This person from a field where reductionist experiments and strong control over variables are possible, makes a sweeping statement about multiple fields where you controlling variables and performing experiments are among the leading research challenges
I was under the impression that something along these lines was already accepted from the perspective of information theory. I.e. a machine that could simulate the universe must at least be composed of as much information as the universe itself. Given the vastness and complexity of the universe, this would make it rather unlikely that the universe is simulated. Unless you want to view the universe itself as a machine that calculates it's own progression. But that is a bit of a semantic point.
Disclaimer: this is not my area of expertise and I probably got some terms or concepts wrong. I am basing this off of 'The information' by James Gleick