I don't know. Barrkel's example seems better to me because there's an obvious practical need for it. The trouble with most of the work on parsing I see is that it's just not hard to hand-write a parser. I used to avoid doing so, and then I wrote one and was surprised: once you factor in error-handling and whatever other meaningful output your system may need from its parser (e.g. text extents for ASTs), the overall complexity of a hand-written one can easily be less than one made with tools at a supposedly higher level of abstraction. And that's not counting the time it takes to learn the tool (which is not trivial, as they don't always have good debugging support) or the complexity cost of having the tool in one's stack (also not trivial, since they typically have their own languages, complicate the build process, and so on). This experience led me to mentally discount the whole field, hence my perhaps overly dismissive comment.