IF the authors have a history of poor work, or no history at all, or a bunch of stuff that has been discredited, then the likelihood that they have solved a big problem declines.

People who do massively important work typically telegraph quality by doing earlier good work. That isn't always true, but it is commonly true.

You could say that estimating a paper's worth by the quality of the authors' previous papers is an approximate solution to the NP-hard "paper quality assessment" problem, which offers an exponential speed-up. :P