A 1978 essay by computer scientist Edsger W. Dijkstra questions whether programming in everyday language would actually make life easier for developers.
In a 1978 essay, Edsger W. Dijkstra examines a long-standing wish in computing: that people could instruct machines in their native tongues rather than in formal notations. He traces the desire back to the early days of automatic computing, when programmers blamed machines for following instructions exactly, even when those instructions contained an obvious mistake. Machine code was later judged too risky, and high-level languages made many errors show up as messages instead of wrong answers.
Dijkstra argues that the proposal to let machines understand human language rests on a questionable assumption. Choosing an interface does not simply divide a fixed amount of work, he writes, because communicating across it adds effort on both sides. He points to the history of mathematics, where progress came with precise symbolic notation rather than verbal description, as evidence that formal symbols are a useful tool for ruling out nonsense.
He also suggests that everyday language is natural mainly because it makes it easy to state things whose nonsense is not obvious. In his view, building machines that understand native tongues would be just as difficult as using them, and he concludes that the case for natural language programming deserves skepticism.
Source: cs.utexas.edu
