Endless Spaced-Repetition Test

[TopCat]

I've read through the manual and stuck on how to configure Scoring & Card Selection to effectively ignore time based tests (hours / days).

Because I work with a small number of cards (10 to 20 at a time) in a short period of time, I'd like to have a test run indefinitely until all cards achieve a target score. The current default settings only run through cards once unless they are incorrect, and then forces me to restart the test.

Does this make sense?
How can I configure the Spaced-Repetition profile to work like this?

 

[mikelove]

I'm afraid you can't do that with spaced repetition since it's specifically trying to review cards at certain intervals.

However, if you switch to 'random' card selection (New Test / Card Selection), you can then set the number of cards to 'endless' and set up your card filters such that cards will be excluded after they hit your target score.

 

[TopCat]

I'm afraid you can't do that with spaced repetition since it's specifically trying to review cards at certain intervals.

However, if you switch to 'random' card selection (New Test / Card Selection), you can then set the number of cards to 'endless' and set up your card filters such that cards will be excluded after they hit your target score.

This seems more like a design constraint vs. a feature. The system assumes people using flashcards are going to only study once per a designated timeframe.

I'm not interested in using a random card selection.
I'd like to use spaced repetition without having to restart a test over and over.

Are there any plans on the product roadmap to provide the option for a score-only spaced repetition test?

 

[mikelove]

The system assumes people using flashcards are going to only study once per a designated timeframe.

But isn't that kind of the point of spaced repetition?

If you want a way to drill cards that you're normally reviewing with spaced repetition, you can do that within our system - just set up a second profile with the settings for that and have it share a "Scorefile" with your spaced repetition one. But if you're reviewing cards over and over again even after you get them correct (and thus their spaced repetition intervals are a longer period of time than the time in which you're proposing to review them) I'm not sure if that's something we could call spaced repetition anymore.

Are there any plans on the product roadmap to provide the option for a score-only spaced repetition test?

It should be possible in 4.0, which allows pretty much completely arbitrary card scheduling systems, but I can't say that totally confidently since I'm not quite clear on what you're asking for here - are these cards that you're reviewing over and over still being 'spaced' in some way? (so you want them to be on like intervals of 30 seconds or 1 minute but then for it to look ahead if it runs out of cards?)

 

[TopCat]

But isn't that kind of the point of spaced repetition?

If you want a way to drill cards that you're normally reviewing with spaced repetition, you can do that within our system - just set up a second profile with the settings for that and have it share a "Scorefile" with your spaced repetition one. But if you're reviewing cards over and over again even after you get them correct (and thus their spaced repetition intervals are a longer period of time than the time in which you're proposing to review them) I'm not sure if that's something we could call spaced repetition anymore.



It should be possible in 4.0, which allows pretty much completely arbitrary card scheduling systems, but I can't say that totally confidently since I'm not quite clear on what you're asking for here - are these cards that you're reviewing over and over still being 'spaced' in some way? (so you want them to be on like intervals of 30 seconds or 1 minute but then for it to look ahead if it runs out of cards?)

I'm not sure what algorithm Pleco uses for Spaced Repetition, but what I'm describing is the Leitner system. The cards move up and down based on proficiency until all cards reach their target level proficiency. My request is to allow Pleco's score system define the target level of proficiency (we set the high end) and each right / wrong response move cards up or down as the score changes until it hits their goal. There is no end to a test until all cards hit their target level proficiency score, and even then, theoretically, you could keep reviewing, mark a card wrong, card moves down, and the process continues.

For simplicity, if Pleco hard set a target level proficiency score into the Leitner option in the app, that would be good enough.

Not sure how other people review flashcards, but I review cards until I learn them completely. Leitner does a good job of driving that, even if I never achieve the target level proficiency score. When the card is learned, it's learned...

EDIT: I hope I'm not come across as whiny and I love your app. Just looking for a little bit more functionality. You guys are great!

 

[mikelove]

I think you could get something close to this behavior now with a 'manual' scoring system.

Start a test with your chosen cards, Random selection, with a Card Filter limiting cards to those within a certain score range, and use a 'manual' scoring system in New Test / Scoring with each correct answer increasing the score by say 100 points and each incorrect answer decreasing it by the same amount.

The behavior you're describing about continuing to review cards at the end of the test would be a little challenging, but you could always make an additional 'review' profile for that which drew from the same scorefile but didn't make any score changes.

But this should be quite easy to do in 4.0.

 

[TopCat]

Thank you for the response. What you're describing is not space repetition. I guess this post should be a feature request for a Leitner spaced repetition option in future releases. Is this possible?

 

[mikelove]

Sorry, I'm not quite understanding how this differs from spaced repetition - how is your proposed system 'spaced'? Do you want the cards to repeat with some sort of extremely short spacing? Wouldn't that prevent or delay you from looping them over and over again?

