Presuming the behavior is real and not an artifact of the model. This is a statistical technique, needing specially selected targets in order to be observed reliably - and also assuming those targets themselves are typical.
There's plenty of observations which can accidentally vanish because of subtle problems with assumptions, so declaring a total refutation is beyond premature.
Like to wit, if MOND is real then you've really got to explain how sometimes it also selectively just bails out on some galaxies apparently[1].
I'm tired of people trotting out "epicycles" to attack theories they don't like: you're gonna be adding a lot of those to get a MOND which can explain all the data as well (which is to say, it's a trite insult and not useful argument).
Iirc, the udg makes sense if the distance to it is off by something like 25% and the orientation of rotation is off by a bit. Those parameters are very hard to measure in UDGs.
> I'm tired of people trotting out "epicycles" to attack theories they don't like
That's disingenuous. LCDM has a long history of failing to successfully predict later observations and adding parameters to fit the data, where MOND has made many successful a priori predictions without any added parameters since the 1980s. This is not just a matter of not liking something, successful predictions vs. post-hoc curve fitting strikes at the very core of what it means to be a good scientific theory. See:
There's plenty of observations which can accidentally vanish because of subtle problems with assumptions, so declaring a total refutation is beyond premature.
Like to wit, if MOND is real then you've really got to explain how sometimes it also selectively just bails out on some galaxies apparently[1].
I'm tired of people trotting out "epicycles" to attack theories they don't like: you're gonna be adding a lot of those to get a MOND which can explain all the data as well (which is to say, it's a trite insult and not useful argument).
[1] https://en.wikipedia.org/wiki/NGC_1052-DF2