I have recently started going through a language mapping that I’ve left dormant for over two years, and in the process I’ve found the need to deal with the oldest clozes, which should be those most likely in need of refreshing. However, I find that none of the selection algorithms are are of any help here, as they focus on the due date, not the number of days since the last play. Thus, using “Due date, oldest first” gives me clozes in a range between roughly 750 and 1850 days of age. And there’s no telling which cloze is the oldest of them all. Of course, it’s very difficult for a user to evaluate all the algorithms, as the process isn’t repeatable.
In fact, I think that focusing on the number of days from last play could effectively replace the algorithms based on percentages, since with the new review-time flexibility (halving or doubling the time between reviews for all percentages), the percentages are no longer fully indicative of the urgency of review.
Besides that, there are other algorithms that come to mind, such as “Most played”, which would address the ones that have caused you most difficulty over the years.
What do you say, @mike and the community?
3 Likes
I just want to point out that prioritising reviews based on percentage completed is an absolutely essential part of my learning process, and that I don’t use percentage complete as an indicator of urgency of review.
I appreciate that this was just a “shower thought” as an aside to your main point, but I want to quash any possibility of changes being made that might affect the percentage complete algorithm.
1 Like
OK, maybe I was a bit too radical 
. But here’s how I’m thinking.
Given, say, a 4-day period for 25%, 10 days for 50%, 30 days for 75%, and 90 days for 100%, you can in fact have the following in your review queue (all due for review):
- 2 days old at 25% (hard)
- 8 days old at 25% (easy)
- 5 days old at 50% (hard)
- 15 days old at 75% (hard)
- 20 days old at 50% (easy)
- 30 days old at 75% (default)
- 45 days old at 100% (hard)
- 60 days old at 75% (easy)
I would like to sort them according to how old they are; do you still want to sort strictly on percentages?
I should point out that the difference between the algorithms may seem subtle if you have a short review queue that you can handle in one day, but if the review queue is huge it does have an impact, as my proposal would allow you to process the most “urgent” items first (in terms of age).
2 Likes
I apologise, I should have made it clear. I don’t only use percentages in the true review process, but also as a “marker” as part of my method the first few times that I go through a collection. Think of it like “not immediately known”, “hints added” etc. I’ll write a separate post describing my method so as to not polute this thread.
You’re additional description here is very helpful by the way, as it really shows why percentages might not be useful for collections that haven’t been seen for some time.
2 Likes