Humans capable of the first part regularly make the mistake of assuming that their decipherment is correct even if it's not internally consistent and fails to adequately explain most of the data.