All of them can be true, while still not providing any predictive utility.

And they can all be false. GP's point is that the reporting always suggests causation when causation is rarely if ever knowable.