For one thing MOND doesn't actually appear to eliminate the need for dark matter[1]. It gets rid of ~80% of the missing mass requirement, but not all of it. Whe

For one thing MOND doesn't actually appear to eliminate the need for dark matter[1]. It gets rid of ~80% of the missing mass requirement, but not all of it. Whereas pure DM can just eliminate MOND. Then you've got the "Bullet Cluster"[2] - where two colliding galaxies have had their observable and dark matter masses apparently separated. MOND can't explain this one without a lot of tweaking, but it's pretty trivial for DM: electromagnetically interacting matter is "sticky" where as gravity only matter isn't. The Bullet Cluster shows a galaxy shaped blob of gravitational lensing exactly where you'd expect it to be if a bunch of non-interacting matter had flown through each other, whereas the electromagnetic matter has interacted and re-shaped. NGC 1052-DF2[3] and NGC 1052-DF4 are both ultra-diffuse galaxies which have no, or very little dark matter. That is, they appear to have normal galactic rotation curves fully explained by their observed visible mass. This works totally fine for DM existing (it's a problem for lambda-CDM though because it's not clear how they could've formed without dark matter, but I mean - we also don't yet know how black holes actually manage to ever merge either yet we do observe them too). This one always seems like a problem to me: MOND proposes a new universal principle of matter, then suddenly we have some matter where it's not doing that. The theoretical problems[4] are somewhat beyond me, but they get well into issues with violating relativity and that's a big one: relativity is stupidly, reliably accurate under every single test we put it through, to absurd levels of precision. Build a better instrument, you can just dial in your precision and get the answer out ahead of time before you launch the satellite which is testing it. Also without careful adjustment you get violations of conservation of momentum (conversely, if MOND is real this would be handy because maybe it means we can reactionless spacedrives). It's worth noting that none of this is implicitly fatal. lambda-CDM could be wrong, a MOND variant could be right. But a list of convenient things MOND explains easily doesn't escape the need to also include the things it can't - and appeals to the idea that DM is being "tweaked" to match observations unnaturally ignores the fact that MOND has to have the same thing done to it to fix within cosmology. [1] https://arxiv.org/abs/1404.7525 https://arxiv.org/abs/1404.7525 [2] https://en.wikipedia.org/wiki/Bullet_Cluster https://en.wikipedia.org/wiki/Bullet_Cluster [3] https://en.wikipedia.org/wiki/NGC_1052-DF2 https://en.wikipedia.org/wiki/NGC_1052-DF2 [4] https://arxiv.org/pdf/astro-ph/0104435 https://arxiv.org/pdf/astro-ph/0104435