Scaling the review mountain part 6 - the race to zero

In 2020, I raced through Wanikani as fast as possible, reaching level 60, and then stopped after a year. However, I later realized that I had forgotten pretty much everything, so I resubscribed and attempted to start whittling down the pile of thousands of reviews I had accumulated.

This is the sixth thread documenting my challenge. You can read about my journey up to now in the previous five threads - Part 1, Part 2, Part 3, Part 4, Part 5.

The journey so far

I started this challenge on May 12th, 2021 with a mountain of 6218 reviews. I naively thought that I’d mostly remember everything and that I could get through all that in just a few weeks, but instead I could barely remember anything from the higher levels, and it was brutal and my review count actually kept going up. After a month in a half, with my reviews up to 6618, I gave in and ended up resetting from level 60 down to level 46.

Even after resetting, it still wasn’t easy, and my review mountain reached a post-reset high of 5319 before it finally started going down again.

Even after my review count peaked, it was slow going for a while, but then my review count started to decline faster and faster and fell like a rock in spring and summer of 2022.

Unfortunately, since then, progress has been stubbornly slow, particularly for the last two months. I think this is because with all the easy items burned, most of what is left is leeches, making things a lot harder, but I’m still making progress slowly but surely. My review count is now down all the way to 675, and I’m hoping it will soon dip below 500.

This can also be seen with my average accuracy over the last six months. After peaking at 83-85% last summer, my accuracy has been languishing in the low 70s. At least it seems to have stopped dropping again for now.

My average review speed doesn’t show a clear pattern over the last six months. For a while I thought it had gone down for good, but recently, it went back up again.

Lastly, here’s a graph of my counts of Apprentice, Guru, etc. over time.

After resetting to level 46, it took me over a year to level up and reach 47 again. Level 48 took the rest of the year (~6 months), and now I’m about to hit level 49 after only three months (obviously, the lower my review count, the faster I can level up, because the new kanji on each level come up for review more often when there are fewer reviews due to randomly choose from.)

30 Likes

What’s your strategy for new lessons?

3 Likes

I just do all the lessons whenever they are available.

Back when I first started I was worried about getting overwhelmed again and I only did a few lessons at a time, and did another batch of lessons whenever I guru’d something (which is another reason why the first level up to 47 took over a year, besides having thousands of reviews to compete with). However, that was back in 2021. Nowadays, I don’t put any limit on lessons.

6 Likes

Really interesting :slight_smile:

Do you think you will hit Level 60 this year again?

4 Likes

Yep. Obviously, things are taking a lot longer than I expected, so I’m less confident than I was. However, I still expect to get the review count down to 0 eventually, and at that point, I should be able to level up pretty quickly.

4 Likes

Aside from WK, how do you feel about your Japanese overall as you start your (likely) final thread in this saga?

Are you interacting with any native material, and how is that going for you? Do you feel like you are accomplishing your original goals in Japanese learning? Asking for a friend.

4 Likes

Some, but not as much as I’d like. Most of my Japanese study time is occupied just keeping up with WK, JPDB, and Satori Reader. But I do consume native content sometimes as well.

I’ve been very slowly reading this story. I can read it easily now (few unknown words), although I still have trouble understanding a lot of it.

I also often watch Youtube videos. For example, this morning, I watched Hayato’s Traveler’s Rest video. I can understand a lot of it (e.g. I was able to deduce the joke where he commented about the cat supposedly being a stray that wandered in, despite the fact that he chose which kind of cat he would get. Context helps a lot, but obviously that still wouldn’t be possible unless I already had a high level of understanding) This one has more new vocabulary due to all the brewing and crafting stuff and even a bunch of kanji I hadn’t seen before (e.g. 薪, お粥, 鍬, 蝋燭).

3 Likes

By the way, if you want to learn more about my Japanese learning journey, I’d recommend reading my log here.

2 Likes

Just curious, now that you’ve probably seen most of the items in your review queue are you ever tempted to spend a couple hours on a Saturday and fully push down to 0/0 just to see it? I notice you usually do a consistent amount of reviews per day and obviously doing a bunch a reviews at once runs the risk of having a huge surge reappear later, but I feel like if I had come so far I wouldn’t be able to resist lol.

