To put it short:
Academic "weight" works much like google - the more citations paper has, the higher it is ranked. Of course a citation from local university journal (probably every university has at least one) is worth much, much less than a citation from Nature. It is calculated from Impact Factor, that is in turn calculated from number of citations. Sort of a closed loop.
This would not be too much of a problem if not for funding. You see, funding, that is necessary for anything non-trivial, is usually awarded according to researcher's reputation - something again based on impact, citations, etc.