Log: personal mnemonic medium

One big tension when writing in this medium is that it feels like I’m repeating myself. I read the book, I write some prose notes, I wrote some prompts about the notes. It feels like drudgery. One helpful approach has been to write cloze deletions like these in Human perception:

And this one from Human visual system:

These type of cloze sentences blur the lines between prose and prompts. If I write carefully, I can write text which does “both jobs.” It’s interesting to contrast this with the more structured, consistent model proposed in RemNote:

Two common patterns I notice:

  • “Two-sided prompts” like the first and fourth example above. Note how the fourth example gives extra context for the first deletion.
  • “Finishing the sentence” as in the third prompt (and sort of in the second, with an extra deletion capturing a bonus bit of detail).

Some related observations:


David Dohan suggests that cloze syntax could optionally include scope—i.e. by ending with |p} for paragraph or |s} for sentence or whatever. I need to think a bit more about that, but it seems interesting!


There’s clearly some tension between creating legible prose notes and creating good SRS prompts. Adding lots of Q/A pairs creates noise in notes, particularly since they’re often duplicative of existing note content. I’ve been leaning much more on cloze prompts to mitigate that issue, but if I’m not careful, they create cards with tons of text.

It’s helpful to internalize that the cloze prompts include the entire surrounding paragraph. Once you understand that, you can place your paragraph breaks to create the cloze prompts you want. But that makes its own tension, since it pushes you to move paragraph breaks to places you wouldn’t naturally put them in prose. I may want to create additional syntax for specifying “capture only this sentence” or “break here for cloze purposes” or whatever.


Used the system for the first time in a serious way, while studying Service workers. I felt supported, something like when doing a pull-up with a positive upward force. Normally, when I take notes on mundane stuff I’m learning like that, it feels bad because I know they’re not really going to help me internalize the material very well. Using Anki feels bad because it feels like I’m scattering what I’m learning to the winds. Writing notes inline, particularly as Evergreen notes, produced a great feeling of confidence.

I also really enjoyed being able to stop partway through to review the prompts I’d created thus far.