Would've been nice to see more effort made:
1. Using the version of a multiplatform game which was reviewed by more publications.
2. Checking to see if Metacritic or GameRankings used more reviews since a larger sample size equals a more accurate average score representative of a groups opinion.
3. Working out the real averages from Metascores and not using their weighted bullshit.
I'm not going to work it out but I reckon at least 10 games from the OP would/should be omitted doing it this way thus giving us a shorter but more accurate list of the 95+ club...
I can't help being overly scrupulous, meh.