I’ve been reading a bit about the Zettelkasten system and there are some interesting opinions about meta data and how best to think about information. This old blog post by Christian Tietze about categories is thought provoking.
Creating categories is a top-down process. You start with the structure and then file the material away. Notes will have to fit the structure. If they don’t, there’ll have to be a compromise.
This is certainly one approach, and one that I've taken several times. It doesn't work well unless the project is very well understood. The alternative is to implement a constant review and update to the project. That's a lot of work to commit to.
So if you model your knowledge management system to fit the way your brain works, you better not start with inventing a hierarchy of categories, top-down. Instead, you’re better of starting to collect notes and see what happens. Let things grow in your Zettelkasten as you let your brain do its work or organic growth.
This is more like how I work now. I use tags in DEVONthink, but those are also folders which are kind of like categories. Flexibility is the key. The entire point of categories or tags, or whatever, is to find relevent information when I need it. To reveal that piece of data I collected months earlier when it is most effective.