This has the same failings as a system of information distribution where an aristocracy is given the authority to decide what is and is not important and correct. This introduces all sorts of random factors having nothing to do with any sort of objective measurement of the publication.

Those random factors don't even have to be nefarious; an expert in an area may not have sufficient sub-domain knowledge to fully understand the work of another author. The hope is that the publication is clear enough that they can, but every publication has to make a trade-off about how much knowledge they assume the reader has.

Each of these random factors chisel away at the intended meaning of the certification. If enough of these accrete, you end up with something that tells as much about a potential publication as a GPA does about a high school student. But then, people depend on things like that for hiring and promotions as well. It would be a damn shame if this is really sufficiently optimal that everyone stops working for better.