Cron for the Mind

On Spaced Everything

Image created by ChatGPT. Prompt: “cron for the mind.”

Intro

---

3 notes before you dive in:

1. None of this post will make sense if you aren’t comfortable with spaced repetition systems or familiar with the idea that they can be applied outside the typical approach of pure memory. See the footnote if that doesn’t apply to you.¹

2. A lot here will be derivative of prior work by Andy Matuschak and Michael Nielsen in this area for a bunch of reasons.

   1. I’ve spent so much time reading their work that they heavily shape my perspective on spaced repetition, memory, education, and learning.

   2. I’ve found that a lot of memory system users who start to experiment with non-memory uses end up trying out similar kinds of things. As I describe multiple times throughout this post and the next, I often feel like I’ve made a cool and semi-unique discovery and find it later in one of Andy Matuschak’s notes or Michael Nielsen’s posts.

Spaced everything

---

10 months ago, I read Michael Nielsen’s wonderful essay on spaced repetition memory systems and I was instantly pilled. Luckily, Michael Nielsen’s essay and the premise of being able to “make memory a choice” were fascinating enough that I quickly managed to get through the numerous barriers to adoption, and to a place where I became genuinely confident that I had control over what I remembered.²

If that was it, and Anki (my preferred memory system) could “just” improve memory of individual facts, the platform would still have been incredibly meaningful.³ When you use memory systems well, they can go even beyond that and serve as a medium for deepening understanding and intuition.⁴

As I continued to use these systems, I kept expanding my range of what I thought they were useful for. I needed to start remembering to fold the laundry after coming home from school and decided on a whim to use Anki to remind me. Soon after, I quickly started using Anki for habits and repeated tasks. Days later, I realized that Pocket just wasn’t cutting it for me anymore as I read-it-later list. I would chuck everything in there but I had no guarantee that anything useful would resurface. On another whim, I started chucking things to read into Anki, and, lo and behold, it worked better than any read-it-later system I had used in the past.

What I was slowly discovering, without explicitly realizing it, is that Anki is effectively a general interface for task scheduling. The idea of spaced repetition helps for memory, but can also be extended to many other uses because it’s effortless. Andy Matuschak calls this spaced everything (where the title comes from).

Systematically, we can generalize spaced repetition to:

• a priority queue of microtasks

  • (for memory tasks, in SM-2: a simple due date)

• an interactive environment which presents sufficiently high-priority tasks

  • (for memory tasks, a flashcard UI which shows “due” cards)

• feedback actions which modify a task’s subsequent priority

  • (for memory tasks, forgotten / remember modify the next due interval)

(…)

But the core concept—automatically arranging and presenting tasks according to some expanding schedule—can be instantiated in many interfaces and domains. I call this notion Spaced everything.

(…)

It’s interesting to imagine a single interface malleable enough that I could define my piano exercises above as one sort of routine, and a SRS memory system as another routine—both special cases of a single general primitive.

In the future, I want to build prototypes of genuine spaced everything systems themselves or see them built well by someone else. I like to think of this future prototype as the “Notion of spaced repetition systems.” Notion’s a powerful system based on general primitives where ease-of-use for beginners usually isn’t a bottleneck for sophisticated users.⁵

Beyond that, I would be thrilled if the concept of spaced repetition permeated more throughout “our culture” and started seeping into all sorts of different systems. As a mechanism for practice, it would be cool to see more habit-tracking apps, sports practice apps, and music practice apps incorporate spaced repetition as a core concept. Outside of practice, I would love to see inboxes (email inboxes, reading lists, social networks/messaging) use spaced repetition more as well.⁶

Common uses I expect for spaced everything

---

I’ve been using Anki as a spaced everything approximator (you can get surprisingly far once make the mental switch away from viewing it purely as a tool for memory) and have often tried to convince my friends to join me in the spaced everything rabbit hole as well. I’ve found that there are a couple of areas where spaced everything has worked well for me and that seem to be the most exciting to my friends.

Tasks w/ fuzzy deadlines

• Examples:

  • “Dedicate a week without much school homework to exploring cybersecurity.”

  • “Shop for new electric guitars.”

  • “On a free day, spend some time thinking about the things you want to have learnt by the end of high school and strategize your future classes and self-learning accordingly.”

