Sorry if this has been asked multiple times. I read a few posts in the forum, and from my understanding, I’m supposed to update my parameters either once a month or when my reviews double. Then, I need to reschedule them using the FSRS Helper add-on.
Is that the correct approach?
Also, when looking at true retention, should I focus on the cumulative percentage, or should my mature card retention be close to my desired retention?
The manual (Statistics - Anki Manual) only mentions that you should look at monthly data. I would assume you’d check the cumulative percentage, rather than only the percentage for mature cards.
Nobody actually really precisely knows about this. “About once a month is probably good enough… in general” is… probably decent advice. Hitting it every single day is not going to hurt you… conversely… it probably is unlikely to help you out in any significant amount.
If you really care about getting the maximum efficiency and optimization and calibration, hit the “optimize all presets” button every day and then immediately afterwards hit the “reschedule all cards” in FSRS Helper add-on… but in 99+% of cases, this is likely to have less than 1% of an impact versus hitting it every now and then and not using such rescheduling, if even that much. To the point that the act of hitting the button itself is likely itself a source of decreasing your studying efficiency, so it might actually be worse… maybe… possibly.
No. You shouldn’t optimize every day. Monthly is enough. Or the semi-mathematical approach is – every time your number of countable reviews doubles.
There are downsides to optimizing too frequently.
Every time you optimize (and save) FSRS updates the memory state of every active card in your collection. If you sync with other devices, that then needs to be synced to AnkiWeb and to the rest of your devices. That means extra time for you to complete your sync, and unnecessary traffic on the servers of that wonderful (and free) resource that we all share.
If you sync between devices, hopefully you never miss a sync, because you could also be turning that tiny error into a huge sync conflict. [The same issue currently blocks implementation of automatic optimization.]
Hopefully you don’t use “reschedule on change,” because your database grows disproportionately whenever you do that.
And then – distracts you, keeps your focus on the program/algorithm instead of on the studying you need to do, wastes your time, etc.