WK should work with script devs and officially implement some of them


I guess you’re not completely wrong there :thinking:


you’re proving my point.



Does that exist yet as a WaniKani script or API service?

(There is my WKStats projection script of course but I would still be interested to know)


It does not


A thing I think WaniKani should consider is, a better way to handle typo checking in English.

  • Rather than No Cigar, “did you mean” should probably be followed up with multiple choices, with trick choices (from blocklist or from JMDict, or from unrelated vocabularies) - if you answer so, you are wrong.
    • This should be better for Romaji meanings as well, to check whether you really know.
    • Honestly, this may entails lots of addition to each vocabularies. Vocabulary meanings are better checked by phrase / cloze tests, rather English translations, anyway.
    • There might be options in settings to turn on Experimental Features.
      • Ignore / Double-Check might be a part of Experimental Features.

It might be possible to implement on mobile / third-party apps, even at this current state of API as well.

I don’t exactly how to calculate dissimilarity between words meaningfully.


That could also be a cool userscript for someone willing to take on that undertaking

I know it’s a super late reply, but that really echoes my impressions.
To follow up on my previous post, I only recently started to manage leeches decently thanks to the Item Inspector script. That one actually helped me to clean up numerous problematic vocab words so far. Huge thanks! :durtle_noice:

I work as a programmer too, and the “aw, but we have to support the feature later on” narration is oh-so-familiar but I’m not exactly fond of it in this case. I mean, it’s not like you’re building a completely new resource allocation app in a rigidly time constrained system. It’s not like you’re adding useless bells and whistles either. These are tangible improvements to the learning process.

As for Bunpro comparisons, for one, I recently confused their API integration with Kitsun (which doesn’t have a direct one). They’re quick enough for me to think others have their stuff, while they don’t.

Generally, leeches is one aspect, but there is self-study, practice, jisho search, searching user synonyms, even silly overall progress bar. Then there are kanji similarities, transitivity pairs etc, etc.
Extra Study had me hopeful, and I was very happy to see it, but it needs to followed up on. To me, it’s not much on its own.


After having used WaniKani for around a year and Anki for much longer, I would pick Anki mostly due to its flexibility, but also due to the Quality of Life features which it offers, like cap on daily lessons/reviews, leech management and several scoring classes for each card. The only thing Anki is not very good at is review schedules and card scoring weights which are very personal anyhow and need to be tweaked.

WaniKani seems like a good, well-structured system (radicals → kanji → vocab), but from my experience tends to suffer from massive diminishing returns, because it’s too restrictive, punishing and distracting with the amount of extra information it loads the user with on top of kanji.

Here’s a couple of things which could be changed to streamline the overall user experience, some of which are covered by scripts:

  • Removing user keyboard input for answers. It takes time, is error-prone and opens up the system to way too many issues (keyboard layout differences, spellchecking, accidental typing errors, etc.).

  • Radicals reaching only Guru stage. They’re useful, but shouldn’t be forced onto users for months of study, especially that for much more complicated kanji they just add noise.

  • Replacing the partial spellchecking system with more meaning answers for kanji and vocab. What we currently have are mitigations - spellcheck accepts imperfect answers, we need blocklists for words which the spellcheck wrongly accepted, allowlists for new synonyms, etc. That’s a lot of periodic maintenance overhead and shifts the burden heavily on the user.

  • To the above, allowing user synonyms when learning items. Since around half of my initial WaniKani journey I had to add a user synonym literally for everything, because the kanji/vocab meanings were either too vague, too restrictive or sometimes simply wrong.

  • Configurable caps on daily reviews. Sometimes it’s very discouraging to see a massive review pile after an unintentional break.


I disagree with removing the partial spellcheck (it would be extremely annoying to get answers wrong due to minor typos), but the rest of this I agree with strongly. WK is way too narrow in what synonyms it accepts for most items, and especially for Kanji. It’s also very annoying that lessons don’t allow you to add user synonyms; that’s the best time to allow the user to add them, but for some reason it’s the only time you’re not allowed to do it.

1 Like

After doing both WK and Anki style decks, I have to disagree with this one. I think keyboard input for short answer items like Kanji and simple vocab is better and Anki style is better for things like sentence cards.

