Need help figuring out how - and how much - to atomize

I started in Obsidian already with a large body of existing data. And I’m loving being able to interlink it (although mostly I just have things sorted into MOCs, so far, so I can find what I want to work with more easily since there are so many notes already), but I’m struggling to figure out how closely my current note structure does or doesn’t resemble Smart Notes, and how best to adjust from where I am now.

Although there are exceptions, the overwhelming majority of my notes are in two categories:

  1. Book notes
  2. Stuff I wrote on Facebook and thought was well-written enough to be worth saving

Some of the FB posts have a lot of duplication because I might have written about the same basic thing 3-8 times, in different places, so they may be merge-able. I think most of these are probably closest to permanent notes; they’re usually fairly developed explanations of various ideas, presented to others.

My book notes are mostly in a kind of outline format. I don’t do a lot of writing in my own words (given that I don’t have trouble “translating” what I saw in a book to the new ideas I want to connect it with, it kinda feels like unnecessary time - but maybe I’m really missing something?). I mostly have highlights (esp. from Kindle books, just because it’s easier that way than trying to “type” on my Kindle), annotated with my comments. I have these broken up by headers, so I can link to specific sections, and I figure that if I need to pull out individual sections to atomize later, I can do that.

Meanwhile, I already have almost 7000 notes. People are already asking why I have so many. LOL And I can’t even fathom how many I would have if I really took every separate idea from my every note and/or post and made them separate. (To put this into perspective, I find that I’m pulling in approximately 10,000 words/month that I’ve written in Facebook.)

Is this a matter of “if it ain’t broke, don’t fix it?” Or am I going to really be costing myself by not breaking everything apart?

How would you go about doing this? Would you treat the Facebook write-ups as permanent notes, and work backward from them? (That is, merge anything that seems redundant, and start linking these back to everything else?) How far do permanent notes really need to be atomized? I know I’ve seen some people talk about keeping them to a paragraph or so – but then I see examples of notes that are very lengthy, with lots of sections.

And on the far end of the spectrum, what do you do with very short snippets of information? e.g. I have something I captured that merely indicates the fact that “betaine hcl is a methyl donor.” Does that really warrant its own note? I feel like a note should at least be longer than a filename! Am I losing something by only throwing this into a larger note about methyls? Is this treated differently because it’s a fact/data point, not a “concept”?

1 Like

One thing to consider is backing up and then just going for it, breaking up all the notes into smaller pieces across the entire vault. And for the notes where you might want to maintain the original undivided note, you can embed the parts. For these, you might consider using some organizational or formatting technique to make it clear that it isn’t an actual original. Nested tags can be useful for something like this.

Another thing is to consider how often you are comfortable relying on linking to headings and blocks. Considering that eventually, based on the roadmap, they will automatically update upon name change, I think it is a very helpful tool to start implementing if you aren’t already. However this has the disadvantage of reducing the usability of the local and global graph.

There are some threads and newish techniques that rely on dataview to, in a way, treat blocks kind of like atomic notes. I can’t really vouch for these but the thought seems powerful.

Personally, I like the idea of breaking everything up quite small and repackaging it in multiple other places and contexts. However, one thing I avoid is copying and pasting content into multiple notes. However, I am thinking of some good use cases for doing so now that I write this.

Good luck. Thanks for the thoughtful posts.


First, everyone is different. A lot of this process of knowledge management is finding what works for you, what works in particular situations, and what your needs are. So, the following is my perspective, and doesn’t apply to everyone, but something might be helpful.

I think that writing things in your own words is crucial. This is a part of The Feynman Technique and also this. Our brains often fool us into thinking that we know things, when we are actually just familiar with them. Most people have experienced this in school. Read, re-read, re-re-read, “I’ve got this!” Then on test day that info we thought was locked in, was merely just familiar, and we can’t recreate it. So the Feynman Technique helps with that.

(Gotta run, may expand on this later.)

1 Like

If by “backward” you’re referring to reference/literature notes, then I would say, no. There’s really no reason I can think of for taking an atomic zettel and then putting it back into a reference note. You can include the source and link to it right in the note itself.

Length isn’t really the metric. It’s whether or not you are diverging in your idea. If you’re wanting to really stick with the principle of atomicity, then I would use “but” or “however” as a gauge. “Apples are good for you” is one note. You can add other supporting material if you like to build it out, but the basic idea is singular. On the other hand, “Apples are good for you. However, some people think too many apples can be bad for you” is two notes. It introduces a different train of thought requiring different support.

You don’t have to get at this level, but it’s the gauge I use 9/10s of the time.

As far as length goes. I think it’s just a general gauge to keep in mind. If you’re writing multiple paragraphs, you might want to check if you’re diverging into different lines of thinking.

Again, this is all if you want to abide by the atomicity principles. You don’t have to if you’re not using zettelkasten. LYT, etc don’t stick strictly to atomicity. Or, at least, atomicity comes later. It’s not a precursor for a note getting into the system.

1 Like

One thing that really helps when struggling with note length is transclusion.
You can have a large note that collects related concepts together by transcluding them.
Continuing @bobdoto ’s example:

# Apples

## The good
![[Benefits of Apples]]

## The bad
![[Apples are bad]]

## The ugly
![[Heritage apples]]

Then each of those transcluded notes can be more atomic, as you see fit, can be linked individually to/from other notes (giving better granularity), but can also exist in the larger context of Apples in general.


So I’m trying the “just link things as I go and see what emerges” approach, but I’m still concerned I might be tripping myself up here and I don’t want to have to go back and clean up thousands of notes if this is a stupid way to do this.

Most of my links seem to be more vertical than horizontal, for lack of a better way to put it. So, for instance, I have these six or eight notes that are about age segregation. They all link to an “age segregation” MOC. And that MOC, of course, links back to all of them (mostly in list form at this point, although I’m not sure if that’s how it will stay long-term).

These six or eight notes are not linking to each other, and that has me thinking I might be missing something here.

I definitely don’t know the answer, but find that it is helpful to believe that there can be a delicate balance struck between notes in the vault where there are admittedly too many and notes where there are too few connections. If you aim to homogenize the linking regularity simply based on number, I have found that sometimes the quality will decrease and the good stuff will hide.

But, when you know an area of the vault is being neglected, I have had some success trying to summarize the notes with a phrase. Then when you line them up, it can be interesting to try to take them at the face value of those phrases and clump them accordingly, intentionally ignoring all that you are aware of, lying beneath the surface. Anyways, I am not saying this is some super helpful method, I just think it is an example of the kind of experimentation that can get things turning. These non-destructive mini journeys are where is Obsidian really shines. This is especially true if developing unique thought processes are as important to you as developing unique thoughts.

Good luck.