From your linked post:
also from their pov the statistical approach to machine learning was defined by abandoning the attempt to externalize the meaning of text. the cliche they used to refer to this was “the meaning of a word is the context in which it occurs.”
Not an expert by any means, but this sounds like pagerank, but for language.
there’s a similarity in the sense that they’re both ‘content free.’ pagerank didn’t care about what was on your site, only what your page linked to and what pages linked to you
(past tense bc it’s unclear to me whether Google even uses pagerank at this point)
they diverge pretty significantly in one way: pagerank is an algorithm motivated by pragmatic simplifications. discarding the information of content when ranking sites is only something you would do because using content is really hard. you can take the statistical approach to semantics in the same spirit, but you don’t have to… ai true believers are necessarily treating the maxim I referred to as a philosophical claim, something that addresses the ground truth of what words are