I'm getting lots of suspended cards in Anki, so I think I should reconsider how I score the cards.

How do you score your Anki cards?

I'm not following on how rating cards wrong would make them suspended, but any of the hard/good/easy options should really only be picked if you actually got the card right. If there's something you got wrong -> fail option. That's how it's designed.

edit: Is this what you're talking about? I'm confused.

TwoMoreCharacters Wrote:I'm not following on how rating cards wrong would make them suspended

He is probably referring to the leeching options in Anki. Where if you fail a card a set number of times Anki will mark the card as a leech and suspend a card. This is actually a good thing, as it helps you focus your time on what your getting right rather then what your failing.

TwoMoreCharacters, Anki is more for drilling facts you already know, not for learning new facts. If your reviewing facts that you have memorized outside of your SRS, you probably wouldn't be failing them as often.

Or perhaps your just grading yourself too hard, like If I miss a single stroke but was thinking the right idea, I'll mark a card hard rather then fail it.

Another thing you could try doing is to cram new cards with Anki, one way is the built in cramming feature (which I think probably isn't as useful as it used to be for this) or just by duplicating your deck and setting cramming like intervals into that deck.

@Daichi

I disagree with part of your comment. To repeat my past arguments, I believe Anki should be seen more as a HUD rather than a static vessel for depositing artefacts. It's the perfect environment/perceptual overlay for actively, continuously learning an item or items, given the customization for designing the card.

Usually when I first add/unsuspend a card, I study it, rather than retrieve it, e.g. I'm learning it/encoding the information before setting in motion the adaptive spaced retrieval. You could, of course, have encountered that information in a different context. There's all sorts of ways to integrate. Explicit study before another, more implicit context, or afterwards, etc.

Also, feedback with or without failure (e.g. flipping the card and studying that information) improves spaced retrieval results. Feedback is good in general after retrieval or when you fail and restudy (which presents a similar situation to a new card), but also when it's corrective feedback, when you make memory errors or have low confidence in certain areas over time. When you execute a successful retrieval you temporarily enter a reconsolidation state where the memory is more plastic, so you can strengthen the encoding again, make adjustments.

Also, when Karpicke and folks talk about spaced retrieval's effectiveness, they're conceptualizing the process of spaced retrieval (study/testing trials, encoding/recall) itself as a learning phase prior to some future evaluation (in the context of school something like exams at the end of a semester, or in research a final evaluation of effectiveness [hence ‘retention interval’]) but for lifelong study, it's sort of an infinite learning phase of that information (with ever expanding intervals).

“... research in cognitive science has challenged the assumption that retrieval is neutral and uninfluential in the learning process (7–11). Not only does retrieval produce learning, but a retrieval event may actually represent a more powerful learning activity than an encoding event. This research suggests a conceptualization of mind and learning that is different from one in which encoding places knowledge in memory and retrieval simply accesses that stored knowledge. Because each act of retrieval changes memory, the act of reconstructing knowledge must be considered essential to the process of learning.” (Karpicke & Blunt, 2011)

Daichi Wrote:Or perhaps your just grading yourself too hard, like If I miss a single stroke but was thinking the right idea, I'll mark a card hard rather then fail it.

I actually do that too quite often, unless the interval is too large.

Like the other day when "暖" came up and I completely forgot about that damn extra stroke above 友. It's not like I forgot the primitives or wouldn't be able to recognize the character if I saw it etc, but if I simply had gone with hard right then, it would have been scheduled for more than 7 months in the future and I would by no chance in hell have remembered that extra stroke by then. So I went with fail for such a simple character.

I'm not dying in reviews because of this though, with such a simple mistake like this I'll usually be able to keep going with easy on it right after failing it once.

nest0r Wrote:@Daichi

I disagree with part of your comment. To repeat my past arguments, I believe Anki should be seen more as a HUD rather than a static vessel for depositing artefacts. It's the perfect environment/perceptual overlay for actively, continuously learning an item or items, given the customization for designing the card.

Usually when I first add/unsuspend a card, I study it, rather than retrieve it, e.g. I'm learning it/encoding the information before setting in motion the adaptive spaced retrieval. You could, of course, have encountered that information in a different context. There's all sorts of ways to integrate. Explicit study before another, more implicit context, or afterwards, etc.

Also, feedback with or without failure (e.g. flipping the card and studying that information) improves spaced retrieval results. Feedback is good in general after retrieval or when you fail and restudy (which presents a similar situation to a new card), but also when it's corrective feedback, when you make memory errors or have low confidence in certain areas over time. When you execute a successful retrieval you temporarily enter a reconsolidation state where the memory is more plastic, so you can strengthen the encoding again, make adjustments.

Also, when Karpicke and folks talk about spaced retrieval's effectiveness, they're conceptualizing the process of spaced retrieval (study/testing trials, encoding/recall) itself as a learning phase prior to some future evaluation (in the context of school something like exams at the end of a semester, or in research a final evaluation of effectiveness [hence ‘retention interval’]) but for lifelong study, it's sort of an infinite learning phase of that information (with ever expanding intervals).

