Because they fed the software different data and the result ended up similar and then introducing data that had nothing to do with the subject.