Resume learning after two-month break : best practices

Anki Help
1

Hi, I need some advice on the best way to resume studying decks which I totally dropped for over two months, leading to thousands of due cards that I cannot study in one or a few sessions - maybe I could but I dont want to inflict this torture to myself.

My best idea so far is to add seventy days to each and every due date and leave everything else unchanged. The task is not as daunting as it sounds - I already tested a KeyBoard Maestro macro that could accomplish it in a few hours, without even tinkering with SQLite 3 which I fear might corrupt the database.

My idea is to return to exactly the same state where I was seventy days ago, when I stopped studying and let the cards accumulate. I realize that my recall of some cards will have dropped significantly and that I cannot expect a cruising speed similar to when I left off but so be it - I do not see a better alternative.

Am I missing something ? Is there a much better plan than I did not see ? Does my protocol have flaws that I neglect today but that can cause disaster in the future ? More generally, what are the drawbacks of my approach that do not yet comprehend ?

I thank you in advance for the collective wisdom and experience of your comments.

All the best. W.

2

https://forums.ankiweb.net/search?q=backlog

Catching Up

I would do this: Create a filtered deck with a backlog and place it in a deck where Maximum reviews/day has a value that is comfortable for you. The review sort order is set to Descending retrievability or Easy cards first or Descending intervals

Deck(N reviews,Descending retrievability)::Filtered Decks(prop:due<=-1)

2 Likes
3

Definitely don’t do that. That spreads the backlog problem to every card in your deck, making everything worse.

You can’t “return to exactly the same state where [you were] seventy days ago” because you don’t have a time machine [do you? :wink:]. As you said, your memories have continued to decay and there’s no virtue in keeping the cards in the same order you would have studied them in if you’d studied them in the last couple months.

A better path forward is to get your overdue cards out of the way so you can resume current studying while you also dig your way out of the backlog. I use a catch-up deck like this – Adressing Backlog - #2 by Danika_Dakika – and it works great.

2 Likes
4

When I had a similar situation I used to use a method called 'learn/filter by urgency '. I chose the cards which seemed to me most urgent to repeat.

Try to learn the cards with the lowest intervals first. And then increase the filter for the interval value slowly when you get fewer cards.The cards with the highest intervals have the lowest priority because you may still remember them. If you get too much cards decrease the amount of cards with a ‘due’ value condition.

So you have to create filtered decks. For example:

is:due -is:learn (prop:ivl<=1 prop:due<=0)

If this filtered deck has too few cards you could increase the amount of cards you want to repeat with:

is:due -is:learn ((prop:ivl<=1 prop:due<=0) or (prop:ivl<=2 prop:due<=-4))

And then:
is:due -is:learn ((prop:ivl<=1 prop:due<=0) or (prop:ivl<=2 prop:due<=-4) or (prop:ivl<=3 prop:due<=-6))

Long time ago I had a similar problem like you, i. e. a really big backlog. My first filter was:

deck:"Babbel" is:due -is:learn ((prop:ivl<=1 prop:due<=0) or (prop:ivl<=2 prop:due<=-4) or (prop:ivl<=3 prop:due<=-6) or (prop:ivl<=4 prop:due<=-8) or (prop:ivl<=7 prop:due<=-14) or (prop:ivl<=11 prop:due<=-22))

And my second filter which I created when the first one had no more results was:

deck:"Babbel" is:due -is:learn ((prop:ivl<=1 prop:due<=-1) or (prop:ivl<=2 prop:due<=-2) or (prop:ivl<=3 prop:due<=-3) or (prop:ivl<=4 prop:due<=-4) or (prop:ivl<=5 prop:due<=-6) or (prop:ivl<=6 prop:due<=-8) or (prop:ivl<=7 prop:due<=-12) or (prop:ivl<=8 prop:due<=-16) or (prop:ivl<=10 prop:due<=-24) or (prop:ivl<=14 prop:due<=-35) or (prop:due<=-90))

Third filter:
deck:"Babbel" is:due -is:learn ((prop:ivl<=2 prop:due<=-1) or (prop:ivl<=4 prop:due<=-2) or (prop:ivl<=6 prop:due<=-4) or (prop:ivl<=8 prop:due<=-6) or (prop:ivl<=14 prop:due<=-11) or (prop:ivl<=21 prop:due<=-21) or (prop:ivl<=30 prop:due<=-40) or (prop:due<=-60))

And so on. Try to decide: What are the most urgent cards? And then add an according rule. E. g., in the third filter I decided that it’s time now to repeat the cards with a due date more than 60 days ago. The interval value should not matter anymore.

