This is commonly called the "gish gallop" - it's much easier to generate falsehood than refute it.

Worse, because of the way confidence intervals work, if you generate enough bad hypotheses and run experiments, eventually you find one where it appears to be true!