See, this is the cool thing about Science: even if you don't understand it the details, you can understand the process: fitting ideas to observations, trying to come up with new ideas to fix where the previous ideas don't line up with the observations, testing to see if those new ideas fit better, deciding the new ideas are not (yet) correct or are qualitatively better than the old ideas.
Lather, rinse, repeat.
I have no idea whether Dark Matter is a better solution than New Gravity (or whatever it's called), but I love the hell out of the idea that people are discussing it and testing it and figuring it out. It's not that Science is always changing — it's just that it's constantly triangulating around what the incredible complexity of Reality.
It beats the hell out of "I want to believe this, so you're wrong" as a means of figuring out the Universe.
New ideas on gravity would vanquish dark matter
New ideas on gravity predict rotation curves of galaxies without dark matter.
I've never liked Dark Matter. It has always struck me as the ultimate physics 'fudge factor', i.e. 'can't make your math work out? throw in some Dark Matter, give it whatever qualities you want it to have, et voila, the math now works!'
+Eli Fennell There is that — though the challenge has then been to hypothesize what it is, why it is, how it behaves, and how all those explanations then still fit into the observational data. Again, Science!
+Dave Hill Actually, to me, it's the opposite of science, it's pseudoscience. In science proper, you can't explain one unknown by another, and that seems to me to be all Dark Matter has ever been, the one unknown that can be bent and twisted into whatever form the researcher wants it in to explain another unknown. A true scientist would be willing to say "Gravity only explains 10% of why galaxies hold together, and we don't know what the other 90% is." A pseudoscientist invents a magical substance that can't be seen or felt, interacts with nothing in any measurable way, and then he imbues this magical substance with whatever qualities it needs to 'explain' the other 90%.
+Eli Fennell I'm okay with ascribing the explanation to something we can't yet understand or characterize … as long as it fits the data, and can be tested against.
+Dave Hill Anything can be made to 'fit the data' if you can make it behave however you want it to.