Everybody should adjust the parameters above to his own needs but this worked for me to get rid of the backlog after some months. (You have to be patient…!)

Thorsten

5

This is the same as saying focus on learning new cards and only then on review cards. This advice is harmful. Do the opposite.

1 Like
6

If I have a backlog of e. g. 2000 cards it makes no sense for me to learn new cards at all. Instead, I repeat the cards I have already learned until the backlog is empty, i. e. until there are no overdues cards anymore. But others may handle this differently.

Do what you want. The above method worked for me several times until today. But of course I did not add new cards.

BTW My advice should be harmful? LOL.

1 Like
7

I know everyone has very strong opinion on the correct way to do this, but in my experience in practice, it doesn’t really matter how you do them and in what order. You can do the full review session, you can keep increasing the review limit as much as you can for the day, or you can stop mid-way. You can do these from easiest to hardest if you’re motivated by the initial speed, or you can it from the hardest to easiest if you’re motivated by getting the hard part over with as soon as possible. You can do them in random order if you want a steady pace throughout the whole thing. You can even mix and match these depending on how you feel at the moment (or that day, depending on how big the backlog is).

You just need to accept that you should stop with the new ones for a while and that your workload is not magically going to go away once you’re done with the backlog – many of these cards will be forgotten and will have intervals as if they were new (or worse), i.e. many of these will feel new to you at this point (but more frustrating, I think).

In the end, the big brains behind the ordering can be “outsmarted” by doing the reviews 15 minutes longer per day than with some “optimal” algorithm for clearing the backlog. I see people so often approach this as a min-max game, when in reality all that time might be better spent using a suboptimal solution that just feels better for you at the moment. I’ve always found the exhaustion a more limiting factor than strictly the minutes count.

3 Likes
8

I am also battling a backlog right now. I am experimenting with the “Ascending retrievability” review sort order. My thinking is that the lowest retrievability cards are most in need of resuscitation. But I’m not using a filtered deck; I simply stop studying when I’m tired out.

That generally seems like good advice. I will note, however, that sometimes doing a few new cards can help with motivation—it feels like I am learning new things instead of merely rehashing old things.

That is always good to keep in mind—doing the reviews is more important than analyzing how reviews could be done. :slight_smile:

9

I think you misunderstood the analogy.

10

Since there are many individual features in the case of backlog: how long has the backlog been going on, your Desired retention, and how much Desired retention is far from optimal, and how quickly do you plan to eliminate the backlog. It makes sense to use FSRS simulator (experimental) and see what is best for you.

11

Yes. But: Learning makes you tired. So the sequence of cards could be more important to somebody than to another person. Therefore, if I had a big backlog today (hopefully not) I would try to filter out the cards in the backlog so that the results would be a mixture of difficult and easy cards. (I did not practise this strategy in the past, but I practise a similar strategy to learn in advance.)

My strategy would be: Trying to repeat a bunch of difficult cards (low intervals, low easyness), e. g. 80 cards. And then, if there is some energy left in my head, repeating only cards with bigger intervals where the possibility is high that I will remember them immediately.

I use a similar strategy to avoid getting a new backlog. If I see in the statistic that I will get many cards in the next 1-5 days (e.g. more than 100 per day) I filter out the cards in a descending sequence of intervals. Those cards are easy for me to repeat (otherwise they hadn’t a high interval). I can repeat some of them (e.g. 150) on a day where I have enough time and energy. Example: If a card has an interval of 30 days and it is due in the next 5 days it does not matter if I learn this card e. g. 5 days in advance.

With the strategy described above I can lower the curve in the statistic so that I will expect only e. g. 70 cards per day instead of over 100. If I had a big backlog again I would do the same for the past.

12

This topic was automatically closed 30 days after the last reply. New replies are no longer allowed.