So I have a deck of 900 cards. Currently I have learned around 800 with a percentage of 97.5
The settings for Anki are the standard except the reviewing limit which is at its limit.
Im learning 20 new cards per day and at the reviews unless i completely forgot the card I always choose the hard option. Now the thing is by doing this my review per day is at around 170 and keeps increasing. I also plan to add a few thousand new cards to the deck so should I change my approach ?
I dont particularly find it difficult with the current method.
That is if I plan to choose the good option as well should I change something at its settings? Also have any of you tried this approach (that is always choosing hard) and if you did how well did you fare? 
Any tips?

People usually burn out with this approach. (imagine with 10-20k decks). And when you think about it pressing "hard" when it isn't is just messing around with the card spreading so of course it'd be detrimental at some point.

About your (good/hard/easy) settings, idk, it depends, what are they? If they feel a bit too close or too far from each other, nothing wrong with a little tweaking. It takes a bit of trial and error sometimes, but Anki is very flexible for that so you can always change them later.

I used to do have "hard" as my standard answer. IT wasn't very smart. I now think it is a lot better to press "good" and rather have the risk of failing a few more cards once in a while. By pressing "hard" you also reduce the ease-factor of the card every time, making the relative increase in interval of the card smaller and smaller. The only way to increase the ease factor again is to press "easy". My deck is now full of cards with a 130% ease-factor (the lowest), which I don't really like because it's difficult to isolate cards that are actually difficult.

I would advise you to only use "hard" for cards that are actually quite difficult. 97.5% retention is very high, so there's really no reason to press "hard". You're only spending way to much time on stuff you already know

[s0apgun]

Totally pointless in my opinion. Learning the definition of a word with a high recall rate isn't going to teach you how to use it better. That comes from lots of reading and listening. Anki is just a tool to prime the meaning and so you don't forget it over time. I always press easy on cards and if they keep failing then they will get suspended to make way for easier to remember words.

Pressing 'hard' every time has the same effect over time as eventually lowering your initial ease to 115 which is a 15% increase for every correct answer.  As people have said above, 97.5% is not very efficient if you are trying to learn a large amount of material.  Peak efficiency happens around 70% and you can learn about twice as much material at 70% than you can with accuracy in the high 90s.  If you must insist on very high accuracy, at least lower your initial ease so you can press "pass" sometimes and have benefit of the algorithm working with you.

(2016-03-01, 7:06 am)s0apgun Wrote: Totally pointless in my opinion. Learning the definition of a word with a high recall rate isn't going to teach you how to use it better. That comes from lots of reading and listening. Anki is just a tool to prime the meaning and so you don't forget it over time. I always press easy on cards and if they keep failing then they will get suspended to make way for easier to remember words.

Do you mean it's either fail or easy? Sure, you won't really know the word until you see it in the wild a couple of times, but isn't pressing easy every single time a little too much? I press easy quite often for words I've seen a couple of times and ended up failing, but doesn't that just backfire if you've just learned a word?

If that works well in the long run though, it would be a nice way to learn more words in a short span of time and focus more on reading and listening. But would that actually lower the time you spend reviewing words? I don't know, since you'll probably fail more as well.

When you say you just do it to "prime the meaning" and talk about the denition of a word, it sounds like you don't actually care much about the reading. I wonder if actually take both reading and meaning into account...

Since I won't really understand a word properly before I see it (unless it has a very specific meaning, e.g. "president"), it seems to me Core is more helpful as a way to learn readings. You don't need real exposure to learn readings except for words that depend on the context, so that's something you can learn for certain. However, you'll still be getting used to the way a word is used once you stumble upon it. So perhaps, even if the meaning is slightly off when you review the word, failing the card might actually be unnecessary.

For example, 短期 means short-term and 定期 means periodic according to Core and Jisho.org. But I sometimes mix up the meaning of these 2 words, even though the readings don't pose any challenge. I wonder if there's any point in failing them when I already know that they have something to do with time, which is arguably part of the meaning.

anki has "fail", "hard", "good" and "easy" for good reason. Seems counter-productive to me to work against the way the tool is designed, either by always spamming easy or always spamming hard. I can certainly imagine situations where spamming 'easy' for everything might be useful (trying to get a massive amount of reviews down to something manageable, for instance) but choosing hard every time is just setting yourself up for burnout. I agree with yogert909 that 70% retention is probably optimal while in a heavy learning phase where new cards are being added frequently.

Flameseek, while some words are obviously leeches and its ok to suspend these or push them out into the future with the "easy" button, I think passing cards (with good/easy) when you have a very vague understanding is probably not the most efficient use of the tool, it's designed to help you memorize after all. I tend to be ok with passing a card (with hard) where the meaning of the whole sentence I'm learning is still a little vague (these always get clearer over time), but only if I'm very clear on whatever the target vocabulary word or grammar point that the card was created to teach means, and the uncertainty is coming from other elements of the sentence. I have found that if I pass cards where I have only vague recollection of the target term/grammar, I end up with a bunch of reviews down the line that are very painful.

All this said, anki is a tool after all. So use it in whatever way is useful. But if you start finding it painful (and usually using tools in a way they were not intended does eventually become painful), be sure to make adjustments before you burn out and stop using it. Piled up cards not being reviewed won't do anyone any good.

