Most mouse GUI actions don't require complex curves even when they appear to. You can see this when you look at GUI recording and automation scripts generated or supported by tools such as Sikuli [1]. The better approaches don't even require precise coordinates for most actions let one complex curves/lines.
What the mouse does provide is a suitably generic input device that allows exploration and discovery. A strict CLI style user interface doesn't tend to have this level of possibility without rapidly becoming overly simplistic. Readline support is a good attempt for most realistic use cases but still requires a learning curve.
Final note for people who think mice are intuitive: back in the day I had to train people on how to use the mouse button to click a onscreen button. The two events are not inherently obvious - they just appear so now because our culture has now adopted it. The bar has been raised.
I tend to agree.
Has someone made a
"considered harmful" considered harmful?
“Considered Harmful” Essays Considered Harmful
Article: https://meyerweb.com/eric/comment/chech.html
Discussion: https://news.ycombinator.com/item?id=9744916
And while I prefer the keyboard for most interactions, most users would probably find the mouse more convenient because the mouse is almost always easy to use without memorising any shortcuts. Every website and program is fully accessible with the mouse, but the same can definitely not be said about the keyboard nowadays.