Open ajschumacher opened 4 years ago
compare:
symbolic reduction fallacy and nominal fallacy (mentioned in #239 book, page 100)
the same way that every useful word is incompletely defined by its definition. the problem with datasets goes deeper, which is that they are essentially definitions without words
Kyle McDonald, https://twitter.com/kcimc/status/1315049704665669632
"What’s a sandwich" as example of (assumed) definitions not being as clear as you think
as a way of introducing modeling?
There's more than one way to do it (C sharp vs. D flat, synonyms, phrasing, denotation vs. connotation)
Multiple referents: "There's a dog" - is that about the world, or about what you're perceiving? Can be even more indirect: the world, the video doorbell picture, what you're perceiving (etc.)
Clarification that probably isn't necessary: Not talking about models of language here, as in descriptivist vs. prescriptivist. https://fs.blog/2020/10/descriptions-arent-prescriptions/ etc.