I’m using Anki to learn foreign vocabulary (with basic and reversed cards) and I’m looking for the sweet spot where the number of new cards added everyday wouldn’t affect the number of reviews in the long term. as the workload decreases with time if no new cards are added, I just want to add enough new cards to stabilize the workload.
is there a formula for this ?
my desired retention rate is 86% and my actual retention rate is 88-90%. the number of cards in review is 10000 (5000 words).
the FSRS simulator never worked on my machine.
actually if you stop adding new cards, the workload decreases with time. so the idea is to use that extra space to add neew words to keep the workload at the same level. so far, I couldn’t find the perfect number and I tend to add too many new cards which increases the workload considerably (but not immediately which is treacherous).
no matter what deck I choose, the graph always looks very suspicious. for example, the number of projected reviews always increases in the first few weeks (and there’s absolutely no reason for that, my workload keeps decreasing gradually). and the new cards/day setting never worked.
Assuming that intervals of learned cards increase, roughly, exponentially, if you keep learning same amount of cards each day, the number of reviews per day will grow logarithmically over time. Meaning that, no matter the actual rate, eventually it will surpass any fixed limit.
If the goal is to keep the overall daily studies (reviews + new cards) constant, the number of cards learned per day will have to be decreased with time, asymptotically tending towards zero.
In practice, instead of constantly recalculating the current target number for the new cards, the same goal can be achieved by setting the overall limit on the reviews in the deck’s settings (and toggling New cards ignore review limitoff). If the limit on the new cards/day is high enough and the new cards are set to be shown after the reviews, Anki will show you exactly as many new cards each day as is required to hit your goal. As the daily reviews accumulate, this will result in the average number of daily learned cards automatically becoming lower, following the theoretical asymptotic decrease.
Interesting note about the mathematics @Eltaurus it sounds reasonable, but I would need to look into it before I’m convinced. In my personal experience it seems to stabilize. In my case around 10 new + 100 reviews per day. Maybe its the fact I delete a note every now and then, idk.
I imagine it depends on personal FSRS settings, and what one is studying.
Logarithm is a rather slow-growing function. You need much more data than this to clearly see the trends behind all the noise real-life data has due to the stockastic nature of the process.
