Spaced repetition works: you review a fact just before you would forget it, and each successful review pushes the next one further out. The failure mode is not the method, it is the backlog. Miss a week and you come back to three hundred due cards.
I do not tune the algorithm. The defaults are roughly:
| Review | Next interval |
|---|---|
| 1st | 1 day |
| 2nd | 3 days |
| 3rd | about a week |
| after that | previous interval × ease (≈ 2.5) |
The aim is not to remember everything. It is to stop forgetting the things I chose.