I hear ya. I'd just like to think that, theoretically, there's some consistency in the ratings just because of 1) a shared and well-documented methodology and 2) sometimes it's the same people reviewing these major events-- i.e., the super-elite survey dudes who really specialize in this.
But, yeah, with 'canes we are lucky. Sure, the reanalysis estimates for early-20th-century cyclones probably have plenty of error-- but at least we know the analysis techniques are being applied consistently across samples by the same committee of people.