The other big problem is that a lot of what made it seem natural and elegant has been progressively ruled out, so you’re left with far less simple and elegant versions which could apply to the universe. At that point you have to ask what the appeal is, you know? It’s not unlike the problems with Inflationary cosmology. Originally it was a simple way to resolve outstanding issues of homogeneity and isotropy of the observable universe. That was based on a model with a very natural PE curve, which has been ruled out. Now we’re left with a “curve” every bit as contrived as simply saying “because” in reference to the underlying questions.
In both cases, M-Theiry and Inflation, they persist because of huge investments of time and energy into them, and because as yet no strong competitors have emerged.