yole's journey in languages

I’m currently in the middle of (reluctantly) migrating my note-taking system from Tana to Obsidian. Given that Tana is still in early access and not many people are familiar with it, I’d like to share some background and information on why I’m doing this.

While I was at JetBrains, I didn’t use any formal note taking tool until I discovered Roam, and it met my needs quite nicely. At some point during my last year there, I learned about Tana, and the desire to try out the shiny new thing combined with a smooth migration experience encouraged me to switch.

As the university year started, I began to use Tana much more heavily. During classes, if I don’t need to use a laptop, I prefer to take notes on paper, as this helps me a lot with staying focused. But everything else – notes on my readings, homework, notes on my projects, random articles or presentations I was reading – I put in Tana. And this is when I started running into Tana’s limitations.

In general, I find Tana’s approach to taking notes better suited to how I think than Obsidian’s. I don’t particularly want to think of “a note” as an individual thing that I have to manage; I’d much rather simply enter information somewhere into the system and then access it through backlinks and search. (And in a way, Obsidian’s approach, where I know that every note is a file that sits in a folder, feels even worse than something like OneNote, where a note is a thing but the storage is completely opaque to me. Even though I completely understand the arguments for flexibility and the lack of lock-in, I guess I spent far too much time managing files over the course of the 30+ years I’ve been using computers, and I don’t want to do this when I can avoid it.)

However, Tana’s “everything is a bullet point” approach, which enables its super-flexible information structure, also leads to the limitations that I’ve run into. The biggest one is lack of support for tables. You kind of can do tables in Tana, if you assign a supertag to a bunch of nodes and define some properties for it, but it doesn’t really work for ad-hoc tables that come up so often in linguistics. I ended up stuffing all the table-like information into bullet points, which was hard for me to read afterwards and not really natural.

The second downside of Tana’s approach is aesthetic. Because everything is a bullet point, everything looks the same, and Tana’s formatting possibilities are quite limited. Obsidian with a good and colorful theme (I’m using Nord at the moment) and with the full set of formatting capabilities afforded by HTML quickly became a much more comfortable environment for me to work in.

A bunch of other downsides of Tana are not as significant, but they add up. It’s not great that there’s no superscript/subscript formatting (this also comes up often in linguistics), it’s not great that Tana does not work offline, it’s not great that customization options are very limited. At the same time, I found myself not actually using many of Tana’s more unique features, such as supertags, search nodes, and especially AI, which I actively don’t want to use and which seems to be a major focus of the development team’s efforts.

So, after yet another not quite successful attempt to find a readable format for my Hittite homework in Tana, and after seeing a pretty Obsidian installation on a fellow student’s laptop, I decided that it’s time to migrate. I’ve partially reverse-engineered Tana’s JSON export format, and I’m trying to contribute Tana support to the official obsidian-importer plugin. (At the time of writing, this is still work in progress; you can find the current version of the plugin here). So far I’m quite happy, but this is still early days; I’ll share an update in the future.