In my opinion, an advantage would be that it removes some steps/friction while writing a paper.
For example, let’s say I’m working on a paper called Draft 1.
I start typing a paragraph and at some point introduce [[keyword 1]]. Automatically, a list of notes could appear in a sidebar that also have a link for [[keyword 1]]. The list could be expanded or collapsed under [[keyword 1]]. If I continue typing and introduce [[keyword 2]], a second collapsible list of notes using [[keyword 2]] appears.
The local graph provides such visibility, but in a way that (for me) is more cumbersome than a simple list of notes with [[linked]] keywords in common. It also quickly becomes crowded the more keywords I use.
Tags are useful as well, but would require additional steps. I could search #keyword1 to see a list of notes with that tag, but would have to search #keyword2 separately, increasing friction and steps.
I could also use Dataview (and often do) or use a simple query block to generate a list of articles with the linked or tagged keywords, but that again increases friction/steps.
Finally, I could create an MOC/index note for each linked/tagged keyword, but this yet again requires me to leave Draft 1/open a new pane in order to have visibility on notes that are connected via [[keyword]] or #keyword.
Maybe I’m being picky, but this would be an awesome quality of life feature, helping me to improve efficiency and workflow.