One major goal of a visual representation would be to enable refactoring. I think as your Tana graph grows and you realize you need to change supertags around or break things down or perhaps introduce new ways of structuring data, it's really hard to get a mental model of the super tag hierarchy and fields (especially when working across multiple workspaces). Then there needs to be an easy way to see what content is impacted by changes. Overall I feel like Tana becomes more brittle the more it grows.
Yep, just seeing the supertags as a graph would already be super useful vs the regular requests of getting a graph of everything, which is hardly useful (as many existing apps have shown)