The truth is this. Dark matter is observationally the best theory that we've got. However it is deeply unsatisfying. It requires fundamentally new physics about fundamentally new stuff with properties that we have absolutely no clue about. It is literally a theory of, "Insert magic cosmic glue here."
Virtually any idea can be inserted. String theory is popular despite having made a single verifiable prediction in decades of trying. So let's say that dark matter is made of strings!
Maybe if the Everett interpretation were more popular we'd theorize that a proper unified theory will have a small gravitational interaction between quantum superpositions. So what looks like dark matter is really the gravitational interaction with the superpositions of the stars in the galaxy that have been evolving since the early universe. Thanks to the fact that multibody gravity systems are chaotic, every tiny variation grows over time until those superpositions just act like a smooth smear, which we can't directly perceive because of quantum decoherence. Is this a reasonable theory? Don't ask me, I just made it up. But I know that it doesn't require any new kind of matter - it just requires a bit of speculation about interpretations of QM and the nature of quantum gravity. And it would look just like dark matter does.
The truth is that we've got a theory, and we know how to fit the data to it. But that theory doesn't integrate well with all of our other theories. Therefore, no matter how well we've made the facts fit, intellectual integrity requires that we remain open to the idea of being wrong. Not so open that we stop pursuing what *WE* think is right. But open enough to recognize that other people's discomfort with the theory is actually somewhat reasonable. No matter how strongly we might think that it is right.
