Which rating is better, mathematically speaking?

It's cool to see the math/statistical analysis behind this, as it's something I've always done just intiutively I guess.

I will always have more trust for a 93% review with 1,000's of reviews, versus a product with a 100% review with only ~20 reviews. Much too easy to manipulate and/or pay for reviews to trust a product with such small sample size of feedback.