As for whether it should be an option within WK, I’m on the fence. I can understand some people disagree and prefer it other way, and it’s good that there’s an API that allows that to be customized, but there’s definitely an argument to be made for it to be rolled into WK.

I think they should keep radicals going all the way to burned, but I think they should stop having radicals at all after level 30. Instead, just build on Kanji with radical components added.

They’ve actually been doing quite a bit of this over the last year or so. I can remember several discussion threads that results in changes to the white/black lists. Of course, it’s always going to be a work in progress but I think the Wk team has been doing a good job of addressing this.

1000000% agree. This isn’t even available via the API I don’t think.

I don’t know how this would work in practice and I think I would disagree with aspects of it depending on the implementation but I’m good with the general idea.


I feel that WaniKani is good on forcing you to type, even if it is English. However, I still stand strong on typo-checking + follow-up questions. (Typo-checking + forcing you to type English correctly [No Cigar] is just a mitigation.)

I totally agree with this one.

Another related issue is, whether Kanji and vocabularies should still be remembered with broken English, after months of study? Unless meaning quizzes are good enough, I won’t acknowledge.

1 Like

Well but I think that’s a fundamental different expectation of what WaniKani IS. “WaniKani” isn’t the SRS. It’s the plan, the set of custom mnemonics and the order you should learn them. The SRS is just the method. You could “do” wanikani IN anki.

I kind of like that the WaniKani team maintains focus on the essential product. I also appreciate that they provide a bare-bones functional user interface for the “I just want it to work with no effort on my part” people, but also maintain an API for all the tweakers and customizers to go crazy with “nice to have” user scripts.


Well, WaniKani team has always been trying to make vocabularies and context sentences accurate; as well as adding more on vocabulary usage relatively recently. Not to mention native audio for every vocabularies.

I have always been using WaniKani as a studying direction - a textbook, as well; just cloned much of it to Anki. But it might also be studied in a different way. (Just that it’s not vocabulary frequency’s order.)


Exactly, that’s what I’d rather they spend time on, accuracy and quantity of content. Not just adding more, but measuring which items don’t seem to be sticking for people and considering altering mnemonics, learning order, beefing up explanations, providing more examples, that kind of thing.

(If it was me, I’d spend some effort on leading more interactive learning topics here on the forum, too, as a value multiplier for subscribers (vs. copying an anki list from somewhere). But un-asked volunteers seem to be doing a decent job of that with the book clubs and stuff. This is a pretty great community)


Right, but if a SRS is put in place, I feel like adding a couple minor features (leech detection, review caps, etc.) to prevent users from shooting themselves in the foot is not a bad idea :slight_smile: .

One of the core issues with SRS’ requiring answer input is that that input either needs to match the key or there needs to be some mechanism to allow ambiguity. We have the latter (in a way), but that’s not full-proof, because language is language and it’s not like every kanji or word has that one exact meaning. I can’t speak for others, but I can’t imagine myself remembering the exact word-for-word meaning of a kanji/vocab item after months of SRSing despite knowing perfectly well what the kanji/vocab can mean. And adding synonyms to every single item is exhausting and makes using the said product long-term painful.


It’s important to keep it simple. Look at netflix for example. Sometimes they add a feature but once they realize it’s not useful they remove it right away.


Which, and maybe this qualifies more as a “shower thought” for that thread, I wonder how the economics would work out to pay someone like Misa to make an exclusive subscriber-only video once in a while on a common error/misconception/interesting question from the forum. And post it here instead of YouTube.

The way I see it, they have 3 things that someone might want to pay for. 1) the content, 2) the community/forum, 3) the app. I (some random unqualified dude on the internet) think #1 is the most important, but let’s face it, can be stolen. After that, #2 is the thing that can’t be stolen or replicated.

Economical issue is a unique consideration, otherwise

  • Initial effort, in exchange for pedagogical value
  • Maintenance - more creation can also mean more to maintain

But it’s tiresome for me to think for others, as well as I am not specialized in this kind of stuff… Providing feedback is OK, but let’s not expect too much.

This topic was automatically closed 365 days after the last reply. New replies are no longer allowed.