Two of the major contributions Ton made though are relevant.

The Blender team did not always accept code or suggestions. This has been a running theme with several people I've known that felt their work and/or ideas were rejected by people that didn't grasp their brilliance. There was a possibly unusual willingness to say no, but it was more discerning than with GIMP which gave off the appearance of vetoing virtually everything. (At one time all GIMP woes would be solved by CinePaint aka "Film Gimp").

But it was combined with the idea of the studio, in order to find out where exactly the pain points are to be addressed. In a sense this is agile software done right, where you get the users and devs alongside each other with a common goal. Unsurprisingly one result is the UI today is not mocked in the way it was 20 years ago, while the GIMP UI has remained a constant point of confusion.

> In a sense this is agile software done right, where you get the users and devs alongside each other with a common goal.

I'm not sure Blender development is "agile" in the traditional sense of the term because from the outside there are signs that it is slow paced with high inertia (specialists in charge of large features, planning over many months, purpose-specific contractors and GSoC projects, features that are delayed until they are ready), but they are certainly successful at delivering copious, high quality and high value features. Let's hope that the good leadership continues.

> while the GIMP UI has remained a constant point of confusion

If only...