I seem to have increased reviews in FSRS, when people claim that it significantly decreased their reviews for the same output. I set my desired retention at 90%, the recommended, and optimized but now I’m spending almost 2x more time reviewing. (I followed the super short guide in the FSRS4anki github basically fsrs4anki/docs/tutorial.md at main · open-spaced-repetition/fsrs4anki · GitHub) I didn’t enable the option to change the card schedules all at once, I said to do it gradually, but still I’m getting more reviews. I’m also not changing much of my card adding habits. Any tips? My retention was ~80% before if that helps.
Well, there’s your answer. FSRS gives fewer reviews for the same retention, not fewer reviews for any possible value of retention. You can set desired retention to 80% and use FSRS for a month or two more, and see whether that helps.
1 Like
Is it bad if I use 80% since that’s the edge of the recommended? I just chose 90% because they said it would work for most people.
I just computed the minimum recommended retention and it’s telling me to increase it lol
Well, what do you care more about?
If you care more about verifying the statement that FSRS gives fewer reviews for the same level of retention, set desired retention to 80%.
If you care more about getting the best bang for your buck, in the “amount of stuff memorized divided by the time spent” sense, set desired retention to 93%.
Another way to look at it –
Why was your retention only 80% using the other algorithm? That’s a bit low.
(A) Because you were learning as much as you wanted to learn and you were happy with how well you were doing. – Then set your desired retention to 80%. You’ll get the same amount of knowledge for less work.
(B) Because the intervals were always too long, so you were forgetting more cards than you would like, and you weren’t happy, but there wasn’t much you could do about it. – Then set your desired retention to something higher than 80% (but maybe start at 82-85%). You’ll get more knowledge with a reasonable amount of additional work.
[Although – @Expertium what would cause someone with 80% retention to be told to increase their desired retention to an optimal 93%? That seems like an onerous increase in workload for it to be “optimal” for someone. Or is it because they are 10pts below their desired retention now, and would need a 93% to catch up? – or in the alternative, just disregard this question, because I’m not sure that I’m meant to understand the deeper parts of compute-optimal. 
]
what would cause someone with 80% retention to be told to increase their desired retention to an optimal 93%?
If you want me to point to specific factors that could cause this, I can’t really answer that.
Then I won’t feel bad not knowing! 
If anyone knows why that is happening please let me know because I really don’t know why it’s doing that and it makes me worried that I’m going lower when anki is telling me to go higher.
I think we’ve both told you the same thing. It’s up to you. Are you happy at 80% retention or do you want to have higher retention than that?
[To be clear, we’re talking about 80% on the day a given card is due. Before the card is due, you’ll likely have higher retention that that.]
There is really no way to know that. There are a lot of variables that go into deciding the workload:knowledge you will have for a particular desired retention value. For example, you might spend too much time with (re)learning cards compared to review cards. This means, if you decrease the desired retention your workload (and workload:knowledge) will rise quickly because of the higher number of (re)learning cards. If all things being equal, you were spending less time with (re)learning cards then the desired retention value CMRR (compute minimum…) gives you will probably be lesser.
This is only an example to illustrate how something might affect the feature and doesn’t mean I am giving you prescriptions for what you should be doing. Also, take this with a grain of salt because there might be some things I haven’t taken into account.
1 Like
“Why was your retention only 80% using the other algorithm? That’s a bit low.”
why you got to be condescending to everyone?
80% is not low for difficult material