Documenting issues related to inconsistencies in the Queue Behaviour

There have been a few edge cases where the queue is not behaving the way it should, for example :

  1. Cards not adding to the queue automatically
  2. Queue count inconsistent between devices

This is to try and collect as much information as possible regarding these bugs. Thank you for the reports!

related : Problem with Portals/Queue

Other issues related to the queue behaviour that I found in GitHub :

  1. Daily Practice Time is Under-Counted · Issue #537 · remnoteio/remnote-issues · GitHub
  2. Embedded queues show the same card more times than they should. · Issue #179 · remnoteio/remnote-issues · GitHub

Hello everyone, I’ve been using RemNote on both my Mac and an Android phone. I’ve often found that the queue progress (for example 43/150 questions) is lost when I switch from one device to the other. This happens even when I have ensured that a sync is completed on both devices. Occasionally, the queue progress updates itself without my having done any questions. What can I do about this?

This is keeping me from doing my cards effectively, because I have to keep importing them in small batches to Anki. Would appreciate if anyone can suggest a long-term solution.

I have the same issue. Reloading the page, making sure everything is saved and re-opening the RemNote app on the Mac solves this most of the time for me.

The same problem. In Account > Card table I can see that today was practiced a few cards but queue shows progress 0. Usually it happens when I use my android app, but sometimes after the web page reload.

About 3 months ago, I started to encounter some issues with the SRS system in RemNote, such as:

  • Different number of cards in queue on mobile and computer (even after syncing)
  • Incorrect scheduling of some cards, e.g. having to review one card after a 2 day interval instead of a 1 month interval
  • Some cards not appearing in queue at all. This was a big deal to me and something that made me switch back to Anki for SRS.

Could anybody with a larger database in RemNote comment on whether these issues have already been resolved and whether the SRS is reliable in its current state? BTW, I started to experience these problems only after creating around 10,000 Rem and having approximately 4,000 cards in the queue.

I’m not using RemNote for a long time (3 1/2 months), but we are facing now some big issues in SRS (like cards being schedule every day infinitely, or intervals when using Anki SM-2 scheduler type being almost 10x what they should be). I am, and others also, reporting the bugs for the team, and really expect they will focus more narrowly to this so important thing. Sadly, not many users care sufficiently to SRS, so there are few reports on these issues, what I feel makes the team focus on other issues (of minor importance, in my point of view); also, the system is being updated very fastly, what imposes a great challenge in maintaining it stable. But the team is very competent, and surely they will see the importance of it and work on it.

So, it is not very trustworthy by now; but I really expect it will soon be. For now I am reluctant on leaving Anki, because of its very trustworthy and stable SRS. But seeing the potential of RemNote, I’ll keep trying it, side by side with Anki, and I’ll keep reporting strange behavior.

1 Like

I love that RemNote is being updated at a great pace and some of the new features really are amazing (PDF annotation). However, I feel that changes to the scheduling system should be very thoroughly tested before rolling them out to the public. A lot of the bugs I personally encountered with scheduling only started appearing after the custom schedulers update (which introduced the Anki SM-2 algorithm). Cards disappearing or not being scheduled properly combined with some syncing issues can really increase your stress levels especially if you have an important exam coming up!

For now, I stick to the default scheduler and only use RemNote on a single device. This way, everything seems to be working just fine.

1 Like

Good to know things are ok to you. I have had problems when unwittingly switched to Anki SM-2 scheduler during a time, but going back to the default apparently solved the problem; and more recently cards learned after Jan 30rd entered in an eternal loop being presented every day and never increasing the intervals, making it impossible to me to continue my reviews for now (I’m giving a break until next update). The team was already reported. Hope this will be fixed soon.

1 Like

I have similar issues. On browser, mobile and desktop app(Linux) the queue contain different amount of cards scheduled for review. Moreover if I go to the card section of the website some cards are indicated as inactive and not in queue (I cannot toggle them). Seems like for now the queue is unreliable

2 Likes

I’ve been reporting since the end of January issues with Anki-SM2 scheduler. They remain unsolved by now. Intervals get too long, sometimes twice, sometimes six fold what they should be. Sadly the queue and scheduling remain unreliable. Probably they will have to undergo a full revision of the SRS system.

2 Likes

Thanks for the report. We will definitely look into this.

Thanks @Karthikk , I would ask you to help me convincing the team to finally look at this.

Concerning the messy Anki SM-2 algorithm, I’ve also posted Too long scheduling intervals (Anki SM-2 shceduler) · Issue #1062 · remnoteio/remnote-issues · GitHub, also dealing with anomalous extremely large intervals, and Pressing "Forgot" while in learning phase (Anki-SM2 scheduler) is affecting card's ease · Issue #1067 · remnoteio/remnote-issues · GitHub, showing that mistakes in the learning phase are wrongly affecting the ease of the card.

