Skip to content

Be Sure to Fail

I’ve been getting back into the habit of doing my Surusu and Anki SRS deck reps, in addition to the reps I always do daily for Skritter. I came to a bit of a realization today, which I am certain has been written about somewhere, but I never really took it as much truth until I started to follow it myself. That is, when doing SRS reps, always aim to fail.

Something about the self-grading system on SRS programs are dangerous; you judge yourself and it is very likely you don’t want to judge yourself poorly. Or, if you’re like how I used to be, you judge yourself just to “get through the reps”.

So, now, I aim to fail each card as they come up. I put high standards on myself (depending on the card, MCD or regular; Heisig or not, etc) and make sure I meet them as I go. For example, tones always have to be right. Not the “oh right, I knew that [even though I got it WRONG] so I can pass it”. Now it’s failed. I would rather have a large stack of failed cards than a large stack of finished cards that I really didn’t learn anything from.

I’ve noticed its much more helpful this way too, and the results go much farther than they did the old way. Of course, I also liberally delete as boring cards come up. So, in your SRS reps, make sure you fail!

10 thoughts on “Be Sure to Fail”

  1. I’ve been doing the same thing for a while and would have to completely agree. I think aiming to fail with reps is especially important when you’re dealing with multiple languages. The whole “oh right, I knew that so I can pass it” excuse is very easy to (ab)use, but it doesn’t help. Speaking from experience, sadly.

    1. Sounds like you and I both had a similar experience. I would always come back across cards I passed in that fashion, and would just continue to pass them that way. Rather bad cycle, really. Later when I would come across the word in some context I would discover I only had a faint idea of what it really meant, and realized my way of studying wasn’t helping at all. So far, this is going much better.

      Thanks for your comment!

  2. I’m the same way right now with verb tenses in Korean. I’m extra hard on myself even if I know I really knew the card, but just slipped up and said it was “went” instead of “goes” I’ll fail it. I force myself to pay attention the tenses so that it’ll just be intuitive later on. I’ve noticed that I’ll slip up less and less now, I did noticed that it tends to happen on easier cards too. I’m not sure why that is, I guess it’s just laziness? But I agree, making sure you fail in your SRS is a good strategy!

    1. I’m quite interested in getting into Korean (we had a class last semester that talked about Korean history so much that I figured I may as well learn a bit of it, seems interesting), so I’d like to know how your progress goes with that! I noticed I would get really lazy with cards sometimes (eyes glossing over and all that) and I’d just pass the cards to get past them. If that happens, I tend to just either fail them to force myself to pay attention, or more often than not, I just walk away and come back to them later.

      Thanks for your comment!

    1. I might have seen you or someone else mention that as soon as they got the tone wrong they failed the card–regardless! I think this is really great, especially important for Chinese/Cantonese and other tonal languages.

  3. Good policy.

    The situation is caused by one of the failings of Anki, in my opinion. It makes it far too easy for the learner to game the system, even when they aren’t trying to game the system.

    I think SRS programs should force binary outcomes (correct/incorrect) and then use other means to determine rep quality (how much time elapsed vs. avg time elapsed), were hints used, etc.

    1. Actually, you’re absolutely correct on that! As much as we’d like (and I know especially for me when I’m just not with it and doing reps to do them), it’s really easy to cheat ourselves to make the reps get done faster.

      I agree with your idea, actually. I would quite like to see something like that in an SRS. For now, though, especially on quite boring decks (learning phonetic systems and such), I keep my rep goal low (say, 5-10 or even less a session) just so I don’t get complacent and let them slide.

      1. The best example I’ve seen for an SRS bases system that had verifiable questions was lingt.com, a semi-defunct website for learning Chinese. I’m also working on an implementing such a system in a website that teaches people to read Chinese. I’ll let you know when it’s ready.

Leave a Reply to zhongruigeCancel reply