I think that the weakness doesn't lie within GIMP itself.

Imagine that you are a car hobbyist. You know your way around a wrench.

But then you step in to an F1 garage or even your local repair shop run by that one guy who inheritted his father's shop in the 50s and has thrown a tool away since the Reagan administration.

It's going to be possible for you to do everything that you know how to do, and even to learn some things along the way, but you're not going to be anywhere near as efficient as you were in your garage where the only tools you have are the ones you regularly use and you know the locations (perhaps roughly) of everything.

The same could be applied across any number of domains. Knowing your way around and ambulance isn't going to go as far as you might think it would in a surgical suite.

Knowing some python isn't going to get your pulls accepted in Canonical, Debian, etc.

Knowing your professors preffered citation methodology isn't going to gaurantee academically succesful searching of The Library of Congress or even the New York Public Library.

etc etc etc

GIMP represents nearly the totality of knowledge relating to image manipulation, and you can lay it out to perfectly match your personal knowledge and workflow, but it simply is not possible to have it automatically laid out to perfectly match everyone's workflow.

Could it be more intuitive? Perhaps, but moving things around now is liable to break the workflows of tens of thousands who have learned to use and love GIMP the way that it currently is.

For instance, having only ever used GIMP as my primary image manipulation tool, I can and do have some of the same complaints against [insert other software] that people routinely level against GIMP. The last time I tried to use Photoshop I spent more time in tutorials and help pages than doing actual image editting because Photoshop is as unfamiliar to me as GIMP is to a Photoshop user.

Interview with Øyvind Kolås, GIMP developer (2017)