Prolonged learning steps

greetings, I want to know, will it be good or effective if I keep my learning steps substantially extended like 30m 2d 3d 4d 5d 8d 10d 15d 20d 25d, as I do not want to indulge in the complex algorithms like fuzz effect, multiplier and all

Controlling the interval manually is inefficient. Human memory is complex and varies from person and material. To find good intervals, it requires to analyze the review logs.

Translated to Ease Factors, the increments would correspond to /, 1.50, 1.33, 1.25, 1.6, 1.25, 1.5, 1.33, 1.25.

For the average person, these learning steps are too inefficient (for comparison: The lowest Ease that Anki ever goes to is 1.3, an area which has been dubbed “Ease Hell”). In addition to that, they are unnecessarily uneven.

This is because of the forgetting curve and how (spaced) active recall affects it:

Very difficult cards might need short intervals. But your steps are too short as a default for all cards; and if they prove necessary, it is probably the material that is at fault for not being well formulated:

Two different approaches:

1. eshapard

1.1 autoLearningSteps

This add-on will write long and adequately spaced learning steps into your deck options.

Mind you that for newer Anki versions, you need to use woolfog’s updated code.

How to create the add-on (it’s easy).

In if l < 28800, you can replace 28800 with the maximum learning step that you want to permit, in minutes. For example, 28800 minutes is 20 days.

1.2 AvgEase

Works independently from autoLearningSteps. Sets the Starting Ease (deck option) to the average Ease of the deck.

Also, the average Ease value gives you a rough idea of how difficult your deck is.

2. FSRS4Anki algorithm

This is the approach I would stronly prefer. The optimizer step can analyze all of your review history and adapt the algorithm to your past memory performance.

Here, it is recommended just set your learning steps to the default 10 minutes and 1 day learning steps over 1 day are discouraged. FSRS4Anki is also compatible with longer learning steps, but AFAIK the learning step setting has precedence, so that the FSRS4Anki scheduler will only apply after a card has passed the learning phase.

To minimize complexity: For the optimizer step, there is an easier to use implementation of the optimizer (the “official” implementation displays a lot of code, which makes it more difficult to find the relevant results):

1 Like

I want to clarify it: I recommend not setting a step of more than one day. For example, if your current steps are 10m 1h 1d 2d , you had better remove the 1d 2d from the steps.

1 Like

This topic was automatically closed 30 days after the last reply. New replies are no longer allowed.