We have found a set of (relatively simple) rules that match (a large part of) reality so well that no matter what we do, we don't seem to be able to get any results that would indicate the rules are wrong or even slightly inaccurate. However, we are almost certain they are wrong, or at least incomplete – but given how well they model reality it would be astonishing if it turned out that we're on a totally wrong track and the actual rules are completely different.