Another reason is encoding confusion. If the text is not ascii, then you somehow should guess its encoding. For compatibility reason a lot of tools and editors

Another reason is encoding confusion. If the text is not ascii, then you somehow should guess its encoding. For compatibility reason a lot of tools and editors may default to an old codepage instead of utf-8 (they may respect utf-8 if you have a bom), add to this legacy code already written in who knows what encoding. Is it a windows codepage? DOS codepage? Linux? ISO-8859? If you guess encoding wrong, you can corrupt text.