have a read of some of Jan Heine’s articles on the subject, there are also various (mostly German?) tests that have been published over the years. The only caveat I’ll give you is that ALL of the published material is lacking in some ways. Jans stuff is very well thoguht out and tries to focus on real-world testing, but leaves his results open to interpretation and difficulty in repeatability and some questions over variable control. The Germans have mostly resorted to drum testing which although very controlled and repeatable, is lacking in the real world aspect and doesn’t take the human on top into account at all.
But in a nutshell, the friction/drag of the contact patch is only one aspect of it, carcass (and tread) deformation, bump absorption and vibration all play a big part. This is why the casing is just as if not more important in some cases as the rubber compound and tyre size.
The circumstances and specifics of the tyres dictates the balance between them but ‘rolling resistance’ is a result of the entire systems and it’s interactions, not just how much rubber is in contact.
This is also why the qualifier ‘when compared like-for-like to a tyre of similar construction and pressure adjusted to suit’ should always be included when saying wide tyres are faster.
ie: a 35mm Marathon Plus is NOT faster than a 25mm GP4s, the two tyres are vastly different in construction and design. But a 25mm GP4s is likely faster than a 23mm GP4s.
Also, remember rolling resistance =/= speed, they are related but you also have to take into account the human on top, and the impact on shaking and bouncing that human around over long distances which mean that ‘speed’ is also not simply about how fast the tyres are, especially once distance and time on the bike come into play, same argument with speed vs comfort, ie: a TT bike and aggressive aero position is ‘faster’ but if you can only maintain it for a few hours at a time before you’re crippled and need to stop/rest then it may not be ‘faster’ over the long haul.