Reading a review from Roger Ebert provided a lot of value to me - I knew generally his likes and dislikes and could use that information relationship to contextualize the review. I think reviews break down when you move from individual to aggregates. What does 34% really mean, maybe 90/100 reviewers hate horror, are under 25, all from a specific city? I really don't agree with your take that on reviews but I do think reviews in aggregate - where there is no relationship to the reviewer -- are problematic.