In what ways can we form useful relationships between notes [LONG READ]

I’m now almost 2 months into Obsidian use, and I’ll say that I’m reasonably happy with where I landed, with slight modifications. As I noted somewhere up-thread, I’ve taken a hybrid approach, with relatively few base folders, and most of my notes going into date-based buckets. So I have a top-level 2020 folder, and in this an 06 folder for June, an 07 folder for July, etc. The main change I’ve made is that each new note starts with a DD designation for the day followed by a short, descriptive name. So notes created today would look like “15 My awesome thoughts on PKM”, where previously I hadn’t prepended the DD. I found a bucket full of alphabetically sorted notes less helpful than date sorted. My daily notes get just the DD, with no label. This naturally sorts everything in the File Explorer by date, which is great, and is perhaps my compromise with the hardcore Zettlekasten folks who prefer the 20200715104223 type identifiers. I average 3-4 notes a day including the daily note, so at the end of the month I’ve got 90-120 files in my monthly folder, which is a manageable chunk size for me. The daily notes, with their very short DD name, make it visually easy to scroll through the list, since the extra blank space breaks things up by day, and it’s oddly satisfying to open a new month with a fresh folder. For me, remembering roughly when I was thinking about something is a useful axis for recall.

I try to make everything a regular note, meaning it just goes into the date based hierarchy with the above naming scheme. I’ve also had a [small] bit of stress over the decision of “should this be a base note or go in the date bucket”, and try to just put stuff in the bucket, but be sure to link it somewhere. Topics will naturally bubble up that want to have a MOC to link them together, so I add them when I feel it’s necessary.


I’m starting from 0 with Obsidian and spent a decent amount of time last week reading interesting lines in this forum. This post really clarified many assumptions I’ve made and the approach I’d like to implement and leverage within my personal PKM.

I’m definitely leaning towards a radical flat approach based on “as atomistic notes as possible”. There is something which however is still now working in my mental model and I’m still pretty confused about… notes titles.

In principle, I’m leaning towards not giving notes titles and just navigating notes through backlinks and graphs. Reasons:

  1. Note titles seem an unnecessary layer and a first, altough inherently preclusive, attempt to classify a body of information/knowledge which I want to develop differently. I’d like the system to supports my thinking at higher levels of abstractions and spur unexpected connections rather than labeling and siloing things.
  2. Considering that I want to take super atomistic notes, in many cases, the title would just be a repetition of the note body.

The question is, would this approach be viable at all or is just supported by my current limited understanding of the topic? Would really appreciate if you can share your thinking.
To let you better understand what I’m trying to achieve, you should know that my use case revolves around:

  1. Consolidating professional knowledge scattered around years of folder based reading and note taking
  2. Develop knowledge in a variety of personal interests

Also, haven’t checked yet evergreen notes and the maps of content which seems pretty crucial concept to develop the PKM along the lines I’m foreseeing.

Finally, could you suggest some intro reading (but not too basic) about
ingormation modelling?

1 Like

I can see the argument for not titling. For me, titles are kinds of cognitive landmarks—when I read a title I can generally remember-imagine what’s in the note. I think that’s useful, but I’ve never tested the alternative. It might be worth trying it out—give yourself a week, leave titles off, and see how it feels.

I will have to get back to you on some readings on information systems/conceptual modeling. One of the gaps in the field is knowledge translation—that is, providing practical insights based on the theory. A lot of what has been done is focused on conventional databases and design for major organizational use, not individualized PKM as we’re discussing. So there might not be much on it as yet! I’ll have a look, though.

@paoloap Andy mentioned that Evergreen note titles are like APIs. The titles become abstractions for the notes itself, and an entire note can be referenced using the title.

APIs is a crucial component to effective high-level programming (higher abstraction level of thinking), since we are reusing fine-grained (atomic) abstractions (note titles). Using this analogy, there are several principles of API design that we can apply to the note title design (separation of concerns, etc). Over time, as our ideas mature, we go back and refactor the APIs (refactor note title and its content) to achieve a higher level of atomicity.

I think that’s one of the arguments for note titles. Personally I employ this approach. My note titles often start out as in-the-moment-conceptualization, and as I revisit them within a day or two I often reflect better and refactor the note titles to something more representative of the notes’ ideas


On a more practical level, how do you link to a note that has no title?

1 Like

IMO that is the crux. As with a book, or any document that carries a title, the title is the ultimate, super-concise summary of it. Like you, that’s how I see and use titles.

