it is empirical that means that you should be able to re-produce the results of a thing or assertion by following the details in a paper.
The public might be able to do it themselves. But the point is, its not about who says what, its about can it be reproduced.
scientist "A" says that the sky is blue because of "x". devises an experiment to prove that. writes up the experiment, publishes it, asserts that the sky is blue because of x, and that the experiment proves this.
Scientist "B" says it bollocks, reproduces the experiment, but also extends the experiment to show that the data also says that the sky is green. Paper is published with data and method.
The process repeats until a consensus is reached where everyone can reproduce the data, and no one can disprove the hypothesis that the sky is blue because of x.
None of that requires asserting bollocks on a chat show. Sure science outreach is great, but its not _really_ part of the method.
