Their "smartness" is very relevant. It's the reason why someone in this experiment would not see the difference and the reason why we shouldn't use this experiment to dismiss differences in fidelity between platforms. Because people as a whole as "dumb" and not a good measuring stick for what is noticeably different or not. Which I remember now is what I was actually trying to argue. But JimmyD was a childish distraction. Reminds me why I don't like arguing in thread like these. But yeah, it's all the same. We aren't disagreeing.
I don't think the point has anything to do with whether or not the games themselves have "noticable" differences. It's about how expectation affects perception. When I watch the vid , I see that the first bit was more about identifying the people who can identify the differences (and kinda what percentage of that group could- which is the part where sample size is relevant in this portion's vacuum,), because the trick test would be useless on the people who couldn't tell.
The findings there have some relevancy as people who know enough to tell differences continued to see differences when there where none. The second half is the more important, imo. Without establishing those who could differentiate the versions, the last bit would fall flat.