Of course it's not the right tool for every job. For large payloads of tabular data, the verbosity adds a lot of overhead with serialisation, transmission, and parsing and then you're better off with something like Message Pack or Protocol Buffers.

YAML is great for lots of cases. Compact syntax, built-in support for heredocs, no unnecessary quoting, value references, clear structure. Stuff like "the Norway problem" is kind of a facepalm but avoidable.