While I'm hesitant to even jump in this thing, don't paint all dynos with the same paint, either.
The Mustang we use is not just weighted drums, it has complex processes that apply resistance to the rollers to emulate more "real-world" conditions. Sure, airflow is always contestable, but when the weight of the car is accounted for, the gearing is accounted for, the wind resistance and rolling resistance of the base car is accounted for (I concede that it's the base car we're using for specs) to some extent, by the HP @ 50 modifier; this value is vehicle-specific, and adds to the reality of the simulation by accounting for how mucht he car wants to decelerate at speed due to wind resistance, rolling resistance frictional losses through the gearbox if they're unusally high, etc.
We can run quarter-miles on the dyno, from a standing start, we can simulate hills and other high-load conditions, etc.
Anyways, stepping back out.
Nate was tuning with ECUtek long before we got a dyno. We got a dyno so we had repeatable results, much safer and more effective tuning, and less 3rd-gear romps behind Sunrise
and I'm done.
Scott, I don't think your analogy is tortured, I think it's a really good comparison. I'm not making value judgments here, just trying to add a bit of info.