www.lesswrong.com/posts/AtEJEQzBnGE6uKJA5/science-like-a-chef

This sentence is a factual overstatement about how science works. Better experiments can increase confidence, but they do not make scientific conclusions certain in the absolute sense.

The National Academies' report Reproducibility and Replicability in Science explains that "results are assessed probabilistically" and that the scientific discovery process is "unable to deliver absolute truth or certainty." The same report says scientific claims gain confidence over time through scrutiny, confirmation, and uncertainty reporting.

So the error is not that experimentation is unimportant—good experiments are essential—but that they are said to yield certainty. In modern scientific methodology, uncertainty is intrinsic and must be characterized rather than eliminated.