Fuzzy habits/making sure I do the right thing in a given situation (most of the time)

The strategies from Tiny Habits and Atomic Habits are great for reliable ways to stick to important habits, but I often have tons of fuzzier and situation-dependent habits where those strategies honestly feel like too much work. Often, for these habits, I’m fine with a tradeoff where I only do the right thing 50% of the time in exchange for less hassle.

It’s incredibly annoying to have daily/weekly/whatever reminders for these kinds of habits (and I often wouldn’t even know when to set a reminder for), which is why I love using spaced repetition for fuzzy habits. The natural spacing works quite well out-of-the-box. At the beginning, short intervals between repetitions help me get started with the habit, and as the card spaces out, it can serve as a check-in to make sure I haven’t forgotten.

• Examples:

  • Q: What should you do every day after washing the dishes?

    • A: A minimum of 30 minutes of guitar practice.

  • Q: What will you aim to limit your daily added sugar intake to?

    • A: 21 grams

  • Q: How will you limit your daily added sugar intake to 21 grams a day?

    • A: Keep a fuzzy running counter of daily sugar intake and do your best to below the 21 gram line.

  • Q: When you go to the grocery store, what’s a good way to avoid tempting yourself and bringing home unhealthy food?

    • A: Try not to get too far inside. Stay on the outside aisles.

  • Q: The next time you write a blog post, what should you do before you publish it?

    • A: Get feedback from at least 2 people.

Alexey Guzey and Andy Matuschak have written about this general idea as well.⁷

Ideas and blog posts

• Ideas: open-ended questions where the goal is to come up with some new answer each time and to jot them down as answers on the back of the card.

  • What are companies or industries that are dying or deserve to?

  • How can we get more infectiously curious, intelligent, and agentic people to want to be elementary and middle school teachers?

  • What does a memory system look like for a driven creative genius who is an expert with the system? (from here)

• Blog posts: most of the blog posts I’ve written this year (other than this one) have come out of incremental writing. Over weeks or months, I’ll slowly expand ideas to include more references, ideas for sections, chunks of text I want to include, feedback from friends, etc. Once I decide that I like an idea enough, I’ll pull it out and hammer away at it until it’s done.

Reading inbox

I genuinely thought that this was a completely novel idea, and got incredibly excited when I first showed a friend how much better it was to use spaced repetition as a reading inbox than an ever-lengthening list. Within a few weeks I found out that Andy Matuschak had given serious thought to using spaced repetition as a reading inbox at least 5 years before me (because of course he had).

He has a more articulate explanation of this idea here, but here’s the short of it: it’s stressful to keep ever-expanding read-it-later lists and it’s good to periodically reread old media (I quite like Simon Hørup Eskildsen’s playlist phrasing). Neither happens if you either read new content immediately and forget it, or save it in an ever-expanding list of things to read where important things quickly get buried by less important ones.

I like the “desk with papers” metaphor that Andy Matuschak talks about in his note:

It helps me to bring some physicality into the metaphor. Imagine your desk has lots of papers on it. You naturally pull out the few which you’re using at the moment, and maybe you set a few aside for special attention. The rest sit, perhaps in a couple of piles. You intermittently look through the piles, pulling out ones which strike your fancy. The ones which sit long enough start to simply dissolve, compost into the table surface. You can always restore them if you like, but if you don’t, they’ll get tilled under.

Beyond this metaphor, I also like to think of tasks and items on my reading playlist as cards in my orbit (amusingly also inspired by Andy Matuschak; this time it’s his Orbit system). The ones that are most important/urgent will stay in a tight loop around me while the rest slowly drift away.

Principles of spaced everything software

---

As it is, I haven’t seen any popular spaced repetition systems that work well for spaced everything. It takes working against the way they are designed to make a lot of these ideas work.

I want to close part 1 with a brainstorm of some of the core principles that spaced everything software should have. I don’t think that this is quite general enough to fit Andy Matuschak’s vision, but it would probably be enough for 60+% of common uses.

• More and easier customizability while creating cards. This is key.

  • Ability to add the first day you want a card to appear.

  • Easier and more advanced flagging

  • Easy and intuitive ways to change the spacing of card repetitions and the way “self-grading” works. The standard “learn” cards should default to some SM-2-based algorithm, but many other uses would benefit from weekly repeats or some sort of doubling. Many won’t need any self-grading (i.e. you might want fuzzy tasks to just appear weekly until you delete the card).