3 Likes

I’ve thought about it occasionally, but it doesn’t seem like there’s much point. Early on, there were a couple times when I did a massive number of reviews in one day to push things down, but of course at the time my review count as so high that it was just a brief dip anyway.

2 Likes

Well, that was a lot faster than I expected. I hit level 49 this morning after only five minutes.

Based on the WKstats graph, each of the three levelups so far has taken a little less than half the previous one. If that pattern continues, it implies that the next level will take me around a month.

Of course, a level up means tons of lessons to go through. In the morning session, I did 30 reviews in five minutes, followed by 60 lessons in 13 minutes (18 total). Unusually, I decided to do an evening session as well today for the first time in months. In the evening, I blazed through 100 reviews in 14 minutes. In the process I also guru’d another of the remaining level 48 kanji, and thus add three additional lessons (+1 minute, for a total of 15).

Excluding the lessons, I went through 100 reviews in only 14 minutes this evening, which is absurdly fast. But it makes sense because I was rushing through them as fast as I could, and I did miss several burns I should have gotten as a result.

I also missed three of the six level 49 kanji that came up for review this evening. Not that that is too surprising, since rushing through 32 new kanji lessons in the morning is not exactly a recipe to remember them. If anything, I’m impressed that I managed to get even three right.

Thu Mar 16 2023:
Time spent: 33m L14
Reviews completed: 130
Reviews remaining: 644
Reviews in next week: 945 (+301)
Reviews in next month: 1348 (+403)
Accuracy: 74.62% (97/130)
Level: 49

Current item counts:
Apprentice: 331
Guru: 378
Master: 330
Enlightened: 882
Burned: 5673

6 Likes

Have you considered doing fewer lessons at a time when they come up?

I know you’re keen to get back to 60, but I wonder if it would actually be faster while you’re still dealing with a fair number of reviews anyway. You’d be more likely to remember them when they come up again, and it might keep the review pile lower over the course of a few days.

6 Likes

If you want to “race to zero”, it’s probably a bad idea to add new lessons, let alone this many. I keep my reviews at 0 as much as possible and I still do max. 10 Kanji lessons a day.

5 Likes

This morning, I got seven of the nine lvl49 kanji that came up for review, including two of three that I missed yesterday. I did 50 more reviews in the evening, but no new kanji came up.

Fri Mar 17 2023:
Time spent: 26m L1
Reviews completed: 150
Reviews remaining: 607
Reviews in next week: 899 (+292)
Reviews in next month: 1322 (+423)
Accuracy: 77.33% (116/150)
Level: 49

Current item counts:
Apprentice: 334
Guru: 363
Master: 339
Enlightened: 872
Burned: 5689

6 Likes

I think when you’ve done new lessons, twice daily reviews are a great idea. Even if you just do 50-75 each time.

2 Likes

When I first started this, I decided to record the total time spent doing reviews and lessons each day. For a long time, it didn’t make much difference, because I’d very rarely have lessons to do, and it would only take a minute anyway. But when I leveled up, I had a massive amount of lessons to do at once.

Therefore, calculating average review speed based on the time taken to do lessons as well causes a massive unfair drop in review speed whenever I level up. For example, when I hit level 49, I did 30 reviews in 5 minutes, and then 60 lessons in 13 minutes. This means the real review speed was 30/5 = 6.0 reviews per minute, but I would calculate it as 30/(5+13) = 1.7 reviews per minute!

Therefore, to make the speed graph more useful, I decided to only calculate it based on time spent doing reviews, not lessons, going forward.

Of course, it is impossible to reconstruct data that was never recorded in the first place. The good news is that I managed to find paper records where I tracked the time spent doing lessons going back to November 6th last year (except when I was on vacation), plus there were two posts in July 2022 when I mentioned the time spent on lessons explicitly (including when I hit level 47).

This means that I have the records for all three level ups, which are the most important numbers, and also have more or less complete numbers going back to November. The numbers before November will be distorted downwards a bit due to not taking into account a random minute or two spent on lessons here or there, but it shouldn’t matter that much.

