A post on “Second Brains” and the PKM software used to make them.
---
If you’re not familiar with the concept of Second Brains, in theory– it’s basically: offload your notes digitally so you can free cognitive space to do the “heavy lifting” such as problem solving or creativity.
This is how the theory goes anyway, in reality, people spend tons of time finicking tools and getting their setup “just right” or plugins for every tiny little functionality that crosses their mind. But all these options included in PKM tools like Notion, Obsidian, Logseq can quickly become a monument to the inner workings of your own chaotic mind rather than assisting in the creation of anything valuable. The system itself won’t make you more creative or productive. The real work is in the thinking, synthesizing, and creating. If you don’t have a clear purpose for the information you are collecting, the system will just become a “digital graveyard” of unused notes.
Let’s say Anne (a college student), creates folders for all her classes , organizes lecture by lecture, pulls out concepts that tie together professors comments etc etc and backlinks them? Does she have a second brain? NO. Okay, now Anne is pissed at me , she incorporates Tiago Forte’s CODE & PARA frameworks into her directory of folders (she can see all her active projects in one spot, nice workflows, maybe even has a little table tracking her work log / study hours, plugins for her research papers like Zotero etc), does she have a second brain now?
still NO.
Now she’s just ignoring me. But let me explain: a second brain is not just a fancy folder of notes, it is a cognitive clone that genuinely offloads work for you. This has not been truly possible in the past which is why “second brain” was always a bit of a misnomer. No actual cognitive work has been offloaded to the software - all you did is create more work that’s better organized for later and save you some time on work down the line. You didn’t offload that work down the line , at most you made it 5-10% faster to achieve.
What Anne has really been doing is building a set of training data that an actual second brain tool like an LLM can actually use allowing her to take actual leaps and bounds in time saved.
Here are some guardrails/principles that keep me personally on track:
- 1.Basic PlainText/Markdown/Org ONLY: I saw that Obsidian recently introduced something called Bases that introduces a database view. I understand no two people have the same workflow and there are people out there that love this sorta feature. But you literally introduce something called a .base file to make it work, what happened to simple plain text? All your core notes are still markdown, but is this extra functionality really creating modularity? Or creating a dependence on the Obsidian Experience? So do your future self a favor , keep your notes in a simple format with no bells and whistles, when you have to open a file from a cloud backup , you don’t want it to get janky just cause all your images are linked elsewhere or backlinks are not usable on your text editor etc. This is also better for LLMs because they also don’t have to parse through files it may not understand. ALL of them understand markdown/plain text.
- 2.The search bar is your friend: I’ve seen first hand , the more buckets (folders) you introduce, the more unwieldy the whole endeavor becomes, I keep 1 degree of folders MAX , I do not nest a topic within topics anymore , each topic has its own folder. EVEN if topics relate , I don’t “connect” them , I search for them. More nested folders creates more layers of organization , all of a sudden it becomes a mini project on its own to see if the current schema is still working for you. Connecting is objectively a waste of time , you should connect in your study notes on paper via a mindmap but in your digital ecosystem? Why when you can search in the blink of an eye!
- 3.Prune what isn’t relevant every quarter: I keep a cloud (or a github) archive when I’m done with using a topic , it prevents clutter and since everything is contained in its each folder , its easy to import and export what I need specifically then the whole archive as well.
- 4.Keep your REAL work out of it: This seems a little counter-intuitive because typically people use these tools as their work desk but like YOU using a to get things done , I think this creates an unnecessary association between your second brain and work. Your second brain should be a freethinking space with no expectation of performance , simply learning , synthesizing ideas etc. Worldbuild your novel in your second brain sure, but do the actual writing in a word doc or google doc.
What about the LLM element of all this? The actual cognitive offload?
I know I know it goes against the #1 warning that AI companies themselves put in their search bars - not everything may be accurate/watch out for hallucinations. But quite frankly I’ve been doing some of this for a while now , as long as you put some guardrails, I’ve never read anything after honing in my prompts that I have felt or known to be extremely wrong. This is because I expressly tell it to pull its data from just the file I attached and that seems to hone it in fairly well.
Here’s some examples of what it can do for you in a way that acts like a true second brain
- -Create flashcards ? Nah
- -Create Quizzes?
- -Ask you to explain a concept / tutor you based on your notes?
Everyone knows this already. How about:
- -Automate your journal reviews: If you keep all your journal entries digitally , use a few command line steps to concatenate everything into 1 master file and whatever your weekly review or annual review questions , ask AI to answer them based on your journal entries and generally speaking, it’ll give you some insights you didn’t even consider, go over everything with the framework or model you were already going to use, and in record time. That is literal cognitive work saved and a game plan for the next X timeframe.
- -Summarize your youtube watch laters / playlists especially for videos where the content is more important than entertainment value , this is major time saved. It’s like a second you on the couch watching these videos so you don’t have too.
- -Parse through all the notes from what it learned in all those videos and you can start asking it to 80/20 the highest priority action items (if that’s your objective) based on what it knows about you or some criteria. That’s real cognitive work , completely off your plate and can feed your real brain insights it wouldn’t have otherwise had. Most importantly - without impeding your learning because you are not asking it to come up with answers , you’re simply asking it do the tedium that had no cognitive benefit in the first place. You don’t actually need youtube videos or listening to podcasts from pop media, you need the information these things unlock so why manually do that? Let your second brain “do” it for you.
*the caveat: Knowledge that is truly important and valuable such as lectures, should still be done manually because you don’t want to impair your ability to learn, this is what your real brain should be for. Let your second brain consume all your junk that entices you that really only has one nugget of valuable info. You are healthy and strong with things to do, let your second brain notes + LLM be the binge watching / reading slob on your behalf. Use your real brain for your best work.
Capiche?
Any questions , mistakes, or things you wanna argue about? Feel free to email me hi@muj.dev