Be Suspicious of Online Movie Ratings

Pretty simple, people see movies they think will suit them early so will rate slightly higher than the rest who see it later / on DVD and rate it less.

There's a type of movie that fools critics and American Hustle was one. I think it was great but it was over rated. But it took a while for that to get out.

Rarely it goes the other way where movies break genre so the keen people are a little disappointed but the rest are pleasantly surprised and the rating goes up.

I miss David Stratton and Margaret Pomeranz.

One is just pure statistics and design. The typical "average" review score is a mean. And a mean tends to bias towards higher scores, especially when the floor isn't zero. Consider Amazon reviews, for example, which is a 1 to 5 star system. Amazon has half-star increments for the average rating, so you'd think that leaves a lot of dynamic range for reviews, but in practice only the very highest averages are worth considering, everything else is a crap shoot. Consider something that is near universally considered bad by reviewers, it takes 3 1-star reviews for every 5-star review to bring the average down to a 2-star. A 2:1 min:max ratings ratio leads to the same average, and a 1:1 ratio leads to 3-stars. A 1:2 ratio (where 1/3 of all customers have an awful experience) results in around a 4-star average. So you have a situation where a 1-star average means something is universally horrid, and 2 through 4 stars are just variations of awfulness, with only 4.5 stars and above meaning that more than 3/4 of reviewers had a good experience.

Obviously the better metric would be a median, or something similar, but that would lower the average ratings and lead to fewer purchases, most likely.

The other big problem, especially with media, is that the ratings are from self-selected groups. In many cases, especially media, the only people who are going to go out of their way to experience something are folks who are going to be predisposed to liking it, while most other people stay away and thus won't even rate it.

All of which is beside the elephant in the room, which is the fact that in art there's no such thing as an objective universal scale of quality. You absolutely cannot place movies, books, television, or games on a linear number line of "quality" and have those numbers be consistent and sensible. The best that you can hope for is a general sense of whether a lot of people like something or not and the intensity of that like, beyond that you need to actually dig into what people are saying, and who, deciding if it's worth the risk to watch/read/etc. and then deciding for yourself how you feel about it.