1 Like

At least some sort of slug in the file name gives a hint, 202007190807 Recipe - Grilled Red Snapper as opposed to just 202007190807 lets you look at files in a directory and have some clue. Even if the slug you use in the filename isn’t perfect, it is something.

Suppose five years from now you have 10,000 notes. Just looking at the folder of files with only numbers as filenames would kind of look like a black hole.
I doubt you would ever regret using some sort of slug in the filename. You might if you ever need to look into that black hole.

Filenames like these below at least give you the top entry points into your network of notes from a folder of files on the file system:


thanks for sharing, conventional db design and organizational applications would be super interesting to explore as well

Thanks for the pointer and for sharing your personal approach Andy was already in my reading list, will delve into it.

Revisiting note titles after a couple of days seems an excellent technique to refine notes conceptualization and enhancing future applicability.

1 Like

thanks for sharing, conventional db design and organizational applications would be super interesting to explore as well

Your reply should have gone to @ryanjamurphy.

Well, here’s a quick one: On Information Quality. It’s kinda tangential, but I think it gets at concepts fundamental to these discussions. It defines what information quality is and how it can be improved.

And a second, on the fundamental cognitive challenges of classifying information in databases: Using Cognitive Principles to Guide Classification in Information Modelling.


Well said, I agree and follow the same process.

It’s always nice to see @ryanjamurphy’s PDFs, full of generous highlights. Thank you for the references!


+1 one that:
Every system or workflow that requires some (pre)post-processing to keep it “organised” will not work when you want that workflow to last for years.
There is also the notion of “ongoing insights” which is the factor that over the years your insights on things change (normally because you became smarter on a subject :wink: )
In a tight system where everything is boxed and separated you will not (believe me) reshuffle anything when your insights change. Nor will you go through years of note to see if you didn’t miss anything.


Where is the spark?

Or to be more precise - what should be the spark igniting insightful thinking when using this or any other PKM?

I’ve read ‘long reads’ by @nickmilo, watched few lectures by Sonke Ahrens on Zettelkasten, tried to replicate IMF set-up after I’ve gathered up enough (i.e. too many) notes, but now what?

What exactly should “spark” that thought? Should you go twice a day through your cards like Luhmann picking random notes (I know there is a plugin) or re-visit/update MOC’s periodically, or stare at the “Graph view” as if the stars in black sky and wait for inspiration?

Or maybe this very work of organizing/management of constantly changing personal knowledge should have a value in it’s self (journey not a destination)? Keeping in mind -


Johnny Decimal looks cool. I’ve been thinking that some type of Dewey system might work for me…might give this a shot.

Imagine today that I write down my thoughts about an object-A.
I title my file “29 My thoughts on Object A”, and I put it under 2020/10.
Three days later, I have a new thought for object-A.
What to do?
It’s only been three days, and I remember “29 My thoughts on Object A”.
I could add this additional thought in 2020/10/29.
Yes, but this thought arrived on 2020/11/01.
And if I put it in 2020/10/29, I betray the rule of time.
If I follow the rule of time, I produce a new note titled “01 An additional thought on object-A” and then put it under 2020/11. And so as not to lose the thread between these two notes, I create a link from one to the other.
Yes, but is this link relevant?
Because this link has no other purpose than to overcome a temporal gap between the sequence of my thoughts on the object-A.
I could eliminates the need a link, if I was willing to deviate from the time rule.
I need to be able to assess whether the link makes sense.
Was the context of these three days really necessary for this little extra thought?
Because this thought isn’t really a new thought, just a little extra thought.
An oversight.
It’s not really an oversight, but it could be like an oversight.
I must know how to estimate the load of difference between the thoughts which concern the same object.
A significant amount of difference would have made me perceive this thought as sufficiently individual. And then I would have gladly placed it in 2020/11/01.
And I would have linked them together.
But when I look at this little extra thought that was rejected in the next month, it seems to me that it belongs by right to the three days before it, like the latecomer of a procession.
Have you encountered this kind of case, and how do you resolve it?
Do I have to relentlessly follow the law of time, to avoid any friction related to the location of things?

Hoping that my message is not too confusing, English is not my mother tongue.

@Nato Yes, I do come across this issue of having additional thoughts down the road, maybe a few days later, and I have to decide whether to create a new note or add to an existing note. Since it’s my system and I’m not worried about the Zettelkasten police coming after me, I just pick whatever seems right at the time.

