I just wanna point out that while everyone is free to design their own interface, it is only valuable if there are lots of algorithms that can exploit it. At the moment, the main source of algorithms is Graphs.jl, and that seems unlikely to change.
Given the lack of maintainer time compared to the size of the codebase, a complete algorithmic rewrite is out of the question. This means we need existing code to work with minor adaptations. In fact, rethinking the interface is already a major undertaking, that may even prove too breaking or time-consuming to complete.
It’s awesome that this topic has sparked a lot of discussion, both here and on GitHub. But I’m begging you: if you have energy to spare, use it to try and find a common ground that is both consensual and achievable. That might require compromising the ideas which would have gone into a solo project, but you get to reap the benefits of a whole ecosystem in return.