I'm aware this would be extremely difficult; maybe impossible.
PS - I see you're at U as well. I work on a datavis team (iobio) in the genetics department (Steph says hi!). If you ever want to meet up I'd be interested in talking more about this.
I know (really, like, I know) how detrimental this would be to anyone's career, and I'm not saying this as a moral condemnation of what you're doing - just curious, as I've found myself that there are many cases in these circumstances where the interests of the author do not align with those of the audience. Just wondering if you feel the same way.
For your broader question---I understand what you're saying, but it's very difficult to edit someone else's writing. That's where "committee voice" comes from: it's the lowest common denominator to multiple authors working together. And often how I visualize things comes from how I look at things, and coming up with a visualization for how someone else looks at things is hard.
Take the OP as an example. This is a long blog post on gears in general, but animated by the specific question "what shape are gear teeth". If I were writing a blog post about gears, I wouldn't start at that place. And then, imagine if this blog post started text-only instead of visual. "Involute" would now be described with algebra, not a picture. The algebra is complex (compare the Wiki at https://en.wikipedia.org/wiki/Involute), and that algebra itself would need pictures. Illustrations and explorables aren't, ideally, something you sprinkle onto existing text.
And I see what you're saying on how it's hard to build on someone else's work, and how what is relevant to illustrate heavily depends on the viewpoint of the author. Still I can't shake the feeling that there is so much duplication. Maybe I should just look at differently. Anyway, thanks for weighing in.
Oh, found it. Noice!
Why is that your default assumption? Seems very strong when GP gave basically no details.
Because that's probably most textbooks.
Edit: Are people downvoting me because I'm actually wrong about textbooks, or is my answer unsatisfactory somehow?
The problem is that anything that is easy to produce and configure with a set of options is going to be very limited.
It is certainly possible to create an environment for producing mechanical and physical simulations with visual tools but it would not be trivial to develop.