Hello all’s,
I’m a PhD student in Deep Learning and I’ve started using Obsidian as a ZettelKasten tool for writing litterature reviews about all the papers I read (and to generate new ideas of course).
I’m hesitating about the best structure. Let’s say that I have ideas structures likewise A->B->C, with C being a sub-category of B, and B a sub-category of A.
Should the note C mention both A and B, or only its direct parent B? Knowing that B will mention its parent A.
A non-trivial question! Not to mention you also left out a third idea: should A mention B or C?
There are many approaches to personal knowledge management. I think a “true” Zettelkasten approach would imply that the mention structure only follows A → B → C.
But many others on here and elsewhere advocate for “living notes.” In this view, you’re trying to grow a set of thoughts over time. So the A → B → C chain will eventually split out and grow in new directions. You might also add D → A and D → C.
Andy Matuschak’s notes (linked again here) are a great place to start thinking about the latter view. Sonkë Ahrens’ Take Smart Notes is a good place to look for more info about the former view.
@Mougatine The question is “in what context do I want to rediscover this idea?”. If it’s important that you remember C when thinking about A then they should be linked. Thinking this through is a big part of building the knowledge base.
In practice when you begin to write you should be pulling a lot of notes together and seeing what ideas you connect (in Obsidian this will most likely be through building a summary/index/MOC page). You are correct that B will provide a pathway to C at this point - question is do you only want to see the A-C connection through B.
Also - there is no such thing as a “true Zettlekasten” approach. No ultimate truth has been handed down from on high. Luhman had a personal paper based system, many others have developed and used something very similar. The fundamentals in all those systems are: 1) read with a pen in hand and make fleeting notes (that are tied to references), 2) later make permanent notes in your own words from those fleeting notes that are valuable, 3) spend time thinking about how those permanent notes connect into your existing knowledge and file them accordingly. Everything else is “what works for you”.
Thank you both for your answers.
I’ll opt at first for the graph-based solution and see if it fits well.
Good idea about MOC, in my context A and B are definitely that kind of note. I’ll prefix them with ‘MOC’.
That said, it would be cool to change the color nodes in the graph view of MOC notes! Unfortunatly there is (yet?) not way to have a custom class (for css selection) per note…
Re: graph colours, no not yet. It isn’t very performant to render the graph’s nodes as elements of the Document-Object Model (DOM), so they are written to a canvas instead and therefore are inaccessible to CSS.
Someday maybe we’ll get graph style controls!
I agree, I just meant as close to Luhmann’s as possible. Far as I can tell his cards referred to one another card most of the time.
You can check this thread, got some answers that might help you