But isn't perception what everyone here argues in favour of because measured metrics using testing in a controlled environment on websites like DxO don't favour Canon?
I don't see many people arguing that here. DxOMark's 'measured metrics using testing in a controlled environment' are just fine (usually, although they have gotten bad copies of lenses, refused to admit it, and then later quietly re-done the measurements on another lens). They don't seem to favor any one brand.
But when they take those 'measured metrics using testing in a controlled environment', put them in a bias blender and whirl them around at several thousand rpm, then take the goo that pours out and feed it to scum-digesting bacteria, then take the bacteria excrement and bake it in an oversimplifying oven at several thousand degrees, then take the steaming pile of baked crap and call it a Score...that's the problem, right there.