Yeah, I deleted it when I realized that
Then I’ll be satirical and demand a sort order using the phases of moon.
“what do u mean it’s illogical? It’s a user making a genuine request. If there’s demand, we add it right?”
I have always felt that I am what I call “doom-reviewing” a set of cards that are not leech cards but still way harder than other simpler cards. This piles up these simple cards and makes them eventually harder to recall so the queue is overwhelmed, but this is my subjective feeling. I don’t know how this pans out on actual simulations.
But is sorting by Difficulty Ascending really much more effective than Retrievability (Relative Overdueness 
I’d be fine with removing a bunch because I think most of them are pretty niche needs, and most of them are just vestiges of early Anki when they probably just chose some sort options arbitrarily based on what was easy to implement, but I don’t find any of them hard to understand. I don’t see the need to remove any. What overhead are they creating?
Well, let’s try our luck
@dae Recommend sorting by relative overdueness.md by Expertium · Pull Request #634 · open-spaced-repetition/fsrs4anki · GitHub
According to Jarrett’s simulations, the following sort orders are the worst for dealing with a backlog:
- Due Date Ascending
- Difficulty Descending
- Add Order Ascending
Let’s remove them and add Retrievability Descending aka Reverse Relative Overdueness instead. It will be a net decrease in the number of sort orders meaning a net decrease in the amount of confusion, and the new sort order is more useful than those.
EDIT: an even more radical solution, only keep these:
- Difficulty Ascending
- Retrievability Descending (Reverse Relative Overdueness)
- Add Order Descending
- Interval Ascending
- Retrievability Ascending (Relative Overdueness)
And remove the rest. These are the best for handling backlogs. Right now there are 11 (!!!) different sort orders. This is extremely beginner-unfriendly.
Also, see this: Ordering Request: Reverse Relative Overdueness - #86 by rich70521
EDIT: there are too many comments here, I made a clean post: Improving sort orders
4 Likes
Lol exactly, and Due Date Ascending is basically the default iirc.
Slightly more, not much
The two I’ll mention are “Card type, then random” which is literally random but we’re trying to seperate siblings. There can be a single “Shuffle” feature that isn’t truly random and takes into account siblings, etc. I don’t think Spotify would have multiple shuffle features, one for each type of randomness.
I’ll definitely take you all’s word for it, but is this really such a headache? Are you guys getting bombarded with questions or something?
Oh, you’re right. That’s something I’ve experienced. I think I briefly started pressing Again on all such cards until I’m at the end of queue so I cam finish them at the end. It’s back in the SM2 era story. Now I get why this sort order would be useful.
Ah ok, I actually went a while without updating Anki and I’m only a few weeks into using FSRS and there are a lot of new sorts on the main decks. The ones I’m used to are the ones that are available when you are making a custom deck. Those are the OGs.
Bro, looks at this:
Imagine that you are new user who just stopped using Duolingo and downloaded Anki 5 minutes ago. What are the chances that you will say “Oh god” and close this dropdown menu? I’d say close to 100%.
Well, I have an ungodly amount of constant backlog. I have never ever been due-review-free the entire time, so this is a small (big ) change that I may consider.
Also, according to which parameters and sorting am I reviewing my cards if I am:
a) reviewing them by their mother deck (in this case VORKLINIK)
b) review them by their subdecks individually
Optimizing the FSRS params of the Motherdeck with its own separate preset leads to different weights, than optimising the separate presets of the subdecks. Wont that lead to a conflict in scheduling
I don’t get what on earth dae was thinking when he added “due date, then deck” and “deck, then due date”. I don’t even get these sort 
1 Like
That is the feeling my proposal here addresses and fixes. Those harder cards have lower stabilities and will get buried faster if you aren’t getting to them fast enough. The ones you do study will show up again later as they come due, but the ones you couldn’t get to will get buried and your decks will stay much easier to maintain.
2 Likes
People probably don’t even look at this. As I said, I worked on this part of documentation and even I am confused what on earth these vicious sort orders are for.
Let them? They can just use the default. Is the goal to make tweaking the settings most accessible? Seems like those interested in tweaking settings are not going to be overwhelmed, and those not won’t do it much no matter how easy you make it.
Hmm, okay this is actually great. And I think if you’re finishing your backlog over several days, this might be a finer method. I still need to organise my thoughts but I’m starting to see it’s merits.
- It’s empowering users to make decisions for themselves by reducing the bad options.
- Making people feel less overwhelmed with the app; we don’t want people to feel there’s an infinite things to learn for this app.
There should be just a sorting option called “backlog mode” where the harder and the easier cards are not necessarily sorted by their difficulty or retrievability but in the manner of a fixed ratio to each other.
For example: 9 or 8 or 7 easy cards to 1 hard card.
@Expertium It is a crazy idea, but it is what I could think of as a potential solution based on that usual solution to backlogs where you create a separate deck for harder or way overdue cards so that you keep on reviewing your easy cards normally while you are churning down the overdue cards and stop the causation of new overdue cards, which is the main point behind this solution.
For example, you made a deck for your overdue cards containing 100 cards, and your deck containing normal cards has 1000 cards. So you are already reviewing in a ratio of 1 to 10 (hard to easy).
Just a food for thought…