I really like the SRS work by , but not sold on the idea of storing all of the repetitions as blocks in the graph, and using time between them to calculate next repetition. Just realized another problem - this makes it very difficult to sync with other system?
Conversation
FWIW this is roughly the approach Orbit takes: a CRDT where repetitions are the units, with an associative+idempotent application function. This permits conflict-free syncing across many devices/clients which may have intermittent connectivity.
1
1
14
Do you plan to open-source your scheduler? I know it's not the most important thing to optimize in memory systems, but it could lower the barrier for other people (like me) to get started with their own systems. I spent a lot of useless time parsing Anki's description and code.
2
4
Yes, probably, though I think the API layer Lucca alludes to below is probably more interesting. twitter.com/lucca_dev/stat
My scheduler's extremely dumb! Quantum Country RCTs have been teaching me that in most cases, recall % is surprisingly insensitive to scheduling…
Quote Tweet
Replying to @adam_krivka @andy_matuschak and @houshuang
I think a good open-source scheduler would free Spaced Repetition from the siloes of single apps and allow the creation of now-unimaginable new use cases of Spaced Repetition.
By using web-tech this could be re-used in dozens of existing ecosystems.
twitter.com/lucca_dev/stat