Tags should have nested hierarchy:
Without this, they are difficult to identify and select. The community has been working on solutions for over a year now, but nothing has come: A Master Tag System (advice and feature requests) (Tag hierarchy)
Evernote popularized it back in 2008, so either you’re deliberately trying to free yourself from this obvious and structurally essential need, or you don’t mind being 15 years technologically out of date.
This feature is not so important as you imply. The feature request has only 3 votes. It is not essential feature and can be reproduced with sets if it is so important for you.
Sets are useless:
They were a very good idea, but incomplete. You can argue that it was just a matter of time before you could code and integrate the Collection concept, but I sincerely believe that you didn’t foresee this need and that the community’s insistence pushed you to conceive this integration. Except that Collections are a blatant example of an overly narrow or rusher vision.
Why is that? Simply because there should be no distinction between Sets and Collections. There should only be Collections, and these could just as easily integrate different Types/Relationships, as a single one.
Making these two entities cohabit doesn’t make any sense; you’ve complicated the structure for nothing. You’ve responded to a legitimate need in a chaotic way.
This is just false. There are many use cases for sets that are not possible with collections. Actually, I could argue that sets are superset of collections.
Our unreadable data file structure is locked in a big humanly unreadable container file. You should have come up with a format that’s readable without the need for your application, and that lets us navigate through a structure representing our data. Even if the hierarchical approach seems complicated by the nature of Anytype, it’s not impossible. Like Obsidian does with markdown, which doesn’t lock up data locally, but secures it in transit/sync.
Markdown is necessarily limited when it comes to integrating and representing Anytype data, but a solution is possible - you’ve simply chosen not to consider it.
For an application that claims digital sovereignty, it’s more than inappropriate to lock up our data. But since you offer Offline-first, I’ll go along with it, despite my disappointment.
If you want, you can write program that would process the protobuf files. The files are unreadable for humans for a reason - you can not store such objects in markdown.