Stop Taking Smart Notes (And Do Something Real)
Stop worrying about setting up a PKM system the right way. Stop worrying about atomic notes and "discovering connections". Stop worrying about the software you're using for it. Just do the thing that this PKM system is supposed to bring into existence. That's it. This is my message. You don't even have to read any further unless you're really curious how I am going to defend the thesis that seemingly goes against 95% per cent of the "productivity YouTube" and what "PKM bros" have been telling us.
But if you do decide to stay, let me explain.
PKM systems serve two purposes that often get conflated. First is storage and (efficient) retrieval of what has been stored. Second, which is often severely downplayed, in my opinion, is learning. These two are the different ends of a spectrum. PKM systems the way they have mostly been touted so far are good for neither of the two. Because they sit closer to the middle of said spectrum. They are suboptimal for retrieval and, for the most part, not very effective for learning. They are neither here nor there. This is Thesis 1A.
But no matter where your actual setup lands on that spectrum, everything must be guided by relevance: storage and retrieval and even learning must serve a specific ongoing project or one very likely in the nearest future. This is Thesis 1B.
Let's unpack that.
Information can be used as some kind of raw input that someone (or increasingly something) needs to process before it gets implemented in a "product". The processing of this raw input can also be the means of learning.
If this processing is seen as means of learning, we can arguably claim it makes some sense. We can probably make the same claim if the information is processed to become a more digestible input for the PKM system. In other words, between the raw input and processing for implementing in a product, there is some processing for the purpose of storage. This is where Tiago Forte's progressive summarization sits (Forte, 2022). Heck, this is where most notes and note-taking practices sit. To claim there is a lot of value in this kind of processing is going to be somewhat problematic in the modern LLM-mediated life.
Let's start with the first case, when the PKM system stores the most "raw" input. This is what normally "productivity" bros do not advise that you do. Because this is not how you're supposed to make "smart notes". Here's a simple example: you find a Medium blog post that you like, so you "clip" it into your PKM vault. So far so good.
I am now going to say something very controversial: this type of input in your PKM is one of the least problematic. It's totally benign. You want to clip an article and store it in your vault? Go ahead. That's totally fine. You're not taking smart notes, but it doesn't matter. I will explain why in a moment.
Such units of input are probably short and reducible to plain text and thus easily and cheaply storable, retrievable, and manipulatable in digital systems. And if it's not seen by your brain as an open loop, I don't see compelling reasons not to store this kind of input. Having said that, if you're collecting stuff indiscriminately, this very specific piece of input becoming suddenly relevant is a total fluke. Now, honestly: How many of these clipped "articles" have you used for something real and tangible lately (or ever)? And if you hadn't stored the input that it serendipitously turns out you now need, how long would it actually take you to find it (or something similar) in a pinch?
By the way, is it me, or do all these "productivity PKM gurus", always assume these are the type of inputs most of us work with? Because for me, the reality is different as my normal inputs are something like Halliday's Introduction to Functional Grammar. All the 808 pages of it in all their glory. And this is where we get to the most problematic side of PKM systems.
We cannot just chuck the whole book into a PKM system. Or rather, technically we can, but this would go against what we have been taught about these systems. So, what do we do instead? Of course, we make notes on certain topics and concepts from the book. This processed output becomes the input that actually goes into the PKM vault. These are "notes" par excellence (a.k.a "OG notes"): "lecture notes" or "book notes" or "article summaries". Stuff like that. Is this what you PKM looks like now? If so, how is it working out for you so far?
These kinds of notes suck balls if they are later used as input for a specific project because no matter how hard you copy (short of just copying and pasting the whole damn thing into your vault), they are are not going to be detailed enough. There's always that detail or idea that you have to run back to the original source for.
But what if I just need this very specific idea or quote from a book, and unless I save it now, I am going to loose it. Ok, fair enough. But then it's back to Thesis 1B: it probably should be more or less immediately relevant; otherwise you can keep saving all of these snippets forever as your vault grows without any particular end (pun intended) in sight.
Then, "dumb notes" also suck as a means of learning because processing taking place in making such notes is pretty shallow and not cognitively challenging enough. But what is? "Atomic notes", of course. I bet you saw that coming.
