While the analysis is fine, the conclusion is almost certainly false. At Google's scale, nearly any iteration made on core products is backed by an immense understanding of their end users and a near unlimited supply of user data.
There can be a million things wrong with user testing, and while I am certain they know more about it than me, I think it's a safe bet to say they're not perfect either and may make mistakes in such a complex topic.
(Small example of what I mean: many people like the new style because it's prettier, but when I use OSM I get to places a lot faster than people using Google Maps, so they might be asking the wrong question even if they did user testing with a million people. But that's just one example of everything that could go wrong.)
> At Google's scale, nearly any iteration made on core products is backed by an immense understanding of their end users and a near unlimited supply of user data.