Let's make the True Retention table look better

Anki FSRS
43

Agreed! Young-Mature-All is enough – Let’s make the True Retention table look better - #36 by rossgb

44

As most people seem to be happy with the Young-Mature-All design I’ve opened a pull request:

6 Likes
45

This shows the problem with keeping multiple graphs for the same data.

(Apparently, reviews happening in filtered decks aren’t counted in “Answer Buttons” but it’s counted in “True Retention”)

1 Like
46

Should we remove “Answer Buttons” after all?

47

Probably makes only sense if anki is limited to 2 buttons for everyone, right? There are sections in the manual describing how often one should press which kind of button (for 4 button users). This graph shows it.

Personally, I’m against removing the graph.

1 Like
48

It still has nothing to do with improving the appearance and utility of the TR table.

Scope creep.

49

The way I understood it is that they are trying to find a better look and are discussion whether merging TR table and Answer Buttons Graph should be merged in order to achieve that.

(because there seems to be some data overlap)

1 Like
50

I’m approaching it from a project management perspective.

The Answer Buttons graph has been in Stats for years. It has been doing fine at being the Answer Buttons graph, even though one person asked a question about it today.

The TR table already existed, but was just added to Stats. It definitely needs to do better at being the TR table because it’s hard to read and understand (and most importantly ugly). When we narrowed our focus to that, we went from proposal to PR in ~30 hours.

Changing the TR table to be something else with different information, changing the Answer Buttons graph to be something else with different information, whether they need to continue to co-exist as a table and graph, and if so whether they need to match … – that’s all beyond the scope and shouldn’t hold up fixing the problem at hand.

[Also from a project management perspective, the decision whether to add the TR table to Stats in the first place would probably have been an ideal time to assess its impact on the Answer Buttons graph. But that’s target-gone-by …]

2 Likes
51

As someone who uses 4 buttons I would be sad if “Answer Buttons” was removed or changed to only show Pass/Fail.

I think it is fine to present the same information in different ways if you are trying to show different things.

For me as a 4 button user they serve different roles:

Answer Buttons

  • How difficult am I finding learning/young/mature cards over a certain time period?
  • Do I tend to find mature cards easier?
  • Do mature cards tend to be more Again/Good vs a spread of difficulties for young cards?

True Retention

  • How good is my retention over time?
  • Am I converging on my desired retention? (my day to day retention jumps around a bit)
  • Is the scheduler doing a good job estimating retention for both young and mature cards? (e.g. not hiding bad mature card retention with high young card retention)

Edit:

Sidenote

From looking at my “Answer Buttons” you could make an argument that there is a flaw in using retrieveability as the only metric in FSRS.

It schedules things so I remember them 90% of the time, but I am very rarely confident in my answers (81% Hard even in mature cards).

You would not be able to get this kind of insight if “Answer Buttons” were removed/changed.

3 Likes
52

Well, I intially thought we would include that extra data. There are ways to go about it including having an answer button breakdown for Pass rating in a pop-up.

53

This topic was automatically closed 30 days after the last reply. New replies are no longer allowed.