Learned cards are new after filtered deck

We have a situation where 50 cards—the entire new cards for a single day—are showing up as new cards even though they show the study history from the day they were learned. For example, here is the info screen of one of the problem cards:

Notice the discrepancy between the review count (0) and the record of reviews, along with time spent. This happened to all 50 cards that were learned on that day, so I think something happened that we weren’t expecting. I will try to recount what we did as best as I can recall.

First, note that we were using desktop 2.1.50 beta 5 (so perhaps a beta issue?) when we added and did the study on these cards. We needed to add new cards to keep up with our schedule, but we were being limited by the maximum daily review limit. We did not want to increase the review limit because then the increase would be taken up by past due reviews, so instead I added the new cards for that day by way of custom study, set up this way:

(I would have rather used “Increase today’s new card limit” as what seemed the obvious choice, but it does not add new cards in this case because the review limit still means no new cards would be added.)

We studied all of the new cards in that filtered deck, then normally I would just delete the filtered deck right after the study so they go back to their main deck. As far as I know that had been working fine on previous days. This time, I think I forgot to delete the filtered deck right after the study, and I noticed it and deleted the filtered deck on the next day, and if I remember correctly, I did that on an iPad. If so, then it might be relevant that that iPad is running a release version, not the beta, of AnkiMobile.

In any case, I did not notice anything unusual at the time I deleted the deck, and my daughter worked through the reviews scheduled for the day. And only now, on the second day since the 50 were learned, have I noticed that they are all new again.

So I am posting for two purposes: first, to report that I suspect that something went wrong such that the learning cards from the filtered deck lost their learning status when the filtered deck was deleted. Possibly the interchange between beta 5 on desktop and release on iPad is related. Or possibly, I have not understood something, as I have never used filtered decks before.

The other purpose in my writing is to ask what is the best way we can repair the damage. This kind of use of Anki is not my usual habit. Normally, I approach it as a long-term study method where there is no target end date and I can rely on the algorithm to just guide me along a sane study schedule. Here, we have a definite end date, and through an unfortunate misunderstanding on timing, we have a limited time to study a large number of cards. So we have to study all the cards as best as we are able, and I am also trying to maximize what can be done without pushing my young daughter too hard.

I am hopeful that we can do something to reset the cards to their learning position. For one thing, it may be helpful that there is a single tag associated with all of these cards and only these cards, so maybe there is some bit of code that we can target these cards with in the debug console. The tag is Joshua_ch22 and all of them should be on the 1d learning step in a learning schedule of 1m 10m 1d 2d 4d with what should have been 2022-03-06 as the due date.

For the moment, I will suspend these cards until I can hear something, so my daughter can continue with her other reviews.

If I remember correctly, this was a somewhat intended behaviour of the V1 scheduler. It has been fixed with the V2 version. If you are using the V1 scheduler, you should consider upgrading (please read the documentation first; I’m just an Anki user myself)

Anki 2.1 scheduler - faqs

I cannot use the debug console so I’m not able help on that regard.
Maybe you could fix it manually, using a filtered deck. A bit clunky, but since they are only 50 cards it should be quite doable.

  • If you’re using V1, upgrade to V2 (to assure that the cards retain their “learning” status whatever you do with the filtered deck)
  • unsuspend those cards, then create a filtered deck with these settings:
    Search: “tag:Joshua_ch22 is:new rated:5”; Reschedule based on my answers in this deck: yes;
    (“is:new rated:5” is only relevant if these cards belong to notes with multiple cards)
  • click ‘good’ on every card as many times as needed to empty the deck (should be twice per card); all the cards should now be at the 1d learning step
  • open the filtered deck options, delete “is:new”, then click on “rebuild”
  • your daughter should now be able to study the cards, inside this filtered deck, at their correct learning step
    (she needs to study them inside the filtered deck because technically they’d be due tomorrow, using a filtered deck allows you to see them 1 day ahead; after your daughter has reviewed all the cards one more time, you can actually delete the filtered deck, and study the cards on the appropriate day in the original deck)

(EDIT: I see you are a frequent poster and quite an experienced Anki user. Sorry for explaining things you probably already know, about V1, V2 etc. I’ll keep them there for reference)


No worries about my being an “experienced Anki user.” I am proficient in the aspects of Anki that I use, but I don’t think of myself as an expert, and in this case, I did come looking for advice, so your advice is welcome. In fact, this most recent project was the first time I have ever even tried to use a filtered deck, not having needed it before.

I followed the gist of your recommendations, and I am satisfied with the results. There was a step or two that I hadn’t thought of that proved beneficial. The debug console may have done something faster or perhaps repaired damage in some cleaner way (maybe), but as it is, my daughter is now caught up and on her way.

Thank you for taking the time to answer.

As to the scheduler, I am using the v3 scheduler (v1 is not even supported anymore in 2.1.50). There may be some issue here that we can still debug, but for my personal needs this issue is taken care of for the moment.

When the top and bottom sections of the card info screen don’t match, it’s usually because of a sync conflict. If a card is marked as modified on multiple devices and then you sync, whichever modification was made most recently wins. If an unstudied card was marked as modified by eg. flagging it or building or emptying a filtered deck, then the previous review state it had could be lost on a sync.