When data can prove the science wrong