I mean, it would be somewhat surprising if the results of yes/no votes weren't positively correlated with the results of more fine-grained voting systems. The counter-example you're suggesting (in which 100% of voters think a movie is merely above average, while no one thinks it's great) seems pretty implausible.