On Heidegger, and the rook that was always, secretly, a bishop

Imagine, if you will, that you are playing chess with a grandmaster. Your opponent fingers his rook, slides it diagonally across the board, captures your queen and lifts the piece away. Nonplussed, you question the tactic. He defends it with vehemence.

Full story on Substack. Video and podcast summaries below.

On the limitations of NotebookLM.

For most of my posts, I also share a summary processed by Google’s NotebookLM. Most of these summaries are decent enough – some are even excellent inasmuch as they shed a new light on the idea. Other times, they miss the point, are well off the mark, or inject concepts or translations, not only unintended, but markedly opposed to the point being made.

My recent Chess with Heidegger post is illustrative. Both the podcast and the video make errors – and I don’t mean the visual ones that occur as well.

As a page footer, I mention that I use LLMs as part of my workflow. This may be anywhere from 0–99%. The podcast asserted that I used an LLM as a significant part of this project. This is incorrect. Specifically, I didn’t feed Heidegger’s text into an LLM and ask it to digest and regurgitate it, parsing out some unlocked wisdom. Obviously, I used various LLMs for supporting content.

The video misrepresented my point about Wittgenstein, directionally off. He expressly does not support the specificity of language. My reference to him rolling over in his grave was triggered by the notion that language is determined by use – and language games.