Btw, is there any way to reset the ease of cards that have already been studied? I think I pressed hard on a good number of cards and now it is difficult to mature them.

(2016-03-02, 10:10 am)PotbellyPig Wrote: Btw, is there any way to reset the ease of cards that have already been studied?  I think I pressed hard on a good number of cards and now it is difficult to mature them.

I don't believe there is a way to reset the ease without resetting the cards' learning data back to zero unless you wanted to go into the sqlite database and change them that way.  There might be an add-on to reset just the ease, but I searched about a year ago and didn't find one.  The best you could do otherwise is press 'easy' every time and it will have the opposite action over time as pressing 'hard'.  You could also look into the interval multiplier.


@OP  This is a perfect illustration of the trouble you can create down the line by trying to outsmart anki's algorithm.

(2016-03-01, 3:44 am)EratiK Wrote: People usually burn out with this approach. (imagine with 10-20k decks). And when you think about it pressing "hard" when it isn't is just messing around with the card spreading so of course it'd be  detrimental at some point.

About your (good/hard/easy) settings, idk, it depends, what are they? If they feel a bit too close or too far from each other, nothing wrong with a little tweaking. It takes a bit of trial and error sometimes, but Anki is very flexible for that so you can always change them later.

Hi Bull007,

The Anki algorithm is generic so doesn't work for everyone (e.g. your memorization day 1 is well above average but your retention in the following days and weeks is below average).

I tried your approach of using only the first two buttons. Slowly the mature card pile began shrinking and it became frustrating. So I can see EratiK's burnout advice. It is incredibly inefficient to keep reviewing cards that are really easy; it also distracts you from cards you should be reviewing.

I see some potential solutions:

1. Manually adjust the algorithm. Perhaps EratiK can make some recommendations

2. I believe that SuperMemo has a smarter algorithm that learns from your responses. I don't know if it is easy to transfer an Anki deck to SuperMemo.

I saw that you asked a question that got closed.  You could ask in the anki small questions thread, but the question could just as well be asked here too.

Bull007 Wrote:How do you guys have your starting ease set at ? I have mine set at 250% which is the basic setting but I feel like the interval when pressing good is a little bit too much. I can still remember them this way but im not that confident that I can retain them that long.  What do you guys think should I set it at or should I not tamper with it ?

First of all, you need to have an idea what you want your accuracy to be.  As I mentioned, around 70% is the most efficient if you are trying to memorize the most amount of material.  However, I find getting 70% day in and day out to be rather demoralizing, so I aim a bit higher to around 80%.  The efficiency curve is rather round, so 80% is still much much more efficient than 90% where efficiency tends to rapidly decrease the higher you go.  This is something that is more up to individual preference than anything.

Once you have an idea what accuracy you feel works for you, it's a little bit of trial and error based on the difficulty of what you are trying to memorize.  The three main settings that you want to change are the learning intervals, starting ease and relearning.  Learning intervals affect your accuracy for new cards below 1 day intervals, starting ease affects your accuracy for intervals over 1 day, and relearning affects cards that you fail.

I use 4-5 learning intervals for most things Japanese and it makes sense to make them even multiples from the lowest interval and 1 day(1440 min).  Something like 1 5 20 75 360.  If your learning accuracy goes above or below your target accuracy, add or remove steps until you are hovering around a value where you are comfortable.

Starting ease is simple but it takes longer to notice the effect.  "Ease" is simply the percent that your interval is multiplied by to get the new interval for a card marked "good".  But anki raises or lowers this per card based on answers lower or higher than "3- good".  I would take note of your "average ease" and your "learning %" in your anki stats graph and adjust up or down based on those numbers.  Every week or two, I would revisit these two numbers until you you are satisfied with your accuracy and progress.

Realearning works the same as learning steps, only anki doesn't break out a separate stat for relearning so you will have to make an educated guess.  Relearning cards usually don't make up the bulk of your workload anyway, so it's not essential to get it perfect, but 2-4 steps seems to be reasonable for relearning.

I hope this helps

[s0apgun]

My retention rate on mature cards floats around 70-75%. Yes, it does cut down on my review times. If the card is being failed too often because it won't stick then I'll look up additional info to add to its definition. If it still is being failed then it will get suspended, which is a good thing. I only pass a card if I get both the reading and meaning correct. I do press the good button too but only for new cards that are added during that session. The reality is that for some words a simple definition will suffice but for the majority of words you will not get a feel for them after just a couple times of seeing them in the wild; you need to see them used over a hundred times to be able to use them.  After using Anki for so long you learn not to sweat the details too much because its a tool to help you and as soon as it becomes a burden to your study habits then you need to change it up. Keep it easy and consistent. Use other sources to that require less mental recall because it will be more beneficial in the long run. There is plenty of words that I know, that are mature in Anki, that I've read and heard so many times. But when it comes to speaking them, I can't get the right use down. Which means I need to see it used in many other contexts. One definition from Anki just doesn't cut it for practical use most of the time.

短期 means short period and 定期 means fixed period.

短 being short　定 being constant　期 being period

I suspect that your Anki settings are fine and the algorithm will work properly if you give it accurate information. I'd just start using "good" and "easy" where appropriate. At some point, some fine-tuning might help, but at the point, that improvement is in the noise.