 

[TopCat]

Take a look at the Leitner spaced repetition system. I did not develop it / it's not my system. It's well documented and has been widely used as a flashcard studying system. I can provide examples if you'd like. PM me if you would like examples.

To respond directly:
Leitner is helpful because it doesn't end a test when you answer all the cards correctly only once. A Leitner test is ongoing until you achieve a target level.

Pleco's spaced repetition test requires you to restart the same test over and over. This is not ideal.

Is adding a Leitner option feasible for a future release?

 

[mikelove]

I'm familiar with Leitner's system, but I believe the original version of Leitner involved reviewing cards at different intervals depending on which box they're in, not running through them over and over again in a single session until they've all hit a certain box. From Wikipedia:

In this method flashcards are sorted into groups according to how well the learner knows each one in the Leitner's learning box. The learners try to recall the solution written on a flashcard. If they succeed, they send the card to the next group. If they fail, they send it back to the first group. Each succeeding group has a longer period of time before the learner is required to revisit the cards. In Leitner's original method, published in his book So lernt man Lernen (How to learn to learn), the schedule of repetition was governed by the size of the partitions in the learning box. These were 1, 2, 5, 8 and 14cm. Only when a partition became full was the learner to review some of the cards it contained, moving them forward or back depending on whether they remembered them.

The system I outlined in my previous post would let you do as you describe - keep repeating the cards over and over again and not end the test until you achieve your target level. It's not technically called "spaced repetition" in Pleco, but neither is it actually "random," those are just the terms we use for that particular option.

The "Spaced Repetition" option in Pleco is specifically for flashcard tests where you're reviewing a large pool of cards over a long period of time and generally only reviewing a particular card once each day - this is how popular Leitner-inspired flashcard apps like Anki and SuperMemo work, and when most flashcard apps talk about "spaced repetition" that's what they mean. That certainly isn't the best system for everyone and it's totally reasonable that you might want a system where you can review a small number of cards over and over like you say, but it doesn't match up with the meaning of "spaced repetition" as we flashcard app makers generally define it.

Is there another flashcard app that implements Leitner in the way that you would like to see it done? Or do you have a way of doing that now with paper flashcards that you could describe to me?

 

[TopCat]

Is there another flashcard app that implements Leitner in the way that you would like to see it done?

Look at the flashcard app Flashonary. It uses Leitner method and behaves the way I'm describing.

 

[mikelove]

Ah, that's an old one, but IIRC they have you take breaks if there are no cards currently due to review - are you skipping through those messages about taking breaks? Or do you want a system in Pleco where you take breaks?

 

[TopCat]

Ah, that's an old one, but IIRC they have you take breaks if there are no cards currently due to review - are you skipping through those messages about taking breaks? Or do you want a system in Pleco where you take breaks?

A popup shows a break, but you can skip right through it. Less barriers to stop & restart is better.

 

[mikelove]

OK.

Anyway, 4.0 is already going to support precise repetition spacing within a test, so if you want to emulate Flashonary in that regard it should indeed be possible in 4.0.

In the meantime, if you choose our deprecated-but-still-functional "Weighted" card selection option in Card Selection / System and configure the other settings as I've described, that will give you almost the same behavior; cards with lower scores will come up more often, as they do with shorter intervals in Flashonary's test.

 

[TopCat]

OK.

Anyway, 4.0 is already going to support precise repetition spacing within a test, so if you want to emulate Flashonary in that regard it should indeed be possible in 4.0.

In the meantime, if you choose our deprecated-but-still-functional "Weighted" card selection option in Card Selection / System and configure the other settings as I've described, that will give you almost the same behavior; cards with lower scores will come up more often, as they do with shorter intervals in Flashonary's test.

Do you guys have a timeframe for launching 4.0? I see it's referenced in the forum going back some years.
The current setup works, but isn't ideal.

 

[mikelove]

No, we try not to announce ETAs for things because they’re so hard to predict.

(great recent blog post on this from a better developer/writer than myself that I’m probably going to be quoting here a lot: https://inessential.com/2019/10/28/no_etas)

 

[TopCat]

I hate to keep beating a dead horse here, but if there is no timeframe on version 4.0, can you please just add Leitner Spaced Repetition method to v3.x.x? I've configured the current app as best I can and it's not providing the type of system that helps me learn best.

An algorithm for Leitner Spaced Repetition is easy to find and implement.

 

[mikelove]

We're really trying not to add anything to 3.x that's just going to be thrown away in 4.0, as this would be - even an "easy" algorithm is still a few days' work which means a few days not working on 4.0.

(we already wasted an inordinate amount of time this summer/fall dealing with iOS 13 / Android 10 plus Google Play's new 64-bit requirement [we already supported 64-bit in our own code, but we had to replace several third-party libraries with totally new / incompatible versions in order to get them on 64-bit], and we're trying not to introduce further delays for anything other than critical 3.x bug fixes)