> Why can't a parser be written?

Because the existing parser is written in truly emacs style: no formal grammar, just a lisp code with a regexp at each turn. Theoretically speaking it doesn't forbid you from writing a parser, but in practice there are no full-blown parsers of org-mode except the reference one.

For many software businesses, licensing is an issue. The spec is GFDL with GPL code samples, a non-cleanroom translation of the elisp parser would (likely) be GPL (or at least arguably enough so to keep lawyers busy), so going and doing some other roughly equivalent markup language instead avoids the copyleft requirements.

So, yes, “too much trouble”, much of it nontechnical.