A dense, but interesting essay by Jaron Lanier on the brittleness of software and what it means for its economics. Most interesting for me is his take on comparing computer and natural languages: The degree to which human, or "natural" language is unlike computer code cannot be overemphasized. Language can only be understood by the means of interpretation, so ambiguity is central to its character, and is properly understood as a strength rather than a weakness. Perfect precision would rob language of its robustness and potential for adaptation. Human language is not a phenomenon which is well understood by either science or philosophy, and it has not been reproduced by technologies. Well, we are working on the later. He brings up the interesting question though on how the inherent imperfectness and ambiguity of language can be part of the solution rather than part of the problem in NLP applications. How can we leverage these properties to build robust NLP applications?