Question regarding speed

WaniKani Requesting Help
83

That’s not necessarily bad for people that have little time. It would just mean that they will have to put effort on certain days more than on others, but having or not having 15 lessons more in a day is negligible in terms of workload anyway (as reviews are the main culprit).

The important thing here is this - if a person is doing WK vanilla and is doing the 15 items per day (for example), will he in the end still achieve the 9.8 days per level average or not as per the math? Meaning, are the 3-4 days of no lessons accounted for in the graphs? Because if they aren’t a person can shave 210 days overall and make his level up time ~6.3 days, which I don’t think is happening/possible. So I would assume at least in some capacity they are.

84

You can’t because there are the SRS levels that need to be respected. 3d10h minimum for radicals to unlock the 2nd batch of kanji from that same level + 3d10h minimum to guru those same kanji, reach 90% and level up. The reason levels 49 to 60 are considered “fast levels” is precisely because you unlock more than 90% of kanji in the 1st batch (aka right after leveling up).At this point, you have most radicals guru’d. That’s the reason this happens. You literally can skip the first 3d10h of waiting and do both the few radicals and all kanji in the same 3d10h.

Well, I don’t see how using the reorder doesn’t allow that either. It actually gives you more freedom.

The 2 main reasons why Wanikani has the vanilla mode and doesn’t have the same tools as the reorder script are:

  • Because they want to avoid people skipping vocab. In general, no one recommends it, but there is always someone that does it.
  • Because the WK team believes that you learning the vocab that uses the kanji you just guru’d right away is the best technique. Sure, it is. But it is limiting.

Basically, the WK team plays it safe like they should very much do. They can’t guarantee that all their users will do the right choices, so they have to choose the safest options for their users by default. But that’s also why they allow the reorder script to exist. Scripts allow flexibility. If you install a script and cheat yourself, it’s on you, not on the WK team :slight_smile:

85

The simulation uses the real dependencies from WK, so the number of lessons per day varies to what is available. There is no reorder in the simulation (yet).

If your rate of new lessons is too small the reordering leads to backlog of lessons, that’s why WK doesn’t allow it as a vanilla option. But even when going slow you could reorder to do a few kanji and the rest vocabulary per day, you can order in different ways during one session. It’s not required, but you can adjust the system as you like.

86

Well everything you have said is exactly my point, that outside of flexibility and making your wkstats look nicer, the reorder doesn’t have a major impact on overall completion times, again if I understand everything correctly.

87

“Look nicer” means more flat, means less time per level on average, means faster :slight_smile:

I would agree that the slower you go the smaller the difference, but if your rate is large enough to finish all items before you unlock the next you will always be faster with reorder, just by putting the reviews that don’t matter time-wise on days where you have less to do.

88

I take faster to mean the overall days spent reaching level 60. Your graph gives 599 days at 15 items a day. Will reorder shorten that? If yes, by how much, relatively speaking?

89

Reoder simply allows you to spread all your lessons out throughout the whole 7 days (of a normal level). Without reorder you have to do all your lessons as soon as you get them (to make sure you do the lessons that are critical for leveling up right away). So instead of having say 20 lessons a day (or whatever the number is) you may have 60 on day one and 40 on day four, and only a few on other days.

90

But the slower you go, the less items you learn per day. In vanilla mode, you could go as little as 4 items a day (these being the 10%+ of kanji) and the unnecessary 3d10h would still be there.

91

I can check later, but it will definitely be shorter with reorder. You are delaying the new radicals and kanji, so when you have an empty review lesson queue you still have to wait 3.5 days until you unlock the next items. An empty queue means unnecessary less speed. If you are slow enough to guru your first new kanji again before your queue is empty then maybe it doesn’t matter. But I think that speed would be 10 lessons or even less per day.

I mean that if you need several months per level then 3 more days don’t matter much :slight_smile:

92

Yeah, true haha

93

OK. Does that mean that the 599 days time-frame is actually inaccurate, and doesn’t account for the days without lessons. Or is it vanilla accurate, just the reorder can speed it up?

94

Let’s say it’s designed to be vanilla accurate, and reordering should speed it up :slight_smile:

It definitely tries to capture the lessons accurately. The simulation doesn’t calculate the time, it actually puts the items as they depend in WK in lesson and review queues and checks every hour if you are doing lessons or reviews and updates the queues. So the results should be equivalent to actually doing it in WK.

95

Thank you for all your helpful replies.

599 is already a great pace considering the number of all items and only doing 15 a day, so really I don’t think that for people that are time-pressed the reorder script will turn out to be that beneficial, but it would be interesting to see how much of an effect it actually has nonetheless, maybe I’m downplaying it.

