One of the mental problems that I think people get in this area is that you can't help but mentally imagine that there's just these few points of data that we need to hit, and so we have a lot of freedom in how to construct formulae that match the observation, so why are scientists being so obstinate about the possibility that MOND may be true?

But that's not the case. We have petabytes of data from space (if not more). Your theory needs to match all of them, give or take noise. The common example of the Bullet Cluster [1] comes up a lot, but it's just one particularly exciting example of the problem. I've seen people create variants of MOND that try to address the Bullet Cluster problem, but you start having to have some sort of really weird gravity memory effect in space or something, where space itself "remembers" that it used to have a lot of matter in it and still has a small amount of gravitation in it despite no longer containing the corresponding mass. Which I suppose isn't necessarily out of the question given that we're intrinsically discussing a non-linearity in gravity, but it's still a bit strange.

By the time you work out what MOND is going to need to look like in order to explain everything we see in space, it ends up losing its "Oh, it's just one more simple term on gravity" nature and starts looking pretty weird itself. It's constrained by the real data to be at least as weird as dark matter itself, because the data itself is weird. Given the history physics has had with "calling its shots" and identifying the necessity of particles years or decades before they are detected, dark matter is something that I consider at least not out of the question.

I also am sympathetic to MOND solutions as well. After all, "the math looked awfully weird by the time we finished working out the theory" also has a long and illustrious history in physics too! What's true is true, regardless of how we feel about how beautiful it is. We also have a Standard Model of particle physics that, whatever its known flaws may be, doesn't immediately obviously have a hole for dark matter in it, so perhaps the "we've called our shots before" argument doesn't apply so well here. There is something to be said for not trying to push a new particle in there. But if MOND is the answer, you're likely to find the final result to have non-trivial weirdness in it, too.

[1]: https://en.wikipedia.org/wiki/Bullet_Cluster#Significance_to...