So all the new widgets are cool!
One in particular has me stumped. The “correct reviews widget”.
Why has Wani Kani called this area the learning zone? Is there any data to back this up?
Dont get me wrong… I would love to be there too!
But is wanikani’s rec to not add more cards until i am there… to cry myself to sleep?
Id love to know where this number came from , and if i should take it seriously.
The your “outside the learning zone” with no data is a litle.. lackluster
You can think of it as our way to encourage you not to be a perfectionist (you ought to get some things wrong), but also to try and get a decent number of reviews correct.
In terms of the studies, they really are all over the place. Generally not lower than getting about 2/3rds of your (thing you’re doing) successful. If you’re successful 2/3rds or 3/4ths of the time, that’s probably fine.
Another part of it comes from where you’re motivated / demotivated. Some people will get 70% right and be like “I’m terrible!” but in reality I think that’s pretty good, so long as you’re challenging yourself well.
I don’t exactly remember how we ended up with our particular range, but a ~70% success rate is fine, I think. Lots of opportunities to learn, and you’re challenging yourself, etc.
I think this is very reasonable and I agree with the motivation, but maybe calling it “the learning zone” is not the best name? Because it makes it sound that outside of this zone you’re not learning which is of course not true. And I know some people really take these things to heart and it can easily be demotivating for them to be told that they’re doing it wrong.
Maybe “the efficiency zone” or something like that would be a better name? I think it would be more accurate too.
I got an 82.37% and it says I’m outside the learning zone.
Google AI search of SRS:
Below 80% Slow down too much time going back over reviews.
Over 90% Going too slow need to push more.
The more new kanji you do at once, the more it will lower your %. Look at month % more than week as it would be a more realistic average.
Also my 83% (month long) comes with 20 day average per level. That’s kinda slow. Its heavily factored on how much previous experiences I have with the words used. How much time I have etc.
If you aren’t beating your keyboard from mistakes, you enjoy the pace you are going, then keep doing what you are doing. Nobody is going to ask what you % at wanikani was as you are reading japanese.
In the end, ever 4 to 6 months see if you are recognize a lot more than you did before. Its hard to see it week to week but after 6 months, you’ll feel it.
But its fun tool to see a % and feel good about it.
Oh one last thing. my 83% @ level 29 includes a lot of long term items I learned. Your 73% probably doesn’t have that many items going thru higher levels yet. Master and Enlighten items make the % of new items less. so the % correct gets higher.
Just my random views.
While I really like the visual of the widget (cute art), I don’t really like this “learning zone” concept. Nor that it’s showing an angry expression when you get 100% on reviews. As a perfectionist, the 85-95% becomes the new goal instead. It’s pretty narrow. 70% isn’t good, 100% isn’t good.
So, I really like the widget as a visual, but I’d prefer it without angry faces or judgement of what’s “acceptable” correct review percentage. Just my opinion as someone who’s long graduated from using WK 
Doesn’t this system tell them that’s not good? I mean personally I’d want to work to land a little higher, but I feel like this thing is going to add to that de-motivation far more than what help you expect from it.
I don’t exactly remember how we ended up with our particular range, but a ~70% success rate is fine, I think. Lots of opportunities to learn, and you’re challenging yourself, etc.
What is the range though? I’m showing as out of the learning zone with just over 80%
I like having the percentage but being shown as out of the learning zone isn’t helpful and just gives me a bad feeling - and I’d generally be quite happy with 80% otherwise. I agree with Redglare that I’d prefer it without any judgement
It says 85-95% in their knowledge guide: Correct Percentage | WaniKani Knowledge
Just for fun, I tried to do a 100% on one item and I got outside the learning zone, maybe because if you get 100% you’re too confortable I guess
Wait, why does the girl get mad at me for getting all my reviews correct? Isn’t that good?
Prolly to encourage the idea that mistakes are okay, perfectionism is bad, and prolly need to add more new vocabs to the review phase.
Plus, now i have the confidence to NOT hit the home button during reviews after making 1 mistake (fat fingered しゅう instead of しゅ, or remembering the kanji meaning not the radical keaning i.e 楽)
The general idea is that, long term and on average, if your accuracy is very close to 100% it means that you’re only reviewing things you already know very well and are not really learning anything new.
That should not really be a thing for just a small batch of reviews though. But if you get 98% accuracy over a full month for instance it probably means you can push towards more complicated content.
it makes sense but also like.. sometimes I just remember really well so the correct percentage for the day is high but this girl still stare at me angrily. That doesn’t feel very pleasant
I agree, it makes more sense for longer period of time
That doesn’t make sense for a course that’s adamant about the benefits of not letting students skip content they already know. You can only manage so many reviews a day.
Currently I’m working on level 24, and out of 32 kanji, I was unacquainted with 14 (couldn’t read/didn’t know the meaning), acquainted with 11 (could read but was unsure/had a vague or incomplete knowledge about the meaning/could recognise but not write and had no concept of their radical makeup) and had a reasonable command of 6. That’s decently stimulating to me, but my accuracy is still over 99% as it was when WaniKani was teaching me N5 kanji. I was surprised the widget was showing an angry face but having learnt about the reasoning I reckon it’s just horses for courses so I don’t mind at all.
Managing 99% accuracy if you are actually learning dozens of kanji per level is very impressive and not really typical, but I agree with your point that WaniKani doesn’t really give you the tools to adjust your pace if you do too well. If your accuracy is too low you can always slow down, but the max speed is artificially capped.
I think the ‘learning zone’ range works well for the longer periods where things average out: 7 days+,
It doesn’t make much sense for the 1 day period where you can never be inside the ‘learning zone’ if you only do a few reviews. (With less than 7 reviews it’s impossible to hit 85%-95%, the first oppurtunity would be 6/7 = 85.7%)
Ideally the 1 day ‘learning zone’ range would widen. Maybe from 70% - 100% or whatever.
It could also give some info, maybe a hint to the blog article. Maybe change the 100% text to something like: ‘very good, but you should challenge yourself more!’
The issue with doing ‘too well’ is that you can only do so many reviews and lessons before you get gated behind waiting periods. Especially in the first few levels. And I expect this would be especially discouraging to new users…
I really don’t think the single-day stat is at all exploitable for this purpose, it’s way too noisy.
I have a kanji writing Anki deck that I’ve been using for close to a thousand days and have done nearly 40k total reviews over this interval. I’m using the FSRS algo with a target retention of 90% and my all-time stats give me an actual retention of 89.4%, while over the last year only I’m at 90.1%. So you can see that the algo is extremely well tuned for my performance.
Yet the daily stats are all over the place. I managed 100% correct yesterday, 86% the day before, 100% again 3 days ago and a very rough 60% accuracy 4 days ago.
Using these individual results to orient my studies would be meaningless.
Right, what the system ought to do in this situation is extend the intervals if your recall is really good, so that you get a lower review workload, but WK doesn’t have variable intervals, only fixed. It seems a bit silly for the system to point out the very high accuracy when it doesn’t support doing much about it…
Simias:
I really don’t think the single-day stat is at all exploitable for this purpose, it’s way too noisy.
It works fine for me, I land in the 88-90 range daily. But that’s with 200+ reviews per day.
You are right about the noise though. But widening the ‘learning zone’ interval if there’s too few items in the time period measured might be preferable to not showing anything due to noise. Depends on the user I suppose.
It’s true, I suppose it’s not so much the time interval as the number of reviews that matters. It should probably not average less than a thousand or so reviews.
