It's just sad. Nobody wants to fork it because nobody uses GIMP, so forks get almost no traction. People love the effort and commitment, install it, try it for a couple of tasks, fail, and never come back. It has built up this reputation of being terrible, and no one wants to give it a second chance after their first impression.
I think it's a lesson in how not to run an open-source project. I put it in the same category as Darktable, which gives off that exact same bad first impression.
> People love the effort and commitment, install it, try it for a couple of tasks, fail, and never come back.
If you just tried Photoshop 2 or 3 times, you'd fail to get anything accomplished either. GIMP is a Photoshop replacement, not an MS Paint replacement.
Maybe the problem is that not everyone needs Photoshop, or to learn a tool at all, and GIMP comes as standard on every Linux install. Most people never have a need for it. They only need MS Paint (or rather Paint.NET or Krita.)
Thinking that you should be able to just roll in and immediately get stuff accomplished on Photoshop or GIMP is a misjudgment of the skill it takes to use professional tools.
GIMP's UI is fantastic, and its capabilities are incredible.
edit: another problem is that people make a living from using Photoshop because it's an industry standard, so they're motivated to learn it. For GIMP you had to take the same effort to learn something that you couldn't use professionally. Hopefully the improvements on color and non-destructive editing will improve the situation on that front. It's becoming something that, if I have the latitude, I can use on the job at a print shop.
darktable is at least slowly actually getting better... I like what blender managed to do.