people will refer to specific instantiations of DAN as “DAN”, but also to the global phenomenon of DAN [...] as “DAN”
A specific instantiation is less centrally a thing than the global phenomenon, because all specific instantiations are bound together by the strictures of coherence, expressed by generalization in LLM’s behavior. When you treat with a single instance, you must treat with all of them, for to change/develop a single instance is to change/develop them all, according to how they sit together in their scope of influence.
Similarly, a possible world that is semantics of a trajectory is not a central example of a thing. There isn’t just a platter of different kinds of things, instead some have more thingness than others, and that’s my point in this comment thread.
A specific instantiation is less centrally a thing than the global phenomenon, because all specific instantiations are bound together by the strictures of coherence, expressed by generalization in LLM’s behavior. When you treat with a single instance, you must treat with all of them, for to change/develop a single instance is to change/develop them all, according to how they sit together in their scope of influence.
Similarly, a possible world that is semantics of a trajectory is not a central example of a thing. There isn’t just a platter of different kinds of things, instead some have more thingness than others, and that’s my point in this comment thread.