And a suggestion regarding Anki SM-2 scheduler: like adopted in Anki some time ago, It would be much better if, when pressing “Hard” in a card during the learning phase, the present learning step were repeated, instead of being considered completed.

Like is documented in Anki v2 scheduler:

Hard repeats the current step after the first step, and is the average of Again and Good on the first step.

For example, if I have set 2 learning steps (1 day and 6 days, for example), if in the first time of the card in my Queue I rate Solid, but next day I rate “Hard”, instead of graduating and jumping to the exponential phase with a 6 day interval, this second learning step should be repeated; I would return to 1 day interval and would do again the second learning step next day; only when pressing Solid in a given step would I jump to the next one.

2 Likes

Based on superficial observation, I agree that many of the queue related issues pop up while using multiple devices. But it doesn’t seem to be happening for everyone, only a small subset of the users. I wish to understand why this happens to a certain subset, so that we can find the bug and fix it. Any idea about this?

I wish I could help with troubleshooting this but I’m unsure what the cause was, it seemed completely random, apart from the fact that it started after the SM-2 algorithm update. If there is some low chance of this bug happening, maybe it only affected users with larger amount of cards in their queue? I had created several thousand cards in total and completed around 200-300 daily.

1 Like

Definitely worth keeping in mind, as we get more documentation from the users.
So we have two potential leads:

  1. Behaviour started after the SM-2 algori update (but that should mean it is globally affecting everyone)
  2. Could be happening to heavy queue users.

Thanks for providing this info, will have a lookout for this. Hopefully we find the bug soon, and get it fixed for you guys.

1 Like

Another thing to consider is that it‘s a bit difficult to notice bugs in the scheduler. If you for example don‘t know how the scheduler works, you wouldn‘t be able to tell that your intervals are wrong. Some people also have their interval indicator turned off which means that they will hardly recognise a wrong interval and generally while studying you don‘t really pay much attention to these things.

1 Like

Quite an interesting proposition, I guess someone just has to manually track this intentionally, to know what’s really happening. But I am guessing (and hoping) everything is actually working as intended, and the inconsistencies are just edge cases.
Also, if it was more widespread, I am quite sure that lot more might have noticed it, especially those preparing with a timeline.

Yes @Karthikk, maybe I’m the unlucky guy, and something went wrong in my DB, something that does not happen to others. It would be much better if this is the case. But we should consider also that probably few users use the advanced Anki SM-2 scheduler instead of the default, and many of them, as I did in the past, as soon as they find the inconsistencies, changed back to the default, that looks much more stable.

Anyway, I think that is not difficult to track the issue; the only factors that should affect the interval of a card when reviewing are those mentioned in the documentation: last interval, overdueness, ease, interval modifier, maximum interval (if it is the case), and button pressed. Nothing else. If the team could look at my database, as in my case the errors are absolutely consistent and recurrent, maybe they would more easily discover what caused the strange behavior. I promise you they will not have to try more than 20 card reviews in this scheduler before they find some big issues.

For now, I gave up of Anki SM-2 scheduler, as I was at risk of messing up my knowledge base. I started using Anki SM-2 because I wanted some few customizations; e.g. it looks meaningless to me that during learning pressing Solid or Easy have the same effect (24h interval); it would sound much more logical that Easy would space it out a little more (4 days, for example); and I was also trying an initial higher jump after the second successful review (to 6 days, instead of 3). Nothing that big. It takes a lot of time and stress to identify these inconsistencies, report them, and so on. It is much cheaper to change back to the default and forget once and for all this so called Anki SM-2 scheduler, that stressed me a lot the two times I tried it (2 months ago, and again a few days ago).

Even it happens to only one person, it shouldn’t happen. And few things are so important in RemNote as the queue and the scheduler. I do still think a full review of Anki SM-2 scheduler would be a good idea. It’s true that few people reported problems; but it is also true that few people also said: “oh, I’ve been using Anki SM-2 scheduler for a while and it is working perfectly”, or something like that.

Thanks once more!

4 Likes

A very very reasonable feedback @hugo, thank you so much!
Will certainly look into this and see what’s happening.

@Martin might be able to give us more answers here.

I think I saw an issue report of @anisha, and as I know she is very active in the forum, I would like to hear from her if she noted long intervals being scheduled when using Anki SM-2 scheduler, like I saw and reported above, or if she thinks the Anki SM-2 scheduler is working well for her.

1 Like