Minimum related review distance


#1

Is there a script to tweak the review order such that the reading and meaning for items are at least a certain minimum distance apart? I keep noticing reviews either coming up again immediately after failure, or with the other half nearby. Yes it may be “random”, but when a related item is close in the review it’s not really stressing me that hard, as I just saw it moments ago.

This is a different ask than the review reordering/sorting scripts I see out there. Really just looking to tweak random ordering to be slightly less random.


#2

As far as I’m aware a script like this doesn’t exist. This should really be done by the WaniKani team as a fully fledged feature though. As you said, immediately repeating a review you just got wrong makes no sense and hurts the learning process.


#3

Does it really? :thinking:

You could also argue that by getting it wrong, you get a call for attention to that item. Getting it right after will use that same momentum to actually learn the word.


#4

Personally, I may not remember an item 1 minute later, let alone 4 hours later. So if I review it 5 seconds after getting it wrong I may remember it, but re-reviewing it later in the same session may result in me getting it wrong again. I think that’s a good thing since it forces me to remember it over a somewhat longer interval. Sure, I’d get the item right eventually, but I think this would result in fewer failures at the 4 hour interval for me since I will have already reinforced the item over shorter intervals.


#5

But I’m pretty sure this line of logic doesn’t really support the argument that it is harmful to the learning process. It may be frustrating to some people, but if one is not properly re-reviewing an item they’ve gotten wrong, then it doesn’t matter if you let it wait until later to review that item or not. That individual simply hasn’t taken the time to learn that problem item.The SRS relies on the premise that seeing a troubled item many times over allows one to take the steps to take the time to learn the item.

Regarding the OP, I believe when one does a review session, the system pulls a fixed portion of items to review at a time and presents them all until the user completes them all before pulling a new batch of items. Otherwise one with 100 review items could possibly be asked the reading for an item as their first review, and then be asked the meaning after the other 99 items have be given… I’m not sure if there is a script that could accomplish this, but it could result in review items that are marked incorrect to be left unresolved before the end of a review session.


#6

I might be saying bs, but I tried to skip some specific reviews during some reviewings to see how many items it would show me. I would only do the reading/meaning and skip the other part. The number of items shown was around 10 :thinking:


#7

That sounds about right. The part you highlighted refers to the hypothetical possibility of a review item’s meaning and reading not being controlled in the way that they are currently. Since the OP wants to toy with the distance between an incorrect review item and the next time one would see it again, then it would most certainly break the system of how reviews are handled by the system. That’s all.


#8

I believe I covered that. I implied that by getting wrong again later in the same session, I would have to review it yet again.


#9

Ugh, I didn’t constrain the problem statement enough. I’ve no desire to chuck the item into the bowels of the review queue, and fundamentally change the WK algorithms. I merely don’t want it next, or second, etc, pick your preferred minimum distance. Positioning it later in the batch of items currently being reviewed would suffice.

When I fail an item, it is because I couldn’t recall outright or reestablish the mnemonic connection. I then read the full mnemonic over again, multiple times, trying to identify and correct the mistake I made. If I am immediately quizzed afterward, this is testing my reading comprehension and active working memory, something I already do myself in studying the item. I have no problems here, and always get it 100%. If however, I put some different items in between, and overwrite my working memory with something new, I feel the recall takes more effort, effectively going further back than my active working memory. How far? I don’t know. My speculation is a more beneficial distance.

Guess I’ll start poking around in the JS, and see whether something like this is even possible.


#10

I was just explaining the functionality of the system. I wasn’t meaning to imply that’s what you were intending.


#11

I was just pointing out how your first and second statements didn’t align to make a point. That’s was all.


#12

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