In this TEDx talk, Physicist and Nobel Laureate George Smoot claims (around 19:00) that inconsistency in measurements of the universe imply either that our universe is not real, that is, that we live in a simulated reality, or that or ability to measure the universe is flawed.
This seems to imply that Smoot is making the presumption that the real universe must be consistent. Why should we expect that measurements in a real universe must be consistent?
This seems to imply that Smoot is making the presumption that the real universe must be consistent.
The underlying assumption that Smoot is making here is that there is a single Theory of Everything that explains the whole universe and unifies all science. Alternatively this view can be expressed as the idea that all sciences will ultimately be reduced to fundamental physics (in the way that chemistry was reduced to quantum mechanics or that genetics was reduced to molecular biology thanks to the discovery of DNA).
So then to answer the question of whether:
Why should we expect that measurements in a real universe must be consistent?
Really comes down to arguments for or against a unified Theory of Everything. If there is a ToE explaining the whole universe, then we should expect all of our measurements to be consistent. If there is no ToE, then there is no reasons why different scientific theories - having different domains of application - need to be consistent with each other. There is a lot to debate in this question, but here are a couple of considerations:
A really interesting book I read on the topic is "A Tear at the Edge of Creation", by physicist Marcelo Gleiser. In it he argues that recent discoveries (including the frequent inconsistencies in measurement that Smoot talks about) indicate that a Theory of Everything is unlikely. He also argues that there is no reason for us to expect such a theory in the first place, and that the only reason people have expected such a theory is a holdover from monotheistic religious beliefs. He argues that the idea of single God who created the universe is so deeply ingrained in Western culture, that even those who don't believe in God still work with the assumption that a single set of fundamental laws must govern the universe, even though empirical evidence points in the other direction.
I don't think George Smoot is arguing that "the real Universe must be consistent".
He's certainly not arguing in a Theory of Everything sense; if he was talking about theory, he would have said it differently.
He's discussing whether measurements must (if properly done) be consistent: If you measure some single quantity multiple times and/or in multiple ways, all done properly, "consistency" means you get the same result each time.
Physicists tend to believe that there really is a there there: That the Universe is in some sense "real" and exists whether we understand it or not, that it operates/behaves/acts on it's own. If the Universe really is a thing, then that thing has to be consistent with itself: there just the one of it, and our measurements are looking at the same thing over and over.
That sounds so obvious that it's hard to say clearly, so let's consider a different case: They Universe is just some values stored in a Really Big Computer that's doing a simulation. Well, in that case, maybe there isn't a single value to a measurement, because the simulation isn't exact, or only gets more exact as time goes on, or values get updated as the simulation works along. Then maybe measuring the same thing multiple times will get you different values.
