Noah Smith holds that there are "four levels of science," with "history" being the lowest level, giving the least confidence in one's understanding of the world. The ultimate level of confidence would be offered by "lab experiments."
But, of course, a report on an experiment is a piece of historical evidence, and thus knowledge about a series of experiments is historical knowledge. Therefore, if the above were true, the paradoxical result would be this: we would be more certain of what a series of experiments proved, than we would be of what happened in the experiments! Or to put it concretely, we would be very confident that a series of experiments disproved, say, the existence of the aether, but only have a vague understanding of what went on in any individual experiment.
And that, of course, is logical nonsense: abstractions from individual events will always give us a weaker understanding than we had of the concrete events in the first place. One can't strengthen a weak foundation by piling ever more floors atop it!