96

Just dropped by to say that you ignored a big if from that statement: having enough lesson every day to maintain your average.

That mean having enough vocab to cover for the 3.5/4 days or worse 5 days (assuming one review a day) it takes to bring the last kanji to guru and unlocking the next level.
I don’t have the exact dependence tree available like @acm2010 (and I’m not planning to make it), but assuming that all kanjis unlock 3 vocab for the level (kind of a random assumption, but it gives approximately the right number of vocab), also assuming (early levels) ~10 radical, 20 early kanji and 10 later kanji

  • You first spend time doing vocab from the previous level, let’s ignore that since you don’t care about leveling speed and they don’t do anything.
  • Then you get to radical. Let’s say you get there fresh (there was no radical mixed in your last vocab session). For 5/10/15 lessons a day, you are down after 2/1/1 sessions
  • Then you get kanji. You will be done learning those after a total of 6/3/2 sessions

We can already see a difference. At 5 lessons a day, you will get the radicals to guru (and thus new lessons) before you are done with the kanji, but that’s not the case at a higher speed.

Now, let’s say you unlocked also 30 vocab as well upon leveling (those that you already have guru’d the kanji for from previous levels; a very optimistic number)
I will put aside the 5 lessons a day scenario for now, since the new kanjis would get in front of those.

You need 1/2 extra session (for 10/15 a day respectively), so you vocab pile will be down to 20/0 respectively.

So, ok, now all scenario have reached the second batch of kanji.
It takes you 2/1/1 session to go through all of it. (in the 15 lesson case, you also consume 5 extra lesson from vocab)

So, so far, you have had 8/5/5 sessions, you have thus guru’d all radicals, plus 10/10/20 kanjis
That means that your respective lesson pile is now 60 (+30 to come)/ 50 (+30 to come)/ 55 (+0 to come :scream:)

Both 5 and 10 lessons per day are thus fine in this scenario (which, again, is optimistic), but 15 per day will have at least one day at less than full :confused:

Of course, if you skip a day once in a while, it might cover for it… but if you skip reviews AS WELL you are back to having the exact same problem.

tl;dr

Anything higher than 15 lessons per day will get blocked with the vanilla method. 15 lessons per day will probably be blocked once in a while. That means that you cannot achieve an average of >15 lessons per day with the vanilla method, thus my statement does not hold for those speeds.

97

Well this goes back to what we’ve been trying to figure out - how much is reorder actuality helping, especially at lower speeds.

If at 15 you get blocked once in a while, that means that the reorder can have a beneficial effect once in a while only as well.

But it was said that you can achieve max level up speed even without the reorder script, only you then have to do all lessons immediately as they get unlocked, and suffer the bad review counts later (and have a perfect schedule and accuracy as well). In that case reorder helps you to even out your lessons and workload, but the point of contention - it doesn’t let you shorten your level up times, it seems. And that’s at max speed where it’s supposed to be most beneficial. I mean it is, only not for overall speed. But again, max speed is not a concern, its finding the best sweet spot of minimal time/effort and maximum effect. 15 seems to be the cut-off point, if you want to go faster probably you’ll want the script, but lets wait and see what the new graphs will show for 10-15 and above.

Maybe it will have the most noticeable effect for speed actually in the middle of 15 and max speed, since in a way max speed is topped off. But looking at the simulations we already have, there’s just not that much leeway for it to shave off, maybe a couple of months at best I would assume.

This makes sense since at the end all the script does is flatten the time per level stat (vanilla chart columns are fuzzy), but it does just that - the overall time in the upper left stays (or rather should stay relatively) the same.

98

You’re really overthinking this. Instead of worrying about reordering or not reordering and how that impacts level up speed, just use the Lesson Filter script to do the same amount of lessons each day. Consistently do radicals, kanji, and vocab every day and you’ll be fine.

99

If I understand you correctly - not a good idea from a learning perspective compared to vanilla.

100

I don’t see why it’s not a good idea. For me, doing only kanji day after day and then only vocab day after day is tiring and tedious (that’s the default ordering). I think it’s much better to mix in different types of items.

101

Obviously reorder doesn’t achieve anything when you do all the ordered or not ordered stuff completely anyway :slight_smile: But I prefer to do 30 lessons a day instead of 120 without losing a few days.

The results for different batch sizes, accuracy is 100% and you do lessons and reviews when they appear (there are too many variables otherwise)

b
b1277×782 429 KB

  • index: lessons per day
  • column 0: days to level 60 without reorder
  • column 1: days to level 60 with reorder
  • column 2: c0/c1

The largest effect is at 20–25 lessons per day with ~10% difference, or ~40 days faster with reorder.

102

Its not nothing, but its quite small, like I assumed it would be.