Here’s a single concrete thing he does that drives me nuts. I wonder if it may be a part of what is setting you off, too?
He overuses the term “unifying.” He uses it three times an episode, to mean a different thing than I would usually mean by it. I really wish he’d cut it out.
I usually see “unifying” as signifying that there is an overarching model that takes some of the complexity of several models, and collapses them down together. Something that reduces “special casing.”
He almost never means that. It’s always adding more, or tying together, or connecting bits without simplifying. It comes off to me like a string of broken promises.
In my notes, it means that I produce a ton of pre-emptive “Summary Here Headers” (for theory unifications that seem to never come), that I had to delete in the end. Because usually, there isn’t a deep shared root to summarize. When I come back to fill them in, all I find is a tangential binding that’s thin as a thread. Which is just not enough to cohesively summarize the next 3 things he talked about as if they were a single object.
I think his “big theory” is actually something more like… spoilers… which I wouldn’t have guessed at accurately from the first 2 episodes.
(I can’t get spoilers to work on markdown, ugh. Stop reading if you want to avoid them.)
Maybe “attention as a terrain,” or maybe something about aligning high-abstraction frames with embodied ones? The former feels basic to me at this point, but the later’s actually a pretty decent line of thought.
I can’t recall any specific examples of him using “Unifying” that way, but what you describe does ring familiar. I think he tends to use verbose language where unnecessary. I’d love to get the Paul-Graham-edited-for-simplicity version of these lectures.
Here’s a single concrete thing he does that drives me nuts. I wonder if it may be a part of what is setting you off, too?
He overuses the term “unifying.” He uses it three times an episode, to mean a different thing than I would usually mean by it. I really wish he’d cut it out.
I usually see “unifying” as signifying that there is an overarching model that takes some of the complexity of several models, and collapses them down together. Something that reduces “special casing.”
He almost never means that. It’s always adding more, or tying together, or connecting bits without simplifying. It comes off to me like a string of broken promises.
In my notes, it means that I produce a ton of pre-emptive “Summary Here Headers” (for theory unifications that seem to never come), that I had to delete in the end. Because usually, there isn’t a deep shared root to summarize. When I come back to fill them in, all I find is a tangential binding that’s thin as a thread. Which is just not enough to cohesively summarize the next 3 things he talked about as if they were a single object.
I think his “big theory” is actually something more like… spoilers… which I wouldn’t have guessed at accurately from the first 2 episodes.
(I can’t get spoilers to work on markdown, ugh. Stop reading if you want to avoid them.)
Maybe “attention as a terrain,” or maybe something about aligning high-abstraction frames with embodied ones? The former feels basic to me at this point, but the later’s actually a pretty decent line of thought.
I can’t recall any specific examples of him using “Unifying” that way, but what you describe does ring familiar. I think he tends to use verbose language where unnecessary. I’d love to get the Paul-Graham-edited-for-simplicity version of these lectures.