but across many consecutive frames, the information is spread out temporaly and can be recovered (partially)
the same principle of how you can get a high resolution image from a short video, by extracting the same patch from multiple frames
It is predicting what the information might maybe have been like.
I get that what you're describing can statisically "unblur" stuff you've blurred with overly-simplistic algorithms.
I can provide you with real-world footage that has "natural" motion blur in it, if you can demonstrate this technique working? I'd really like to see how it's done.