The idea itself is great in many ways. Especially on the surface. If I wanted to make productivity content on YouTube to potentially see good engagement metrics, this is what I would promote. It looks very sexy. It looks very smart. Something that would set you apart from all these clueless normies still making dumb notes.
But on a more serious note (no pun), let's see whether these atomic notes live up to the hype.
Right off the bat, I am going to wholeheartedly acknowledge the importance of rewriting concepts in our own words. As Ahrens (2022) argues, writing is not just what happens after you learn something; rather, writing is "the medium of all this work" ("Introduction" section). He emphasizes that if you want to truly understand a concept, you must translate it into your own words, because "thinking takes place as much on paper as in your own head" (Ahrens, 2022, "Everything You Need to Do" section). Writing the note is the learning process. I have no qualms with this because I simply can't. I have been doing this for some time now and will continue doing that for the concepts that I am actively trying to learn.
What I am going to say, though, is that for real learning to take place (if this is the goal), a note has to be rewritten and/or actively recalled a few times. At least, in my experience. Do people following this approach actually do that? Maybe yes, maybe no. Also, this is no longer about a PKM system per se. In terms of a PKM system, this atomic note is actually an input now in a storage and retrieval system. Will these atomic notes make good "pre-processed inputs" for "secondary processing" if you happen to need to actually implement them in a project? I would say no, they will not. Same reason as with "dumb notes" above. If anything they will contain even less detail from the original source, which you will still have to run back to, when push comes to shove.
Totally food for though: Why bother organizing these kinds of notes into systems if the goal of producing the note was producing the note? Ok, as stated before, with plain text files, storage is cheap, and you may actually find use for the note later. But, again, in my experience, you don't use the note per se, you use the understanding that facilitated by the process of producing the note. I hope that makes sense. The existence of the note it the vault will not hurt, but it's not sacred either.
But then there's the real issue that I personally have faced and am facing. If you're producing atomic notes based on two-page Medium posts or even books like Ahrens' How to Take Smart Notes (every opportunity to go "meta" on something is never lost on me), it's relatively cheap and feasible. On the other hand, the amount of processing required for complex inputs like my Functional Grammar example above is immense. With inputs like this, even dumb notes discussed above would take weeks (I know that because I have done that). Atomic notes would take forever. I am not saying you don't need to make them. If nothing else, they are are good for leaning. That we have already agreed on. What I am saying is that Thesis 1B must be applied here. Ruthlessly. Unless you have very compelling reasons to produce these notes because they have immediate relevance and you're actually cognizant of the fact that you should be doing this mostly for learning rather than storage and retrieval, you may potentially be wasting a lot of time. Very likely for nothing. And this comes from someone who speaks from experience.
Incidentally, this is something that Tiago Forte also emphasizes:
The best way to organize your notes is to organize for action, according to the active projects you are working on right now. Consider new information in terms of its utility, asking, “How is this going to help me move forward one of my current projects?” (Forte, 2022, Chapter 3, "Organize: Save for Actionability" section)
The problem is that even though he repeats this message multiple times throughout the book, it somehow managed to be lost on me, as what I took from the book when reading it a few years ago was the PARA system with it's respective categories and maybe a bit of progressive summarization.
So what is the takeaway here? What do I suggest we do? What do I actually do, now that I am on the path of healing? Here's the shortest version possible.
Don't overthink notes and note-taking. The notes are not for note-taking sake. They must serve a purpose. If you need notes to make a project possible, just do what is required to move the project, not your note-taking system. Copy and paste stuff into a loosely organized structure of folders lest it turns into a complete mess. But even if it does, f&ck it. Use search to find what you need. Or delegate it to Claude. Rewrite some parts of your raw input if you must and leave others alone. Dump the raw inputs into NotebookLM and make it do all the heavy lifting for you, when you need something specific from these sources. Don't make your notes pretty. Rewrite stuff to learn, but don't overthink it's place in the PKM system. Basically, learn what's actually relevant and produce stuff that matters.
References
Ahrens, S. (2022). How to take smart notes: One simple technique to boost writing, learning and thinking (2nd ed.). Hamburg: Sönke Ahrens
Forte, T. (2022). Building a second brain: A proven method to organize your digital life and unlock your creative potential. Profile Books.
Originally published: 2026-06-18