Why aren't car MPG ratings be tested on real roads?

[gimp]

http://www.usatoday.com/story/money/cars/2014/06/12/ford-gas-mileage-lowered/10383307/

This happens over and over and over and over...

Just hop in the damn car, drive it for three thousand miles, let us know how much gas it used. It's not a difficult concept. No estimates, no formulas, just real road performance.

Yes, real road performance is not ideal; you have to brake for traffic jams and you have to wait too long at annoying red lights and you occasionally have to hit the throttle to be safe. You have to exceed speed limits and you have to drive up big hills. That's what the customer wants to know; nobody cares how well the car does on a treadmill.

Yes, the shortcut is much easier. So is just lying and making up numbers. A 7 mpg drop is a pretty big fucking deal, and if you're dropping 7 mpg or if you're fixing a rating twice you may as well just pull a number out of your ass, and be honest about it.

/rant

[matchewed]

Um, because in order to truly measure something you have to control for variables.

[Jamesqf]

Yes, and there's a lot of variation between drivers.  I've never had a car that I couldn't get significantly better than EPA mpg from.  I suspect Ford's only doing this to forestall complaints from people who don't get the rated mpg because they drive like idiots.  Or indeed, because (like certain neighbors of mine) they will let the car idle in their driveway for 15 minutes every morning before driving a few miles to work, or will sit idling in the long line of parents waiting to pick up their too-special-to-ride-the-school-bus little darlings.

[gimp]

I agree, matchewed, except the real world has uncontrollable variables. When we design our products, we take the variables into account... then we test them in the wild, using them as any normal person might use them, and find a bunch more issues due to variables we never considered.

When we figure out battery life, we start with a large complex calculation that requires several PhDs and a bunch of software. Then when we manufacture the device, we build a reference design and take it home and actually use it until the battery dies and time how long that took. I don't know what the various other OEMs and ODMs do but we make sure it performs up to the theoretical spec.

[matchewed]

I agree, matchewed, except the real world has uncontrollable variables. When we design our products, we take the variables into account... then we test them in the wild, using them as any normal person might use them, and find a bunch more issues due to variables we never considered.

When we figure out battery life, we start with a large complex calculation that requires several PhDs and a bunch of software. Then when we manufacture the device, we build a reference design and take it home and actually use it until the battery dies and time how long that took. I don't know what the various other OEMs and ODMs do but we make sure it performs up to the theoretical spec.

Did you watch the video? They do real testing as well. In fact real testing for one of the variables was how they determined there was an error.

I'm just not sure what your beef is on the MPG testing process. You seem to think it's inadequate because an error was made. If I fuck up using a ruler it doesn't mean the ruler is bad at measuring right? The measurement method is consistent and sound. That's what matters. If the method isn't followed or an error was made then the method in of itself is still sound.

[BlueMR2]

I've got no problem with the testing process.  Even on the older models (with the supposedly impossible to reach standards) I almost always get better MPG than the sticker says.  With how loose the standards are today a drunk monkey should be doing better than the sticker number.  If you're getting worse than that, perhaps hiring a drunk monkey as a driver would be appropriate.  :-)