"Writing a really general parser is a major but different undertaking, by far the hardest points being sensitivity to context and resolution of ambiguity"
About this Quote
“General parser” sounds like the sort of tidy, platonic object mathematicians love: one mechanism that can swallow any input and yield the right structure. Nelson punctures that fantasy with a programmer’s bruised realism. The work isn’t “major” because of raw complexity in the abstract; it’s major because language and meaning are booby-trapped by the world around them.
The phrasing “major but different undertaking” is doing quiet rhetorical work. It separates brute engineering (making something big) from the qualitatively messier task of making something socially and semantically aware. The “hardest points” he names are not speed, not memory, not clever algorithms for their own sake. They’re “sensitivity to context” and “resolution of ambiguity” - two problems that refuse to stay inside the sandbox of formal rules. Context means you don’t just parse a sentence; you parse a situation: prior statements, implied goals, genre conventions, even what a user probably meant. Ambiguity means the input is often legitimately multi-parseable, and correctness becomes a choice that has to be justified, not computed.
Subtext: the dream of a universal parser is a category error. Parsing isn’t only syntax; it’s interpretation, and interpretation drags in pragmatics, knowledge representation, and human expectation. Coming from Nelson (closely associated with interactive fiction), this reads like a field note from someone who’s watched “natural” language collide with brittle systems. The warning is elegant: the hardest parts are the ones that look least like math and most like people.
The phrasing “major but different undertaking” is doing quiet rhetorical work. It separates brute engineering (making something big) from the qualitatively messier task of making something socially and semantically aware. The “hardest points” he names are not speed, not memory, not clever algorithms for their own sake. They’re “sensitivity to context” and “resolution of ambiguity” - two problems that refuse to stay inside the sandbox of formal rules. Context means you don’t just parse a sentence; you parse a situation: prior statements, implied goals, genre conventions, even what a user probably meant. Ambiguity means the input is often legitimately multi-parseable, and correctness becomes a choice that has to be justified, not computed.
Subtext: the dream of a universal parser is a category error. Parsing isn’t only syntax; it’s interpretation, and interpretation drags in pragmatics, knowledge representation, and human expectation. Coming from Nelson (closely associated with interactive fiction), this reads like a field note from someone who’s watched “natural” language collide with brittle systems. The warning is elegant: the hardest parts are the ones that look least like math and most like people.
Quote Details
| Topic | Coding & Programming |
|---|
More Quotes by Graham
Add to List
