stumbled accross a claim today in the comments from this article.
http://www.denverpost.com/commented/ci_11936454?source=commented-business
he links to:
http://www.rhapsodyingreen.com/rhapsody_in_green/files/optimal_ethanol_blend_level_study.pdf
Guy claims: "The University of Minnesota along with U of North Dakota did a study not long ago of highe ethanol blends in normal (non-FFV ) late model vehicles and found the mileage actually went up at a "sweet-spot" between 25-30% ethanol. This is with no modification to the vehicles. So imagine what happens when you include turbo-charging, or run in an engine with higher compression ratios( greater than 10 to 1) making full use of ethanols' higher octane numbers. There's also a longer term study( 2 years) done by the UofM that shows no significant problems with using an E20 blend in most non-FFV vehicles, even in cold climates ."
dyno testing doesn't replicate real world, and they're skewing the mpg figures somewhere in that measured/calculated formula they have. Keep dreaming guys...