You can have consistent results on a dyno. There was a recent discussion about this, and the consensus was that you want the results to be in the same format, like Dynojet SAE, you want the conditions (weather), and you want some datalogging through the OBD port.
This could all easily be done. The problem is expense. It would be expensive to buy all the tunes and to pay for the dyno time. Unless the tuners wanted to donate their tunes and a shop wanted to donate the dyno time.
Since this is unlikely to happen, I will continue to look at the dyno results since they are the most accurate measurement of actual gains available. Eight different people with eight different tunes can all say they love their tune, but that does little for me. Even if 100 people buy one tune and say they like it and 10 people buy another and say they like that one, I would remain unconvinced that the tune 100 people bought was any better than the tune 10 people bought. More popular, yes.