Hmmm...I'll have to look at this later since many of the linked graphs are blocked here at work. But the quick perusal of the paper itself leaves one overarching question: why not?
His case that ethanol supporters exaggerate the substitution effect is pretty solid, for sure.
His claims that it's had no effect whatsoever are much less so, and he doesn't profer any explanations for why that'd be the case. My educated guess would be a demand-pull situation where ethanol's cheaper price due to subsidies creates a higher demand that negates the offset. In essence, ethanol's "benefits" cause people to want to use it more, thus eliminating any offset gains. See also, e.g., lighting (more efficient lightbulbs individually use less electricity, but people then use them more or use a brighter bulb - rather than receiving the same amount of light at a lower energy cost, they use the same amount of energy for more light). But that's just a guess, and given the relatively inelastic nature of gas, doesn't seem like it'd be enough to completely negate the substitution effect entirely.
The obverse of this theory that ethanol doesn't replace any oil is that if we took away all ethanol tomorrow, we wouldn't use a single drop more of oil than we already use to make up for ethanol's disappearance. Which may or may not be a true statement (since we can't exactly test it), though anecdotedly I find it hard to believe that would be the case -- I currently fill up with the 10% blend since it's the cheapest; even after accounting for any mileage gains from going to pure gasoline, still seems like I'd be using some amount of oil more than I currently use, even if it wouldn't be 10% more.
So, unless there's a magic bullet waiting for me to see in one of the graphs, I'm skeptical to the extent he argues there's no substitution effect at all. I fully agree the effect is exaggerated by proponents, though.