[3 minutes later]
I’ve been known to add an addendum to the end of an existing note which looks like this – a dividing line and a bracketed date. If I ever care in the future, I can see that I added the thought after the original note, but I’m not sure this will matter much in a year.

As a matter of practice, I try to have all my notes linked to both time and some topic. The folder structure I outlined naturally provides the time axis, and if I also link each note to a relevant topic MOC then I have both a semantic and a temporal link. Note and Block transclusion makes your concern about when to combine notes less stressful, I think, because you can capture the thoughts as individual notes when you have them, and then just include them in a single note via transclusion, so the individual bits all appear in one place. I’ve found that this makes it easier to not worry about how things are divided.

1 Like

When I am confronted with my chaos of notes, and the means to find my way around, I experience a lot of hesitation between the semantics and the temporal. You seem to have found a good compromise between the two.

My discovery of zettelkasten is recent and I am digesting what I have learned on the subject.
It’s fascinating, but I don’t know if it’s what I need yet.

I’m thinking of a flexible system, with which I can concentrate on the content, while minimizing the energy expenditure related to the management of this system.

The difference in applications accessible between Windows and Mac is edifying.

I’m trying to build what’s good for me but I’m not satisfied yet, there is a lot of confusion, unstable localities.

There is a purely practical dimension, but also a sensitive and psychological one.

Thanks for your sharing.

1 Like

@Nato, here are some random thoughts on the problem from my perspective:

Imagine today that I write down my thoughts about an object-A.
I title my file “29 My thoughts on Object A”, and I put it under 2020/10.
Three days later, I have a new thought for object-A.
What to do?

There are no absolutes to this problem because it depends on several factors:

  1. Do dates represent creation or modification? Can it be either?
  2. Is the note anchored in time? (Does it represent an event?)
  3. Does a later thought serve a purpose being a separate note?
  4. When is it important to have a trail of modifications?

On creation and modification dates

The importance of either creation or modification date depends on the context. I think that both are important; sometimes creation date is fairly meaningless, and sometimes it’s critically important. If the note represents things that should be “maintained” in some way (for example, a packing list,) then modification date is more important to me because I don’t really care when I started the packing list, but I do care when I last updated it. In your specific example, for me personally, it wouldn’t matter that I had the thought three days later if it serves no real purpose to separate that one thought.

On notes anchored in time (representing an event)

Let’s use the example of going to a doctor appointment. You’d have a note for that visit with all of your thoughts. If something occurred to you three days later about that visit, what you’ll do with it depends entirely on the relationship to the visit. If the thought really is about the visit (for example, you forgot one thing the doctor said,) I would absolutely add it to the existing note. By my way of thinking, it doesn’t matter that the thought came three days later. It’s about an event anchored to a specific place in the timeline. I think that it would serve a disservice to do otherwise. However if the thought is at all tangential, it would be a separate note with perhaps a link in the original note. Again following the same example, let’s say the doctor told you your Triglycerides are too high. You recorded this in your visit note. Three days later, you remembered that you read an article saying that there’s preliminary evidence that Cinnamon lowers Triglycerides. This would be a separate note with the details / citation. You could link this to the visit note if you want the association to be stronger than simply having similar keywords. In this way, when you revisit the original note, the train of thought leads you right there.

On later thoughts being separate notes

I think that part of this decision is intuition. You must ask yourself if that later thought really serves a separate purpose. Using the phrase I used above, is the additional thought at all tangential to the original note? If yes, it might need to be separate.

I could add this additional thought in 2020/10/29.
Yes, but this thought arrived on 2020/11/01.
And if I put it in 2020/10/29, I betray the rule of time.

I don’t believe in a rule of time. The thought is either strongly associated to the original note, or it isn’t. Unless you’re doing research into how long after you’re exposed to a concept do you continue to have novel thoughts about that concept, I don’t see the value in an absolute rule that says every thought must be timestamped when you have the thought.

On the importance of modification audit trail

There may be contexts in which it is important to you to see modifications to a note over time. If the changes tend to be singular and not often, these can be managed in the note itself if the data is amenable to being date-stamped within the note. But if it is expected that there will be many changes at once, it might be better to strike a new note as a copy of the previous one, with a new timestamp, containing that latest version. The version list could be maintained in a MoC, or the versions could be a linked list forward and backward, or both.

I hope this helps.