Clarify what optimal retention means

Anki FSRS
21

I don’t think I’ll be using that.

■ New 0 0%
■ Learning 0 0%
■ Relearning 0 0%
■ Young 14 5.28%
■ Mature 245 92.45%
■ Suspended 6 2.26%
■ Buried 0 0%
Total 265

Desired retention 0.96

How many days should I simulate?

26 = 0.76
365 = 0.86

22

this is going over my head…

23

Came here after trying to figure out optimal retention myself. A few points, just to add to the discussion:

  • Having only one custom parameter as opposed to three is a big improvement. I never touched optimal retention before because it seemed too confusing. :clap:
  • Days to simulate is still not clear at all: :man_shrugging:t2:
    • Even though I looked, I did not find the info/helper box before I came here. (Wasn’t there small (i) symbols before? It did not appear to me to click the text on the left, far removed from the actual field I can change.)
    • If the wording will ever be changed, I propose the following:
      • Change Days to simulate to refer to the future (I thought it was how far back it considered my deck learning history): Future timeframe to simulate (in days) or similar.
      • In the long helper text, first answer the question of what the only number you can change means and how they should treat it. That’s why people click the helper text (if they find it), right? Then tell them what to do with the result. E.g.
"`Future timeframe to simulate (in days)` determines how far in the future you will simulate. For long-term decks, use a large duration (e.g., 10 years), for short-term decks, use a shorter one. (...followed by general explanation)"
  • In general, the helper text seems too technical and not intuitive enough.
  • The last sentence in the helper text is very helpful: Setting your desired retention lower than the optimum is not recommended, as it will lead to more work without benefit.
3 Likes
24

To read the info tip about “Compute optimal retention”, you can click here:

image
image700×91 2.94 KB

Granted, I can’t imagine someone thinking, “Hmmm, I don’t know what this does, let’s click this inconspicuous symbol that isn’t anywhere near ‘Compute optimal retention’”. As for clicking on the text (which also shows the info tip), I’m fairly confident that the majority of Anki users are unaware of such functionality. I wouldn’t be surprised if most users have never seen the info tips.

I don’t think “Days to simulate” needs to be renamed, it probably wouldn’t make it more intuitive. I don’t have a good idea how to make this feature intuitive, which is why I made this post.

1 Like
25

That naming is just more confusing.

26

I think you can’t really hope to make everything very intuitive for users. Some of it will take figuring out. When I started using Anki I read the manual first and then what https://refold.la had written about Anki and then watched MattvsJapan’s excellent video on Anki Optimisation.

I don’t know if most people start out with community guides and youtube tutorials but a lot of us do. And yes you are right a lot of people might not even consider pressing that top-right button on encountering something they don’t know. Instead they might do a search on Google or YouTube. Well my point is we need not have to spoonfeed every important piece of information. People can figure out stuff on their own if the info is easily Google-able.

1 Like
27

That “Days to simulate” is somewhat unclear to me. I’m getting slightly higher optimal retention for a deck when I put a number >365 in “Days to Simulate”. That’s more accurate? Or that depends on whether I continue using this deck for that amount of time?

28

The main information that is missing from the current variable name:

  • That we’re considering days into the future.

This is not written anywhere and not self-explaining. I thought it was looking back and deciding how much of review history to take into account. Now it’s easy to understand when you already know what it means. But I think a new variable name needs to make this clear.

29

I wouldn’t really say “future” or “past” in this case. It’s not simulating what will happen with your current cards in your current decks, it’s simulating a review history that is similar (in some mathematical ways) to your real review history, with cards that are similar to your actual cards, but not literally based on them.

30

We could rename it to “Minimum recommended retention” to make it more clear that users can set their desired retention above this value but shouldn’t set it below. So far this seems like the only change that could make things clearer.
@L.M.Sherlock @Danika_Dakika thoughts on “Minimum recommended retention”? I think someone suggested this before, but I can’t remember who or where

2 Likes
31

It’s suggested by me, in discord.

2 Likes
32

I like that better as a name – it’s a bit clearer how you would use it.

33

But isn’t that number the “optimal” retention for which workload/knowledge is minimum? Minimum has the connotation of “Anything above it is just as fine as this number” (Think Minumum Wage, Minimum Retail Price, Minimum Support Price). But I maybe countering my own point that “Everything doesn’t need to be explained to the user”.

34

It still is a minimum recommended retention:

  • Going lower makes no sense: You would study more for less effect (lower efficiency, lower absolute retention).
  • Going higher is a reasonable choice: You trade efficiency (i.e., you a have lower knowledge/time ration) for higher absolute retention.
3 Likes
35

@dae seems like renaming “Optimal retention” to “Minimum recommended retention” is considered a good idea, so I would suggest renaming it. I’ll close this issue after that, since I can’t think of any other way to make this feature more intuitive.

1 Like
36

Logged on Clarify what optimal retention means · Issue #3118 · ankitects/anki · GitHub

37

@dae This reminds me that Historical Retention has the second r in caps. I noticed this is stylistically incoherent as the first letter in this position isn’t normally capitalised in Anki. Other users probably do not mind this but can we still get that fixed?

38

Thanks for catching it!

1 Like
39

By the way, I’m curious if we can get the data for how much “knowledge” (as defined in FSRS) we gain per hour etc. It would be intersteting for me to compare this with the data on how much vocab I learn by reading to see what is more efficient method— aiming for higher retention or extensive/intensive reading.

1 Like
40

Interesting, this could be a stat in the Helper add-on. I made an issue: [Feature Request] Knowledge acquisition per hour stat · Issue #382 · open-spaced-repetition/fsrs4anki-helper · GitHub

1 Like