What is the level distribution on here?

I reset a few levels back thinking it’d take off some of the reviews that had piled up (I had a few thousand), and none of them came off, lmao. I reset for nothing. Made it back on track though.


A month is nothing. My longest level up is like a year, because I was in Japan and had no way of paying for the subscription from overseas. (; u;) (This is why you should just get lifetime I guess…)


Relative change in number of users per level. Might be interesting for seeing which levels are “harder”. Level 60 is excluded because it is +1210%


I went to school in Japan, and a few of my other foreign friends there were also using WaniKani, but the highest any of them went before giving up was around levels 23-26. That’s when they all just got sick of it. Looking back at 23 in particular, I see a few items that still come back to haunt me to this day. I understand now.


My longest is also a year :flushed: I was also in Japan haha! I could still pay but the issue was that I was taking some kanji classes and I couldn’t balance wanikani and regular kanji classes (and more classes of course). So I took a ‘break’ (very long break) but I don’t regret it at all. I got back into wanikani pretty easily once I recovered from my 2000 review backlog and now I got used to taking regular breaks, usually every 10 levels, to solidify kanji and vocab I learned


That’s awesome! My kanji classes started at N2 level, even though I was probably N4 in kanji at that point, so I needed WaniKani to survive. ;__; It’s not too bad to take a break from it when you’re actually in Japan surrounded by kanji though. I bet you run into stuff on WaniKani you learned in Japan anyway.


Wait, how do you read this graph? Are the levels 58 and 59 “harder” than 57?


I think both extremes are bad. If there’s a big increase in users between levels then it probably means people gave up more on that level, but if there’s a decrease then that many didn’t continue to the next level.


Ideally it would be pretty even and then hit a wall at 60. Actually, ideally there would be more than sixty levels.


I’ve reset a few times now, from around 30-31 lvl


Was it helpful?

1 Like

Can you scatter-plot the highest level achieved vs. the number of posts made in the forum (or posts read, or some other activity metric)? I wonder if there’s any indication being involved in the community helps people stick with it.


Eh, there’s a big difference between “seeing” kanji in Japan, and actually being able to read them, or check your comprehension of them (i.e. actually “learn” them). Kanji really aren’t something you can absorb simply by being around them a lot, unless you already know a lot about the language (enough that “new” kanji are scarce, which makes them easy to single out and focus on briefly).


I could probably put together a script to cross reference that tomorrow 230185602321088514


Try making a graph that shows how many people are at x level and beyond, that way it’s possible to more clearly see how many people have made it to that level rather than how many people are on that level. I think it would be a great motivator for a lot of people to think something like “Wow, only 3000 users have ever made it this far.”


Here’s a different visualization, showing the percentage of users that ever make it to a level. I’ve clipped off the free levels, and added a trendline that shows it really is an exponential trend. Basically, on average 6.8% of users will “stop” on a level (assuming everyone stopped right now, which isn’t totally right).

Another graph showing the percentage of users that reach a level then keep going.


Surely the value at level 60 should be zero?

Unless someone’s been holding out on us…


Hey, I like that last one. Once you get over the level 5 hump, chances are you’ll make it. I guess the big exponential drop-off shows the same thing (the 60 value is not much less than the 5; the big drop happens right away), but it’s not obvious.

Oh wait, I’m reading that slightly wrong. The 5’s go on to 6, not necessarily ultimately to 60.


Oops, it’s an off by 1 error. Classic programmer mistake :sweat_smile:


What it does show is that, in general, the further you go the less likely you are to quit, at least up until ~30. After that, the dropoff rate per level plateaus (ignoring the noise).