But you missed some of my points (which I didn't think were silly at all)
1. Averaging out results from different machines means you don't end up proving that ONE moped is more/less efficient than ONE motorcycle - which is all a head to head race would prove. Unless you had several mopeds and several motorcycles.
2. Motorcycles and mopeds have different speeds at which they reach maximum efficiency, and things like idle gas usage would affect them differently. So a lower speed path with lots of stops might favour one machine, whereas a different path would favour another machine - so you'd need to do a few different paths to get any real result (again with different motorcycles and mopeds) This would be the preferred thing I admit, but not likely to happen... whereas just averaging out peoples everyday riding results isn't terribly hard to arrange.
One machine against one other machine will only prove anything for that particular brand compared to another particular brand.... not give any real results for the general machine type. It'd be like having one random person sample a drug from the street and one other person samples another drug... and whomever gets cured first proves the drug works. Doesn't really give any real info.