So are JSON and YAML. The point is whether you have a programmatic and structured handling of a document's content, or whether it's random, where every line and word can demand for a different parser.