It’s probably more of a case of being unknown. In the extremest of examples, a great work could have been seen once and the guy who reviewed it just give it low ratings. The work therefore is low rated, despite being a great work.
A remedy to this is to use something like Bayesian average...