Curious Q: What is the minimum negative?

How many reviews and lessons are needed to cause a negative growth of either and in what interval?

Sounds like an easy Q - it’s not. Every lesson and level unlock new lessons (and reviews). Every review unlocks another as well (1+1 if incorrect). I.e. Number and Interval of Review and Lesson.

I got this question quite a lot from people looking at my notoriously high stack of unfinished lessons and reviews, but now I got curious myself. What are the numbers? I figured someone might have done the math. If not, I’ll see if I can come up with a formula.

An example could look something like x=L^N+((R^Nb)*incorrect_count) or simply 6 lessons&69 reviews every 9 hours.

Cheers Screenshot 2020-10-27 at 06.00.42

1 Like

Sorry, run that one past me again?

What’s getting negative growth? How’s that caused by having reviews to do?


The only way to get your lessons to increase like that is by skipping vocab. Otherwise they will stay static at whatever level you’re on as long as you don’t do new ones.



I think part of the issue is that you have to factor in the level of the cards since apprentice, master, enlightened, mastered, and burned all return at different frequencies. Many users like to keep their apprentice count at around 100 to prevent major growth. The main way to decrease your numbers is to do large batches of reviews fairly correctly. I got my current count to firmly decrease for a few days when I had about 600 reviews by doing 300 over the course of a day. This was with about 60-80% accuracy. I had 600 reviews Sunday, 450ish max yesterday, and 450ish today (before I’ve done any work on them).


I’m honestly curious to see if anyone comes up with an equation, but I’m not entirely sure if it’s possible as it also depends on accuracy rate as to whether a review is passed or not, which is not a set value and is rather hard to predict. It also, I would imagine, change depending on the review level, how long ago you saw it, and other environmental factors that would be hard to predict. But hey, I’m not a mathematician, and it would be cool to be proved wrong in this case.

The easiest answer to this question is not an amount, but having a regular schedule - if you’re always doing the same amount of reviews and lessons a day and maintain a high accuracy, then your counts will go down, if not immediately then at least eventually.


Hey @Kumirei and @prouleau I know minimal things about coding, but I think it involves math sometimes. Any idea what a formula would look like for this? I think one of you made a forecaster before it became a part of the dashboard, didn’t you?



Isn’t this oxymoronic? Or just plain contradictory? That wouldn’t be growing that would receding … You can’t grow backwards.

Things do not double if you get them wrong, they merely go backwards on the scale of appr-guru-master, etc. There is no infinite increase in reviews, that number will always be finite.
So, as I understand it, you can stagnate, (you could have every single lesson in your review queue at once, for example) but you can’t dig yourself into an infinitely increasing hole of “backwards progress”.

Right? Just my thoughts as I try to understand what on earth is going on

It’s an economics term. Means “shrinking”, basically.


So … how do you shrink the number of reviews and lessons?.. As in… by doing them?

I guess if I don’t understand the terms I should stay out of it, sorry, I was just confused :sweat_smile:

1 Like

Yeah, that’s where I got lost too. I think the concern is that doing lessons causes reviews, and doing reveiws causes lessons, so it seems like it should be a runaway system, buuuut… it’s really not.

@CDR-Strawberry seems to understand, though - perhaps you can explain it to the class.


I believe this person is asking for the minimum number of reviews one would have to do to see review count decrease day-to-day instead of increase. For example, having done 300 reviews on Sunday, 50 on Monday, and 200 today has made my review count go from 600ish->450->300something. I’m meeting more than the negative minimum since my reviews are shrinking. In my case, if we figured out an answer to this person’s question, it would mean figuring out the number of reviews I would need to do daily to make my numbers go from 600->599->598, or the minimum amount of reviews to keep my review count from to go about WK in the most leisurely and slowest way while still making progress.

Doing any lessons would make reviews incredibly difficult to figure out, so I’m not going to try that bit. Although assuming you do one lesson a day, you could factor that into the equation, but then you’d run into trouble when you guru anything you haven’t previously guru’ed because that would generate an unknown amount of lessons. I kind of find it an interesting question, but I don’t have the time and energy to first reteach myself the math necessary for this and second to actually make this formula.


A formula is nice and all, but in your case I’d instead recommend using the reorder script you are already using to only do new vocab lessons for the next month or 3 (if you don’t want to rush it then you should go with 10 per day max) and then do around 200 reviews per day, as tat will keep your reviews low.

Also. A general formula with the only input being #reviews and #lessons is impossible, as it always depends on what stage your reviews every one of your items are at, what kind of lessons you are doing (vocab kanji or radicals), what level you are on, how many kanji you are away from leveling up, how often you get which review wrong etc.

But at the point you are at, why not just start over. You messed yourself up big by severely misusing the reorder script. The only way to get off this mountain is like I said before to just stop doing new kanji and radical lessons for up to 3 months

1 Like

One approach to this might be to look at the total number of reviews that would be required to burn all of your current active cards and use that as an analog for how many reviews you have currently. The change in your number of reviews required to burn your active items is roughly determined by this formula:

ΔReviews = 2 * #incorrect - #correct + 8 #Lessons
So basically, as long as you get >66% correct on your reviews and don’t do lessons, your total review count (and at least eventually your daily review count) will decrease. You can add some lessons into the mix while keeping your total review count decreasing as long as ΔReviews is negative. The more reviews you do and the better your accuracy, the more lessons you can do.

For lessons, as long as you’ve stopped using reorder scripts, you should stop getting new lessons pretty quickly. So that should just decrease over time as you do lessons. Doing vocab lessons can’t lead to more lessons so those would be a be a good thing to start on as Sinom suggested, if you insist on keeping with the reorder script. But I suspect removing the reorder script will start you out with a ton of vocab lessons anyways, as that’s the only thing you could be skipping while progressing through levels.



I imagine something like my Expected Daily script would give a figure for this. That script doesn’t take either accuracy or lessons into consideration, though


This is not me. I have no idea of what kind of forecast you have in mind.

I think they mean rfindley’s Ultimate Timeline

1 Like

Honestly I don’t really know either because I didn’t use one before they integrated it into the dashboard.

Maybe? Your expected daily sounds like it might’ve been what I was thinking of though.