Why does the perfect note-taking system seem to be such blogging catnip? And the post always basically says "here's my system", never "here's why taking notes is valuable" or "here's something objectively valuable that was enabled by my note-taking system".
All joking aside, append-and-review does seem like a nice pattern for maintaining attention on a big heap of odds and ends, which is probably useful for a researcher like Andrej Karpathy.
Because in a world where everything decays, including most physical objects and your health, your ideas preserved is the only thing that may remain of you in the end.
if that's your goal, I don't think you can count on future generations to try to deconstruct your notes unless you've made some other pretty historically significant contributions. you should be writing essays and academic papers.
In the future it will be easier to mine private notes for novelty.
I've developed a pretty unqiue approach to naturalistic non-arbitrary universally binding morality that has fixed my Ai alignment issues (without being able to retrain their model off their weird utilitarianism), but I'm not highly motivated to share it, it'll get around eventually if humanity doesn't implode.
I find it pretty mysterious, and am starting to think it's distributed bike-shedding. I'd wager most notes, if they are ever taken, are write-only. Seems like a distraction to me.
I have been a huge note-taker for many years, but it's mostly about tracking projects and tasks at work and home that I need to be accountable for. Whereas a lot of the recent trendiness around note-taking seems to be more like, looking for a system that is going to capture every insight you have or interesting tidbit of information you encounter, and this is going to reveal things to you.
But what people seem to find is, if a system requires a lot of work and doesn't show any benefits, they give it up pretty fast. Which is why a super simple system like TFA's is probably the only sustainable thing if you just want to remember "stuff" you hope will be useful later.
The problem is the appification of doing, data/structure should have apptributes, we should live in the structure not the function, I have a personal client that I've solved all my issues with, I've got a fair amount of polish before I can release it, but it's effectively my OS at this point.
The nice side effect is that other than chatting with agents it solves the issue of getting sucked into feeds as everything external is a single feed curated by my cluster of ai agents.
I also use a single text file. I have developed my own notation to give it some structure [1]. I have a parser for the notation that creates tree of the document. Then I write various programs that walk the tree and do cool things. I have been happy(didn't feel like I needed anything else) with my system for some years now.
Now if notebook program is watching the file then it will send the code block to jupyter server and write results to `#out{}` "bag". And file will look like this.
```
[x*x for x in range(10)]
```
#out{
````````````````````````````````````````````
[0, 1, 4, 9, 16, 25, 36, 49, 64, 81]
````````````````````````````````````````````}
After using Apple, and OneNote (which suddenly became unusable thanks to syncing becoming paid without warning) and others, I noticed in some cases the software was wiping the bottom of my large notes (or maybe not syncing them). Confirm indeed that the bottom of your increasingly large note is still there. Also need some method to backup or verify integrity of content added, because Microsoft wiping OneNote sync is one of many ways those services betray your memories and leave you empty handed.
This seems almost uselessly simple to me. The “cognitive overhead” of a list of notes feels trivial considering this is a person who managed to put their words online.
The issue isn’t cognitive overhead, it’s not having rituals to review and refine your thoughts. Everyone has to jot down ideas from time to time, but if you never take time to stop, review, and organize your thoughts then sure it’ll feel like a lot of cognitive overhead.
The person is also quite good at specifically putting their words online in a way that others can benefit from them. (Enough so that it’s a bit of a running joke[1] when he quits his job and has time to write some more words.) That skill is generally difficult to transmit, so if they’re saying something in that direction it could be worth listening.
I do it similarly, but I still start another note every week, as it “feels cleaner”. I find that the global search works pretty well on Apple Notes, and by splitting by week I get a sense of around when I might have taken that note, which is helpful sometimes.
This system seems quite similar to sending messages to oneself on Signal/Telegram/whatever. What I like about using messenger apps is that every note gets a timestamp and that messenger apps are, in my experience, more polished than note-taking apps.
I used to like Reflect Notes which had an "infinite " scrolling daily note. Are there any other similar apps to that? it's kind of nice to have everything laid out on one screen but i need a little more structure than Karpathy's single note which feels more brittle somehow
Next up: everybody claims this the best way to do note taking. Single note on Apple notes is now all the rage and single-noting startups crop up in droves.
Give this man a breather honestly.
Ok but todos have a time cost. It takes time to watch a YouTube video or read a book or make a slide deck. Todos often have deadlines as well. You can't capture all of this in an unstructured text file unless you create your own grammar (like what Andrej showed). Even after that, you need to visualize what's in progress and what's blocked because of some other todo, and before you know it, you have reinvented a shittier version of Linear.
By the way, here's my note-taking system: https://renormalize.substack.com/p/my-markdown-project-manag...
All joking aside, append-and-review does seem like a nice pattern for maintaining attention on a big heap of odds and ends, which is probably useful for a researcher like Andrej Karpathy.
I've developed a pretty unqiue approach to naturalistic non-arbitrary universally binding morality that has fixed my Ai alignment issues (without being able to retrain their model off their weird utilitarianism), but I'm not highly motivated to share it, it'll get around eventually if humanity doesn't implode.
If I happen to indeed forget, I’m one grep away from finding what I wrote about the topic based on some vague keyword.
But what people seem to find is, if a system requires a lot of work and doesn't show any benefits, they give it up pretty fast. Which is why a super simple system like TFA's is probably the only sustainable thing if you just want to remember "stuff" you hope will be useful later.
The nice side effect is that other than chatting with agents it solves the issue of getting sucked into feeds as everything external is a single feed curated by my cluster of ai agents.
Its basically an OS for creators.
Checkout the video: https://youtu.be/CpcsOiETgxA
[1] https://github.com/PratikDeoghare/brashtag
Apologies for low quality of video and code. :)
-----------
Example file:
Now if notebook program is watching the file then it will send the code block to jupyter server and write results to `#out{}` "bag". And file will look like this.(I append, the author's really prepending. Anyway...)
When the note gets too long, I cut and paste it to what I call the big note: a 127000-line, 4.9 MB text file I've been maintaining for 14 years.
Trivially searchable, can get context from neighboring notes (What else was happening around this time?), and easily parsable when necessary.
The issue isn’t cognitive overhead, it’s not having rituals to review and refine your thoughts. Everyone has to jot down ideas from time to time, but if you never take time to stop, review, and organize your thoughts then sure it’ll feel like a lot of cognitive overhead.
He also managed to do quite a lot of other things: https://en.wikipedia.org/wiki/Andrej_Karpathy
[1] https://news.ycombinator.com/item?id=39365638
It is called the append and review note, so I think the blog author engages with your point and agrees with it?
It’s also much more similar to how you would take notes on paper or in a notebook.
I just wish some note taking app would have a setting that allows me to open a note at the bottom instead of the top!
[0] https://heynote.com/
I may keep separate append-and-review topics per major area (work, personal, cooking) but that’s about it.
Usually in form of an outline / list, append in the front, and with deeply nested sub-points, as I “discuss with myself in writing”.
my dude you’re literally on Hacker News right now
1. Editable description 2. Comments