This message:
Can we update it according to new data from simulations? This was calculated using default parameters and .90 desired retention.
The default limits can also potentially change.
The ideal defualt review limit should be either 176 or 196. But I’m not sure which one. This depends on whether reviews here already include new cards or not. @L.M.Sherlock It’s your code, can you tell us?
It’s better than a fixed rule for sure, but the ratios depend on FSRS parameters. You should treat this information more like a cookbook than a law.
EDIT: theoretically, it’s possible to run a simulation every time the user changes parameters and/or desired retention to determine the ratio, but it would be too slow to be practical.
I seem to recall coming across users in the past with a review ratio close to 20x new cards. I’m not sure you can come up with a number that works for everyone, and the existing message was more about setting a bare minimum, not exactly matching the new card count.
I think a limit that is not a round number is going to trigger some users’ OCD.
say again? lol.
@dae, what do you say now? we are already running the simulations in latest beta so shouldn’t be too much of a hassle to use this here. we will have to round the numbers though.
@A_Blokee in case you’re interested, we are talking about making this more precise:
I feel like it’s wasted effort. Users will ignore it anyway, and it makes limit adjustment an expensive process.
in my understanding, we don’t need to run a simulation everytime limits are changed, we just need a rough estimate for review/new ratio for a particular DR/params and that can be done a single time.
It still needs to be calculated once each time you open the screen, or cached, which is complex. I don’t feel it’s the best use of our time.
Fair enough, what do you think about increasing the review limit though? Anki now has load balancer, so the occasional peaks are already handled by that. Review limit just hides from the user reviews that they should be doing otherwise. (Something like 400 is not too much for students).
The table posted above seems to imply that the current 10x estimate is fairly close to the standard desired retention. I was trying to target the common case, not the extremes. And ultimately, it’s intended as just a baseline/starting point - the user is still expected to adjust the limits to match their actual workload and available time.
There’s a psychological aspect here. I wager that the majority of users who pick up Anki are daunted by the prospect of even 200 cards/day at the start. My goal with the existing message was to give them a reasonable idea of what they can expect with the default 20 cards/day, not an upper limit that may be more daunting.
