively underestimating my memory and this is a normal situation?
I also saw some similar forum posts, but they either had this happening due to using hard when they should use again (I don’t do that) or only had this problem happen for new cards (my new cards have normal FSRS intervals)
Hmmm… It’s not about the “maturity” of the card – it’s about you grading it Easy the first time. Now, I’m thinking that someone should have looked at your parameters before giving you advice about your grading habits!
Your parameters are just a bit unusual. FSRS predicts that after you see a card the 1st time, you should study it next in a day or 2, regardless of whether you grade it Again, Hard, or Good – but if you grade it Easy the 1st time, you don’t need to see it for over a month. I guess that’s not very shocking if the material is very easy (which also explains why your Difficulty is so long.
But it doesn’t look like these are the parameters you had when this card was introduced. Did you recently re-optimize? Have you run “Evaluate” on the parameters?
Huh so I guess I should drop my habit of hitting easy for new cards, and instead just stick to good? Also are the cards I reviewed with easy just a lost cause then and I shouldn’t try to optimize their decks, and should instead just use default FSRS parameters? (which appear to give reasonable intervals, from my testing)
I was just using default FSRS parameters for a while, but eventually switched back to SM-2 algorithm by accident for quite a while. (accident like I set it up with the method where you paste in code for the scheduling, but then I forgot to keep clicking the “reschedule” button on the FSRS helper, and my cards didn’t appear to reschedule according to FSRS without pressing the button, so it was just like I was using SM-2)
So yeah this card was definitely not reviewed using these parameters I only switched from my accidental SM-2 algorithm usage to FSRS today.
My parameters changed by a very small amount compared to my other screen shot from before due to reviewing some cards, and then optimizing:
Should I still stick with it, even if I’m 99% sure that I will not remember that card on its new due date from FSRS for hitting a good? (and its the same pattern with some other cards)
Doesn’t it make more sense to just switch to good, because I was only using easy for new cards before due to the fact that SM-2 does show flashcards more than they need to be shown short-term?
I won’t try to select a grade by looking at the intervals.
How is that? The card previously shown was rated Easy the first time. If most cards similar to it were rated in the same way, that data won’t be used now for initial stability if you’re rating similar cards Good.
I suggest you read the “Technical explanation of FSRS” article and the linked “ABC of FSRS” here: https://expertium.github.io
