Add a 5th grading button

Anki Suggestions
1

The proposed 5th button would be in between Again and Hard. It would be called “Almost”
For example, this button is meant to be a weaker version of again.

The usages for this include, when you get a card that asks something, you answer incorrectly, but your answer was very close. For example in a medical context:
What bacteria causes this disease?

Hard/Good/Easy: Staphylococcus Aureus
Almost: Staphylococcus Epidermidis
Again: Bacillus Anthracis or No answer

I believe, that if we have 3 correct answer buttons to differentiate with, then we should have another for incorrect answers. In my opinion, I believe there is a huge difference between not knowing the answer at all and being very close to it.

I propose that the almost answer button would require the re-learning step, same as the again, just it has weakened “consequences” compared to again. Such as, if again changes difficulty by 30%, perhaps almost can do it by 15% instead.

While I do agree, more buttons can cause answer paralysis, this can be solved by enabling pass/fail buttons.

2

I know (from the Discord server) that you’re very committed to this idea. This is the first time I’ve seen you explain it in any detail. I’m not trying to poke the bear, but if you want it to be seriously considered – let’s flesh it out.

[I say all of this with the knowledge that the battle right now is between whether 4-button or 2-button grading will be default – so the likelihood of a 5-button option is very slim.]

  1. If you find that close answers are acceptable in real life – when you use this knowledge for whatever purpose – doesn’t the Hard button already satisfy that need? In that “close enough, but needs improvement” situation, you wouldn’t want to lapse the card.

[Because if a close answer isn’t acceptable in real life – that should clearly be graded Again, right?]

  1. Are you proposing this for SM-2 or FSRS? It would need different formulas/calculations for each algorithm, wouldn’t it?

  1. So the proposal is 5-button or 2-button grading? And 4-button would no longer be an option (without an add-on)?

  2. Aside from your belief that these cases (Again and “Almost”) should be treated differently, is there any basis in brain/memory/spaced-repetition science (or any analysis of Anki-user data) that suggests these cases should be treated differently?

  3. Have you considered building this (or asking an interested dev to build this) as an add-on first? That seems like it might be a necessary step to work out how this button would integrate with each algorithm (in terms of formulas, optimization, and what the presence/absence of those grades means) – not to mention how it would fit into the database/review log.

1 Like
3

  1. The issue with using the hard button for a close answer, is that hard is for correct answers. A close answer is incorrect. A recent incident in which a close answer would have helped me a lot is when I was studying my Greek deck. I was asked to translate “camp” into Greek. The word for camp in greek is κατασκήνωση (kataskinosi). I last saw this card 3 months ago and it’s interval slowly increased from a few days, to a month, two months and finally now 3 months. I did answer it correctly, because I said instead of “κατασκήνησω” (kataskiniso), which is a word that does not exist of course. Most Greek speakers would understand the error I made. I would be able to recognize that this word means “camp” if it was an article, or if it was my reversed note type card. I made a very close mistake by swapping around the last two errors.
    The issue is with again, now it treats the card so strictly. I haven’t gotten this card wrong yet I see it more often than I personally believe I need to. If there was a “close” answer button, perhaps I would see the card a third as less, or even half as less.

    This is not the only scenario that this has occurred in.

    1. I am proposing it for FSRS only.

    2. I was attempting to say that the answer paralysis problem is a bit of a divided issue. A lot of users I know have no issue at all with 4-button grading, a few I’ve asked have no issue with 5-button grading. Those who have issues with 4-button grading have opted into using the pass/fail buttons aka 2-button grading. So I believe, if one was to have an issue with answer paralysis regarding 5 buttons, they would also have issues with 4 buttons, which in either way, they would enable the pass/fail buttons.

    3. I do not have any basis at all or analysis, because I am not sure how one could objectify this to find data. I am very open in attempting to look at data or anything, but I am not sure how I could find it.

    4. I used an AI agent to help me code it as an addon. It is very rudimentary, but it works in this way.

    For an example, a card has 30% difficulty. If I were to hit again, the difficulty would increase by 30%, going to 60% difficulty. The close button would increase the difficulty by half of the again button, so it would increase by 15%, causing difficulty to go up to 45%. In the review log it just acts as I had pressed again.

4

I wonder if their idea was inspired by jpdb.io, which uses a system exactly like this (except it also has “never forget” so technically 6 buttons). It’s not a science, but jpdb.io is well-regarded in the Japanese learning community as having one of the best algorithms, and it’s sometimes mentioned alongside Anki as the better alternative algorithm-wise. I’d say Japanese learning community is the most tech-savvy and resourceful out of all language learning communities so I trust there’s some truth to that judgment. I don’t know which algo it’s based on but I don’t think it’s SM-2.

But like you, I’m not hopeful this suggestion will take off given the Anki community’s fervor for 2-button grading. I just know I’ll definitely use the 5th button myself. I see decision fatigue as a bell curve: It is the most difficult to decide when my heart feels like the result sits in the middle of two available options. Decreasing the options is one way to help, but many people overlook that more options also helps (fewer “in the middle” situations). With jpdb.io I’ve never had trouble deciding which button I should click, because my heart always aligns with one of the available options.

1 Like
5

Just to mention that the original SM-2 algorithm (and to date), failure included 3 grades. Null, Bad & Fail, being the null not present in the UI, as its use should be sparing. The difference between failure grades is negligible in the actual versions of the algorithm. It is closer to being a qualitative classification for the user.

6

as someone who often has “almost” cards and who understands the frustration, i know where you’re coming from and i even jokingly once mentioned that this button should exist, but in reality i don’t think there’s actually much to be gained from such a button. while it’s true that you’re technically grading the card harsher than it deserves, but in practice the card should quite quickly converge to reality anyway. i think that the difference between again and almost will be barely noticeable with FSRS, and should theoretically get even less noticeable over time.

1 Like