I didn't say a trainer couldn't be engineered to have power on it match power on the road at the same speed. In fact most trainers probably have at least one speed where that accidentally happens. I just said there is no reason to assume it. Think about a trainer with selective resistance. Clearly you can ride that trainer at the same speed but several (five?) different power outputs. A high end, really well engineered trainer could perhaps map the speed vs. power curve of road riding through tuning of the resistance mechanism or maybe even using a composite mechanism (magnet + wind + fluid). With three contributions to resistance it is likely you could develop coefficients for each component that would closely trace the on-road experience. But wait a minute. That would be at only one effective headwind speed, right? Different headwind, different speed/power curve. Oh and at only one slope too. So you see, it really is hard. Unless the whole thing was programmable and you could put in your wind and slope particulars and get a custom resistance. That would be cool, but not likely any time soon. Is it?