anki-delay-siblings VS free-spaced-repetition-scheduler

Instead of messing around with ease factors, I'd suggest the Suspend siblings until mature add-on for new cards, and Delay siblings to spread out the current ones. That should pretty directly get what you actually want, with your siblings introduced at different points.

Once they're all in a single note, then I do more-or-less the opposite of your approach, by pushing the siblings as far apart as possible. I use Delay siblings for mature cards, and Push Paradox for new cards, although I'm strongly considering moving to the newer and maintained Suspend siblings until mature add-on.

Because they're all siblings, I can use Delay Siblings and Suspend siblings until mature to spread them out, so they don't hit all at once.

You can't make cards from separate notes treated as siblings. What you can do, though, is generate several cards from the same note. For instance, I modified my Basic note type to have 10 pairs of Front and Back (Front 2, Back 2, Front 3, Back 3, etc), and an Extra field to put on the backs of all the cards. That way, I can put a bunch of questions together in a single note and generate separate cards for them. Sibling burying, the Delay Siblings add-on, and the Push Paradox add-on help me spread them out further.

Rescheduling siblings like this addon does should be a built-in feature. Without this you can get a yard with a multiple-year interval, and then get the reverse card the next day. This doesn't accurately capture if you've learned this card because your memory was just refreshed the day before.

Why doesn't anki make something like this add-on a standard feature? https://ankiweb.net/shared/info/1369579727

Some cards contain information that can be found in others without necessarily being siblings. I want two separate cards to become siblings so that I can automatically delay one of them, by using an addon like this one. This would fix the problem of reviewing card A and then later in the same study session guessing another card B correctly just because card A contained related information.

Sorry if I'm being annoying, but there's one more thing - take a look at this add-on and see if it's compatible with FSRS.

... https://github.com/open-spaced-repetition/fsrs4anki/wiki/The... ...

I'm not sure I believe we understand our own learning/memory anything like enough for this not to be total pseudoscience? Reminds me of A Beautiful Mind.

Libraries: https://github.com/open-spaced-repetition/free-spaced-repeti...

All are MIT licensed I believe, Anki is primarily AGPL

Just read the wiki ¯\_(ツ)_/¯

The model used by FSRS: https://github.com/open-spaced-repetition/fsrs4anki/wiki/Free-Spaced-Repetition-Scheduler

Also, sorry for nitpicking, but I just checked the code here, and I saw that you changed the formula for post-lapse stability, but you didn't update the formula in the description here.

I have another one, also about difficulty. Look at the long "Stability updating formula after successful review" formula, there is a term D-b. This seems very counter-intuitive. A large value of D corresponds to an easy card, but thanks to that formula it will produce a very small change in stability. A small value of D corresponds to a difficult card, but according to that formula it will change stability a lot.

So I checked this and it seems that there's a whole bunch of different parameters (initial difficulty, initial stability, etc), and a lot of them are constant. So the next step would be to use some kind of optimization algorithm, like gradient descent, to optimize those parameters based on user's review history, right?