When talking about AI/human augmentation, we often gloss over the greatest/most used augmentation that everyone uses—notes. Notes are crazy because they function as an external memory source and the open design space for them is basically infinite. This probably requires a deeper dive into 1) What is the fundamental purpose of notes and 2) How can we optimize them to better suit our needs in the current day.

“I think a writer’s notebook is the best way to immortalize bad ideas. A good idea sticks around and sticks around”
- Stephen King (i think)

I really like this quote because it makes us question fundamentally what notes are. If good ideas are the ones that stick around, why do we need notes? It brings to light the main reason for notes—both callback or inspiration. Any truly important ideas will probably already have planted itself deep inside your head.

problems with notes

• You can either have good recall with analog notes (you know exactly where/which page its stored in) or a large volume with digital notes
• People usually don’t remember 90% of what they read—this includes lectures etc.

Lets outline what an ideal note taking system would look like:

• Perfect recall of past ideas—if you have a vague concept of an idea, it can pull it up automatically
• Allows for creativity + knowledge discovery—you can look through past notes for inspiration
• Can store functionally infinite data BUT still not violate the first principle of perfect recall (ie. we aren’t there yet in the case of digital notes)

examples of “good” note taking systems

Here I’ll talk about a couple different systems that I found interesting through my time researching. IMO they have very good thesis’ for how to resurface notes (learning/reading/attention hijacking etc.).

This is strictly different from taking ephemeral notes to remember more—usually these notes are good to communicate (hand off post-it notes etc) but they don’t exist to store ideas.

zettelkasten

This system gained traction because of Niklas Luhmann, who was a famous psychologist. Essentially the main way to store ideas were to have small “zettels” which were big titles + small bulleted notes on index cards. There were two types of notes—literature and fleeting—which were to be revisited in X amount of time. Then they would be transformed into permanent index cards that could be filed away.

The reasoning behind it was the people forget/don’t synthesize ideas—which is a huge problem in current society. Unless you have a test/assignment, its especially hard to have an event for crystallization of ideas.

You could visualize the note flow as:

pocket notebook

The main thesis here is that phones are distracting and at odds with taking notes. A pocket notebook on the other hand analog. This allows for expressive notes + being able to read the notes later for fun… rather than social media.

They can also be used in any context—from meetings to sudden bursts of inspiration while walking. This analog approach forces more deliberate and thoughtful note-taking compared to quick digital captures. However, the main tradeoff is that these notes aren't searchable or easily shareable like digital formats.

Here the note flow is more linear, but has much less friction than the zettles:

programmable attention (anki)!!!

Essentially this is another really interesting note taking system, where you can use spaced repetition to functionally program your memory—tiny tasks leading up to larger goals. This hijacks your attention mechanism to learn more and have better concepts.

The downsides are pretty bad though—its quite hard to find structured anki notes for any specific subject, it optimizes for memorization NOT understanding (ie. you know every word in the language but not grammar), and in general there is a decent bit of friction when getting started.

The note flow here is strongly linear, with no loopback (it doesn’t inspire new notes etc.):

comparison

Each system has unique strengths that shine in different contexts. The zettelkasten system excels in organization and idea synthesis, making it perfect for academics and writers who need to connect complex ideas. However, its lack of searchability and scalability means it can become unwieldy as your note collection grows.

The pocket notebook approach offers incredibly low friction and excellent recall due to the physical act of writing. You can pull it out anywhere, anytime, making it ideal for capturing fleeting thoughts. But like zettelkasten, it struggles with searchability and scaling beyond a certain volume of notes.

Digital spaced repetition systems like Anki offer the best of both worlds in terms of recall and searchability. The programmatic approach to memory makes it extremely effective for retaining information. However, it falls short when dealing with transitory thoughts or contextual understanding—it's great for remembering facts but not as good for capturing the nuanced context in which those facts exist.

You could visualize the tradeoffs like this:

ummmm… so ai

I know how overhyped AI is, but we have to consider it here. AI has fundamentally altered how we approach data in a couple ways. The main way is that we can go from unstructured ⇒ structured data extremely easily. The treatment of text as data that we can semantically understand is relatively new. We can now understand content, without the need for titles/tags/shorthands. The data can actually be read & recommended to people based off of CONTENT.

The way most algorithms work, are that they’re content blind. They look at people you interact with + similar creators that people with similar interests to you have, and recommend that content. The twitter algorithm (at least some time ago) didn’t know the content of a tweet—it only knew the reactions/interactions with them. Fundamentally this meant that without a large social group, recommendation algorithms were useless.

my rendition of notes

What if we have more unstructured notes? Think audio, video, short jotted down ideas, that could automatically aggregate into long term structured notes. Although this is already done by Mem/other AI note taking services, it has one key disadvantage. Memory retention. I won’t intimately know the structure of my notes if I don’t create it.

The solution? A recommendation feed. First you start with unstructured short term notes—these are automatically clustered and aggregated into substructures. Once enough time has past/the notes become more mature, they transform into long term notes.

Now the feed. Imagine a UI, similar to twitter, conditioned on your short term notes that recommends you snippets of long term notes. Systematically, you would get paper ideas, topics from classes, interesting conversations, that are slightly tangential to the short term notes that were ingested. You could have a liking system, that takes into account which ones you find interesting.

This way you can hijack attention, but positively, engaging yourself in past conversations + media that you were interested in. This could inspire further research ⇒ which gets auto ingested into your note system ⇒ and might get recommended to you later.

I argue that this 1) seems like where note taking is going and 2) utilizes AI, not for generation, but for recommendation and understanding. This way AI can keep track of all the busywork, while laying out the creative process to humans at the most important places—having conversations, reading papers, skimming data.

You might even have an external recommender that crawls the internet for you—recommending papers etc. that match your current vibe.

The main downside here I can think of, is the lack of analog integration. Sharing ideas through handing pieces of paper is pretty efficient + ripping paper off a notebook and putting it onto a board. Ideally I think a majority of the data coming through the ingestion engine would be images/video/audio and very little of the text would be typed in yourself. Imagine a world where you write out your ideas on a notebook ⇒ quickly take a snapshot ⇒ automatically gets sent to the ingestion engine.

My ideal workflow would be:

I have an interesting idea ⇒ I jot it down on a tiny pocket notebook/any local piece of paper ⇒ take a quick picture ⇒ I forget about it/go to my recommendation page and read snippets of interesting blogs related to what I just jotted down (further cementing it in my mind)

Another really interesting one would be:
I read a paper and annotate it up a little bit ⇒ go onto my recommendation page where I see other papers I’ve read that are slightly similar + conversations I’ve had with people on cognition that are similar ⇒ I start drafting out a paper idea or continue down the rabbit hole with external recommendations

Overall, here are my takes on notes + how I think it could be changed. This is something I love to talk about in any conversation. Feel free to always text me cool organizational systems + other ways people create ideas.