[jeroenhd]

A large amount of tests includes rendering text and basic elements correctly, which is an incredibly difficult problem. Getting JS to render right is one thing, but preventing bugs like "Google Maps works but completely breaks when a business has õ in its name" requires a lot of seemingly useless tests to pass.

Fixing a few rendering issues could fix all of the tests that depend on correct rendering but break, so I think the rate at which tests are fixed makes a lot of sense.

https://wpt.fyi/results shows that even the big players have room for improvement, but also has a nice breakdown of all the different kinds of tests that make up the score.