When using FSRS, how different should each deck be to warrant different optimization parameters for each. For example I have a variety of misc decks from various subjects (biology, chemistry, trivia, pixel art, world cultures etc) they don’t have a ton of cards in each one individually, as they were just passing interests, and have similar-ish retention rates (a difference of 5%-10% ish for each deck’s retention), so would it be fine to just have the same parameters for all or should I optimize each one on their own with separates parameters?
There’s no precise rule for this. Mostly, you’re looking for a big difference in how difficult the cards are for you to answer – sometimes that reveals itself in your retention levels, sometimes it doesn’t.
Just remember you can Edit > Undo your way out of whatever changes you make. My trick – right before I start experimenting, I sync. In case of catastrophe, I can force a one-way sync and download that version. But it has the added bonus of resetting the Undo queue, so I can Undo everything, and not have to worry about going to far.
I later realised a better method is to go to your deck, clone the preset, save/save to all subdecks, then do an optimisation on the new preset. If the deck size is small, and you don’t get better results, simply undo all the actions.
Try the method here. But if you don’t have too many reviews in each deck, you’re probably not going find value in presets for each deck.