Polygon gave Bayonetta 2 a 7.5. The Metacritic average was 91.

The math aint mathing. Bayonetta 2 closed with a Metacritic of 91. The aggregate of 80+ major outlets averaged to 9.1/10. Polygon's number? 7.5. That is a 1.6-point deviation on a 10-point scale, which in score-distribution terms is the kind of outlier you plot on a chart, circle with a red marker, and use as the example slide in a freshman statistics class titled 'this data point is doing something different than the others.'

Let's break the review down by word allocation. I counted. Of approximately 1,400 words, roughly 30% are dedicated to camera angles, costume framing, and the protagonist's silhouette. Combat mechanics get 22%. Boss design, the actual structural backbone of a character action title, clocks in at 6%. For a genre whose entire selling proposition is the mechanical depth of its combat encounters, that distribution is inverted. The review weighted cinematography over content by a factor of five. Five. The math is the math.

Look at the comparison set. PlatinumGames had shipped four character action titles before this one. Frame-pacing locked. Combo system layered. Witch Time mechanic intact. Every other major outlet landed between 8.5 and 10. Mode: 9. Median: 9. Mean: 9.1. Standard deviation across the top 30 reviewers: roughly 0.4. Polygon's 7.5 sits four standard deviations below the mean. In any other dataset that's a measurement error. In game criticism that's a Tuesday.

Here's the issue. You cannot take a 50-hour mechanical action game, allocate the bulk of your word count to a discussion of costume framing, and then arrive at a single number that's supposed to represent the whole work. The numerator and the denominator have to be talking about the same thing. They weren't. Polygon ran a content-criticism essay through a number generator and printed the receipt. That's not a review. That's grading a calculus exam on the handwriting.