I think you are not reading my post fully, I said, yes there could be a difference of 3% say in a lab but out on the road you put your foot down for 2 minutes and that difference is gone. Is it effectively day and night in reality, not at all.
I guess being IT then to be the same things have to have an identical code frame or the like. Which is fine but being a statistician you dont look for things to be identical, you try to measure a difference, which is different to measuring if something is identical. In the real world, it is almost impossible to find two things you can test alongside which makes the test identical, which is why you have statistical difference testing. Simply being 1% different (be it chemically or otherwise) is not actually different given certain variability from a statisticians point of view. Nice clash of careers as understanding the same concept me thinks ...
There is an interesting article done by a uni in Australia on its cyclists. They tested them exercising until they lost 5% water content in their bodies, then rehydrated them with nothing, water and sports hydration drinks. Clearly the ones that took on water did better than the ones who started exercising again without anything. Actually the sports drinkers did better than the water drinkers - advertising you often see on TV now. Cut and dry you might think. They then repeated the test, this time the three groups took on water, sports drinks or nothing at all intravenously so they wouldn't know which group they were in. All three groups independently of each other performed equally well.
So yes, you can try cars out on the road, you can draw conclusions for yourself, but be aware that you yourself are a bias and it can be a big one.