Is there any realistic hope that in 2025 or 2026 FSRS will be able to fully replace traditional manual learning steps in Anki? I’m wondering whether it’s even theoretically possible for the FSRS algorithm to handle early-stage learning intervals algorithmically—just like it currently does with review scheduling.
Right now, learning steps (like 10m 1d) are still static and need to be manually set. Could FSRS eventually make those dynamic too, adapting based on card difficulty and user performance? Or is this fundamentally outside what FSRS can handle due to limitations in its current structure or how Anki is built?
I’m working on FSRS that can handle same-day reviews properly. No idea when or even if it will be done. And even if it will be done, I don’t know how difficult it would be to make it work well in Anki, since right now Anki uses whole day (integers) instead of fractions. That part is up to Dae (main Anki dev) and Jarrett Ye (FSRS dev).
But does that mean it’s theoretically possible to have FSRS for short-term memory? Could learning steps eventually become a thing of the past at some point?
IF me and Jarrett figure out accurate formulas for FSRS and IF Jarrett and Dae change how Anki works so that FSRS can use fractional interval lengths instead of integers, then yes.