It's important to distinguish the ordinary observations, even those that are mathematically described (e.g. ancient astronomy) from an approach that rigorously reality-tests everything (i.e. science).