You're defining dark matter to be "however gravity looks" which is my point. That's not necessarily helpful. You can't be sure that by winnowing down the parameter space, you're aren't just making an equivalent statement saying like "we are fitting these points to a nth order polynomial and the derivative must be allowed to exceed this value instead of being constrained". In other words, constraining the properties of DM, especially in the direction of lowering the upper bound on it's granularity and lowering the upper bound on it's interaction with normal matter, could artificially increase its independent explanatory power.
That is true in general, but the theory "hey what if there was lots of invisible stuff" was developed long before further study confirmed that "gosh it sure looks like there's lots of invisible stuff". So you'd need a really contorted idea of gravity (at this stage heavily backfitted) to get the results that just fall right out of dark matter, even as conceived of before we got further confirmation.
I'm sympathetic with the idea of "invisible stuff" - classical energy is famously "invisible stuff" that has eventually gotten explained away in different ways, and has a couple of corners that haven't been battened down. But it's also never been categorically described as something with X properties, it's always been accepted as "something unaccounted for for which we need SOME OTHER model to explain", even after it became Noetherian (you still need to define the field).
Maybe I don't understand your usage but it is rather confusing that you call MOND heavily backfitted and not dark matter. The traditional complaint about dark matter is that it is highly parameterized, requiring an empirically fit factor for each galaxy observed. Am I missing something?
I think that where we disagree is in that last bit, where dark matter needs to be fit to each galaxy. That's absolutely true, but that complaint goes for regular matter too! Every time the location of some body was predicted from the motions of what we could already see, the theory of light matter was fit exactly to the data. What makes it stay within the bounds of Occam's Razor is that before ever noticing oddities in the motion of planets, we could say that probably matter would be clumped up, and could affect other matter.
Similarly dark matter has to be fit to each galaxy. However, before doing the fitting, dark matter suggests that galaxies would behave not just heavier than their matter content would suggest, but also that they would be lopsided. That is the first prediction you'd make as soon as you were told to consider invisible matter, even without seeing that the data fits.
More importantly, dark matter also elegantly explains the behavior of other massive objects, like galaxy clusters, which MOND struggles with. The figures that roughly work for galaxies don't predict anything like what is seen in clusters, suggesting that they work by coincidence on the more common types of galaxy. That's not to say we don't have more to learn about gravity; to my admittedly limited understanding weaker flavors of MOND could explain the effects of Dark Energy. Dark Matter is just a more elegant explanation of the extra mass issue.
No, it's not. It's explicitly multimodal. We estimate material distribution of baryonic matter by measuring light output as a proxy and then derive their presumed gravitational effects by observing their motions, which is a totally different method. Now you can argue that those estimates are poor, but it's categorically different from dark matter, where there is only one measurement that drives both its distribution and its presumed effects.
> More importantly, dark matter also elegantly explains the behavior of other massive objects, like galaxy clusters, which MOND struggles with...
This is a bit like saying a seventh-order polynomial adjustment to this 1st order fit "elegantly" explains these data points, so we should throw out the 2nd order fit because it makes us question the 1st order law we started working with.
To shore up my point from earlier, I was specifically talking about how we spotted Neptune without ever seeing it by observing the stuff we could see, just like how we identify Dark Matter. Looking back I definitely didn't make that clear, plus the fact that we already knew that normal matter existed muddies the waters further.
Now to your points specifically. Dark Matter needs 0 modification to explain superclusters, normal clusters, and individual galaxies all at once. The basic principle is hard to swallow, obviously, but it made predictions about the sort of deviations we'd see that turned out completely correctly. While certainly if you could just plop dark matter anywhere you wanted and hold it 'stationary' you could overfit to any data you wanted, the models that work have dark matter right alongside regular matter, driven by the same forces. The only reason its possible to derive such models at all is that galaxies behave so much like there's more 'stuff' there that we just have to fill in the blanks. Brute forcing a dark matter solution to a world driven by MOND would be at best a feat, and more likely close to impossible.
Total side note, I was snooping in your profile and saw that you wrote Zigler. I actually just today tried it out, funny coincidence no?
