Evidence on the Priority Queue Issue
Following Littlefish strategies I’ve been working on a new collection since a couple of days ago in order to find out if the previously mention priority queue issue affects every user, and not only me because of the way I like to import Q&A pairs (trough XML structured Q&A by theme, text Q&A and transfers of collections). What happens normally to anyone using SuperMemo in a more orthodox way.
This pictures show the result of my research on the issue. I get a 56% priority on a folder created when just starting the collection (which belong to the parent category) things do get messier as time goes by. The moment I saw this behavior was three months ago, I found a folder with a priority way superior to 26% – high priority get lower percentage values in SuperMemo. In my collection items are autopostponed if they get a priority of 30% or lower so this folder I found was making some real high priority question get postponed.
Previous post on the issue: Dimissed elements interfere high priority items and Priority queue issue
Hi, you asked on twitter to post priority queue screens. Here’s mine: http://vroobelek.iq.pl/temp/sm2006-priority-queue.png
I scrolled through the list and found first yellow – dismissed element on position 6796 out of 12374. But you can see that there are many high priority elements which were remembered and then forgotten.
Thanks for your investigation. I never paid attention to priority queue and the impact of forgotten/dismissed elements. Now I chose the following way to get them out of the top priority list:
1. filter priority queue to see not memorized (child->filter)
2. “process browser -> priority -> spread” between 70 and 100.
Correction: priority -> spread works only for forgotten elements and not for dismissed – their priorities won’t change.
Sorry for now showing my priority queue. I’m not using it too much (time to catch up :-)).
But Yeah, I think this should be considered as bug since Dismissed elements are those you’re never supposed to go back to in the outstanding queue and daily repetitions. Although it is possible to undismiss it, I don’t think anybody does that.
I suggest reporting it to SM help.
The same might go for forgotten items/topics, but in this case I’m not totally sure whether they should be ignored in the priority queue, since a user might want to temporariliy hit forget on them and want to relearnd them just to restart the learning process from scratch.
I looked at my priority queue, and within the top 3,000 items, I had about 10-15 dismissed items.
So does the priority queue determine the order of flashcards or incremental reading topics?
For example, you have to review Flashcards A, B, C, D and E in one day. This is the priority value for each card (Assuming that there are other cards in the database, not only A through E):
A = 5
B = 2
C = 8
D = 1
E = 11
So does the priority queue decide that the order would be: D, B, A, C, E ? I want to make sure I understand it properly.
IF the above is true, then if B were not an item but a dismissed element, does that dismissed element simply become part of the dead computerized “landscape?” The dismissed element exists within the database but it is ignored if/when encountered by the Supermemo machine? If so, I’m not that concerned. Maybe it causes Supermemo to move a little slower, but because computers are very powerful, I think any loss of efficiency would be very small.
Again, my understanding could be flawed.
In fact Nattan, what I’m saying here is that because of a flaw in design some elements could be postponed or filtered out because there would be dismissed elements displacing them and they shouldn’t. I’ve taken some time in order to see if the same flaw exists currently and it does or if this problem only is apparent after a long period of time.
Well no, this problem happens rather easy, if you import many q&A pairs, or by using XML. I wonder if there are enough people using it?, I do in order to import question data bank info in other to track them down with out scanning, I prefer XML as you can design the whole structure easily outside SuperMemo. SuperMemo 2008 makes contents building much easy this days so probably I just got used to using XML – I should probably use incremental picture reading now.
Thankfully, SuperMemo author have acknowledge it as a error of design and hopefully they will remove this or hacked altogether, but, like you probably already know, this might take “some” time.