“... research in cognitive science has challenged the assumption that retrieval is neutral and uninfluential in the learning process (7–11). Not only does retrieval produce learning, but a retrieval event may actually represent a more powerful learning activity than an encoding event. This research suggests a conceptualization of mind and learning that is different from one in which encoding places knowledge in memory and retrieval simply accesses that stored knowledge. Because each act of retrieval changes memory, the act of reconstructing knowledge must be considered essential to the process of learning.” (Karpicke & Blunt, 2011)

A lot of your posts nest0r, have such big words all in a very compressed space. They often feels like the problem I'm having with my Japanese sentence deck, I can read the words, but I have trouble understanding the sentences as a whole. XD So I guess this is just a wordy way to say, I don't exactly understand what your trying to say here, but I'm certainly interested.

Anyway, I'm not saying I know everything.

It just occurred to me that perhaps it's the idea of SRS = Spaced Repetition System/Software that causes problems. Spaced Retrieval is more accurate, as it's the combination of the retrieval practice, the cue→target/question→answer, with adaptive spacing, that is important.

(Well, what resolve once noted as ‘passive review’ is interesting also, and I wrote on that some in this thread [as ‘ungraded’ cards], though I didn't make the connection to resolve's piece ‘On Learning’ till just now, but even that entails a measure of active learning per repetition. Especially when it's swaths of entirely new text.)

It's funny, I took those phrases, ‘passive review’ and ‘active recall’, from when I first learned about Anki, and for a while applied them as ‘recognition’ and ‘production’ (renaming ‘passive review’ as ‘passive recognition’ and using ‘production’ only for ‘output’), even though I was talking of two variations of retrieval, and what I considered ‘passive recognition’ and ‘active recall’ was really a matter of card design/cue strength/goals/etc... 面白い.

Edit: @Daichi, a really simple summary: Anki is an interactive learning ecology, not a static, separate storage facility, i.e. it's for both newly encountering information and renewing previously encountered information. And with stuff like Rikaisan/Realtime Import (and MorphMan for tracking knowledge, a tracking capability that will hopefully expand in the future), the boundaries dissolve even further (re: the illusory separation between SRS vs. non-SRS that some attempt to construct rather than seeing the SRS as an overlay/supplement/scaffold, a flexible HUD).

This question really depends on what you're comfortable with for your accuracy scores and whether or not you're piling up so many reviews that it's becoming hard to have time to study anything new. I'll describe what I've gone through with my regimen in the last 4 months.

In my sentence deck I started out grading myself a bit too hard. It was fine at first because there were so few cards being rotated, but after a while I just had no opportunities to add new material because I had too many cards coming up for review everyday.

My long term retention was somewhere around 97%, and my daily reviews tended to be in the 92-93% range. I was learning the material really well, but after about 3 months everything ground to a halt at like the 2500 card mark or around lesson 65 of Kanji in Context. I went from adding a new lesson every other day to adding a new lesson every 4-6 days. It was infuriating.

So I loosened up my rubric a bit, and am a little looser with it. If I miss a card then it's auto fail like it was before, but I'm being a lot more liberal with the medium/easy button. My new strategy is as follows:

1. If I ace a new card first time I see it then I mark it easy.
2. I now mark most cards as medium in an attempt to give myself more of the benefit of the doubt when it comes to my own ability to remember new material.
3. If it's a card with atypical readings, difficult grammar, or difficult to remember meanings for words then it's a hard.

This has brought me back to the land of the living where I'm adding new lessons at the pace of a little more often than every other day. Some days if there's few enough cards up for review I'll add multiple lessons.

My overall accuracy has taken a bit of a hit because of this with a lot of 80-85% days. My long term retention hasn't fallen more then 2-3% though. I'm still solidly above 90% retention for mature cards.

<rant>Every thread now I see people giving advice how to take shortcuts. If I miss a stroke, I don't fail it. If I forget the story I put it on the front of my cards instead. I don't bother with stories, just memorize the primitives. etc. etc. Whatever happened to just doing it the original way... it does work you know.</rant>

Anyhow... for answering in Anki, the 'default' answer is usually good (i.e. space bar). For example, for a new or failed card, this defaults to '2', and for young/mature cards to '3'. If you fail whatever your failing criteria is, obviously '1', and only if it's really very easy then '4'.

If you're failing cards a lot of times and they get suspended because of being 'leeches', It's a sign that you are doing something wrong on that card and you should do something about it - like you have a story that doesn't really work for you and you should think of a better one - or in the case of vocab you need to look up more examples or make up a mnemonic to help.

It depends. The longer answer is, for vocabulary, if i know the word J->E E-> readily, i hit space. If i don't know it fail. If i fail cards, depending what my test criteria is, i write down the word, and | or sentence, the translation, and re-learn it. I only hit very easy, if i can be absolutely sure, that i know the words, or cards in general. If i don't know all the information on a card, the exact formulation of the translation for instance, i still push easy.

You have to know what you want, push easy, if you can say you know a card, and if you fail it several times, remove it. Learn it again, add it, and review it. You don't have to go hardcore, though. Because not everything you add to Anki, is so important, that you have to remember it.

The newest version of Anki has some great improvements, in that it adds a step, before new cards go into regular reviews. I can't wait to see the final version of this, and how this improves reviewing with Anki.