In my humble scientific opinion, what you call a scripted destruction/animation and dynamic destruction are just the same kind of physics approximation, but on a different precision/scale/bias ...
And to continue with my analogy with RT, current games (with raster engines) are often more beautiful than RT tech demos, because of art, carefully designed textures, hand crafted pre-calculated lighting, etc. Even if the raster engine has more "bias" and is not as close to reality than a RT engine, it may look better due to clever tricks ... Yet I do not see a lot of gamers advocating the use of RT engines in games because those engines are less biased (if it was the case, I guess RT game engines like Arauna, Brigade and others would stand a chance against the FrostBite engine et al) ...
Same for physics : I don't see why it's bad to use clever tricks (scripted animations, biased physics, just enough particles to feel real, etc.) if it's to allow physics on the largest user base possible (and thus possibly offer physics that impact gameplay) ...
I sure would prefer broad adoption of "unbiased" physics in games (whatever that may mean), as much as I would prefer competitive RT engines for games, but it's not gonna happen soon. I fear it will stay a checkbox feature because it's too expensive to run on CPU, and even on low/mid NVIDIA cards ... "Scripted" CPU physics are perfectly fine as a mid-term solution.
(As for those who say physics are more efficiently computed on CPU and/or don't scale well with cores/parallelism : I though physics computations where embarrassingly parallel, and thus where particularly suited to GPUs ... Am I wrong ?)