Why do we take notes?

I frequent Obsidian Forum’s Knowledge Management section as a note geek. I want to read other people’s experiences and participate in discussion on how to write better notes.

Recently, user perrinyong created a thread and in the lead up to the actual message, wrote:

As LLMs made knowledge easier to obtain, I started wondering whether maintaining those notes—and publishing a blog—still had value.

The short answer is: yes.

To understand the longer answer, we need to start by discussing why we take notes. Of course, there are different reasons and a lot of different types of notes. The goal of a grocery list (one type of a note) is very different from a daily journal (another type of note) and that’s different from a knowledge note based on research.

In this blog post, I will be mostly focusing on notes about knowledge and understanding. Maybe there’s a niche space somewhere for people to discuss how to make good grocery lists but I don’t hang out in them. And personal journals and diaries are very personal and I think for that, it’s even clearer “no, don’t use LLMs there” from me.

Let’s take a look at a question in the original message: “whether maintaining those notes still had value [in a world with LLMs]”.

The goal of taking notes is not to create more notes. The goal is to consume and retain new information and gain new understanding through a process of thinking.

A great note combines information from source material(s) with our own thoughts, filtered and processed through both conscious and conscious thinking processes. The cognitive act of transforming source information into our notes does not only help us remember it better but helps us build a larger understanding that crosses across the knowledge stored in multiple notes.

When we write a note and add it to our system and connect with other relevant notes, we learn about how pieces of knowledge we’ve acquired overlap, complement and interact with each other. These connections between ideas help us discover new thoughts and fill in the gaps.

I often do double processing for this reason: I write the initial notes down with pen and paper and then rewrite them to my digital notes. Usually I don’t rewrite them word-for-word. Sometimes rewriting helps me find a better way to communicate the idea and sometimes the original thoughts end up being split between different notes.

If you use LLMs in the input phase, you skip the thinking process at entry. You accumulate notes but you don’t have a lot of idea what’s in them, how they relate to other notes and you don’t actively think about the ideas in them so your neurons don’t fire and accidentally find new connections.

If you use LLMs in the output phase, to generate some insights from your notes, you skip the thinking process again. You only get a result — and I’d argue not often a very good one either. A big part of learning is understanding how and why the outcome is as it is. Similarly, you can use a complex calculator like Wolfram Alpha to solve complex mathematical equations but you likely won’t gain much understanding from the results and thus you’ll always need to rely on that instead of over time accumulating understanding so that you can solve them yourself and understand if the results make sense.

Even if you ask the LLM to explain things, there’s a big difference between reading an explanation and doing the thinking and reaching the outcome through understanding. The second one is of course way slower, especially in the beginning, but it builds up knowledge and understanding. The more you know about a topic, the better thoughts you can conjure about it.

When I speak and write to software developers about writing a blog, I always talk about how documenting things for yourself is valuable because you end up with a specific guide for how to solve something that is written specifically to match your understanding. The same goes here. Even if you’d write about something that’s easily accessible knowledge (from documentation or encyclopedia), writing it in your own words is massively helpful.

It also exposes you to the gaps in your knowledge. Our brain is good with fuzzy things. It’s easy to think we understood something until we put it to a test. The first and easiest test is to write it in your own words. Your brain is excellent at fooling you while thinking but putting words on the paper makes them so tangible it’s hard to skip steps without noticing.

Writing a note thus becomes a litmus test for your understanding of a topic.