• You should be able to easily create card “templates” for common use cases.

  • There are pre-made defaults that you can edit for standard use cases like tasks, habits, reading inboxes, and of course memory.

• A “today” interface where you can review all the cards for the “day” regardless of what deck they’re from. As you review, you can opt to “skip” and leave a card in the today section that you can check up on afterward without changing its spacing. A card that was left in the today section at the end of the day will just carry over to the next day. The idea is that if you see a read-it-later card or a task card that you want to follow up on, you can let it sit in the today section as you continue moving through the rest of your cards for the day.

Acknowledgements

---

I’ve hopefully already made it incredibly obvious how much Michael Nielsen and Andy Matuschak shape every angle of the way I think about memory systems. Thanks also to Jacob Goldman-Wetzler for introducing me to both Michael Nielsen and Andy Matuschak and for pilling me on memory systems. To the several other anonymous friends who let me bug them with ideas for empirical experiments on weird uses of memory systems—you guys are real ones.

1. For an introduction to spaced repetition:

• Must-read if you’re completely new: https://augmentingcognition.com/ltm.html:

  • To put this essay in more context, I would highly recommended that you at least skim through these: https://notes.andymatuschak.org/z8v56RCUFx6Zp6sBG6mTL95 and https://michaelnotebook.com/ongoing/hiums.html

• https://ncase.me/remember/: A very fun intro to spaced repetition. It could be helpful to go through this and get a general idea so that you can skim through and read the most interesting sections of the links above.

• https://gwern.net/spaced-repetition: If you’re a Gwern fan, this is another great introduction with a rabbit hole of links, research, and useful tips.

2. If effective card-making is a barrier for you, see some of the following. A couple of hours spent upfront will have a large ROI.

• https://andymatuschak.org/prompts: An interactive introduction to card-making with a light subject.

• https://quantum.country/: A great example for getting more intuition on how to make good cards with a more “academic” subject.

• https://notes.andymatuschak.org/z9xavmmNq7xvNqzpnJ3HFXx: Another list of tips for card-making.

• https://super-memory.com/articles/20rules.htm: A list of tips for card-making.

3. I remember writing about using Anki to learn some econ concepts in my journal a couple of weeks after reading Michael Nielsen’s essay. As I was writing, it hit me that if I used Anki correctly, I would basically never have to experience the feeling of “oh wow that’s cool/important, hopefully I remember it” ever again. “Make memory a choice” felt like a nice-sounding but near-hyperbolic slogan at first, but when you truly believe that your memory is a choice, it can genuinely change your relationship with learning and curiosity. I’ve always had mild intellectual curiosity, but I also always felt like I was wasting time whenever I followed my intellectual excitement because I would forget most of what I’d learned. At the risk of sounding cringy, I had an emotional reaction to this realization as visceral as the ones I have to beautiful cinema or beautiful music. Since then, Anki has been the vehicle of the growth in my intellectual curiosity and excitement.*

*Though not just because of memory. I’ll talk later about Anki’s power as a reading inbox, and that’s probably had just as big of a contribution.

4. This is something that Michael Nielsen emphasizes in “Augmenting Long-Term Memory”, but you can see more examples below.

• https://cognitivemedium.com/srs-mathematics

• https://notes.andymatuschak.org/z4KxfCZPkVEf2R8nayLJZBG

• https://jacobgw.com/blog/tft/2024/05/12/srs-intuit.html

5. There’s a quote by Michael Nielsen in “Building a Better Memory System” on this that I love.

Software design so often focus on the first few hours of someone's experience. Yet what you really want is to max out the experience someone is having in their thousandth or ten thousandth hour of use. Pianos seem designed primarily for experts and only incidentally for beginners. If you were designing the piano with modern software design practice in mind it would have 8 white keys, no black keys, and no pedals. It'd be easy to play some simple songs, and that's it.

6. There’s an interesting tension here. Of course, you want a malleable general interface to really unlock the abilities of power users (see footnote above). But I think that the “seeping into other interfaces” aspect is just as important. Spaced repetition memory systems themselves have serious barriers to adoption, and I think it would be even harder to see the value for these more general systems unless you already have some experience with memory systems. For many people, they’ll just be blank vessels that’ll take significant amounts of time to fill up and days or weeks before demonstrating any noticeable gains.

7. I had enough to write about here that I ended up fleshing this footnote out into a separate post.