One of my big tasks for the next year will be to better understand how to write well in the mnemonic medium—and how to help other authors write well. Starting to collect notes here:
Conversation
When explaining SRS’s poor adoption, I often say “it’s hard to write good prompts.” That’s a bit misleading: it’s relatively easy to write prompts which encode simple facts—which is all most people imagine doing. What’s hard is writing prompts which develop rich understanding.
2
2
15
As writes: “Anki skills concretely instantiate your theory of how you understand; developing those skills will help you understand better.” Paraphrasing: to be a virtuoso author in the mnemonic medium is to be a virtuoso in understanding, in theory of knowledge.
1
1
13
I have to remind myself of this regularly because it’s too easy to write tips about how to write SRS prompts for simple facts. The harder—and much more important—challenge is in understanding how to encode complex knowledge in this form.
7
1
26
You might find this paper on "Knowledge Components" useful pact.cs.cmu.edu/pubs/Koedinger
Essentially, the specific learning intervention depends on the type of knowledge. Anki/SRS probably works for everything in some sense, but maybe is not optimal for all types.
1
2
6
Replying to
Thank you, will read! (don’t worry: I’m certainly don’t believe SRS prompts can best encode all knowledge—but I want to better understand their limits)
Replying to
For sure. It's a few years old, so there might be new evidence. And I wasn't worried... I'm actually more interested to know what things SRS is not good. I think it can be used for most things, and the things it's not good for are probably fewer.
1
1
2
It’s a good prompt. One weak domain seems to be in building intuition around complex dynamic activities (“how to surf,” “how to dance,” “how to debug")—compared to e.g. simulation, projects, apprenticeship. Can probably help, ofc, but limited.
3