Here is a graph showing both the old and new numbers. The old calculation (dividing reviews by time spent on reviews + lessons) is in blue and the new calculation (ignoring time spent on lessons) is in red.

The two lines overlap until November, because there is no lesson data before then. As you can see, there are huge drops in the blue line at the end of December and again now, when I leveled up. I was surprised that there was also a noticeable divergence even in between them though, just from the random minute here and there. I had no idea it could make a meaningful difference in the numbers.

I went back and edited all the posts where I could find time spent doing lessons. (Exact list below). I’m planning to record it as e.g. “18m L13” meaning 18 minutes total (as before) with 13 minutes on lessons.

Here is the list of entries I backfilled:
2022-07-06: lessons: 1m, reviews 40m, total, 41m
2022-07-21: lessons: 20m, reviews 18m, total, 38m
2022-11-06: lessons: 1m, reviews 17m, total, 18m
2022-11-07: lessons: 1m, reviews 18m, total, 19m
2022-11-09: lessons: 1m, reviews 15m, total, 16m
2022-11-15: lessons: 2m, reviews 30m, total, 32m
2022-12-03: lessons: 1m, reviews 18m, total, 19m
2022-12-08: lessons: 3m, reviews 14m, total, 17m
2022-12-11: lessons: 1m, reviews 5m, total, 6m
2023-01-01: lessons: 11m, reviews 33m, total, 44m
2023-01-16: lessons: 1m, reviews 18m, total, 19m
2023-01-21: lessons: 1m, reviews 19m, total, 20m
2023-01-22: lessons: 1m, reviews 19m, total, 20m
2023-01-26: lessons: 1m, reviews 19m, total, 20m
2023-01-27: lessons: 1m, reviews 5m, total, 6m
2023-01-28: lessons: 1m, reviews 16m, total, 17m
2023-01-30: lessons: 2m, reviews 20m, total, 22m
2023-02-09: lessons: 1m, reviews 20m, total, 21m
2023-02-14: lessons: 1m, reviews 20m, total, 21m
2023-02-15: lessons: 1m, reviews 19m, total, 20m
2023-02-22: lessons: 1m, reviews 21m, total, 22m
2023-02-24: lessons: 1m, reviews 20m, total, 21m
2023-03-09: lessons: 1m, reviews 13m, total, 14m
2023-03-16: lessons: 14m, reviews 19m, total, 33m
2023-03-17: lessons: 1m, reviews 25m, total, 26m

6 Likes

This morning, I got 1/3 lvl49 kanji, but one of them I did remember and only missed due to a missing synonym (I put “recommend” instead of “encourage” for 奨), so it was really more like 2/3. In the afternoon, I got 5/8.

Also, this morning, 根本 came up for review, and I managed to guess it due to seeing 根本的 recently in 悪役令嬢が私をいじめるのには訳があるそうです。, so that was cool.

Sat Mar 18 2023:
Time spent: 34m
Reviews completed: 200
Reviews remaining: 532
Reviews in next week: 833 (+301)
Reviews in next month: 1288 (+455)
Accuracy: 75.50% (151/200)
Level: 49

Current item counts:
Apprentice: 324
Guru: 362
Master: 336
Enlightened: 862
Burned: 5713

6 Likes

24 burned today!

5 Likes

Had an unusually large number of burns (20/100) in my second session today. Also below 500 for the first time today.

Sun Mar 19 2023:
Time spent: 33m
Reviews completed: 200
Reviews remaining: 461
Reviews in next week: 773 (+312)
Reviews in next month: 1242 (+469)
Accuracy: 85.00% (170/200)
Level: 49

Current item counts:
Apprentice: 314
Guru: 353
Master: 338
Enlightened: 850
Burned: 5742

12 Likes

Mon Mar 20 2023:
Time spent: 18m
Reviews completed: 100
Reviews remaining: 443
Reviews in next week: 746 (+303)
Reviews in next month: 1226 (+480)
Accuracy: 76.00% (76/100)
Level: 49

Current item counts:
Apprentice: 308
Guru: 356
Master: 332
Enlightened: 853
Burned: 5748

7 Likes