On 7/16/25 8:53 AM, glen wrote:
> I had intended to only address Dave's assertion "trapped within a narrowly defined model". But I'll try to tackle 2 objections at the same time. Again, my target is this "everything's a metaphor" bullshit.
> 
> "Familiar" is a problematic term, here. Both Dave and Steve invoke the "definite" (ala Feferman's "what is definite"). When we use formal, schematic systems to translate a method from one domain to another, it's fine to call that "metaphor" at a cocktail party. But it's just not. Unbound/a-semantic terms are not metaphorical terms.
> 
> Now, Steve's right to separate (A) from (B) because "explaining" is different from translation, at least in the naive science/knowledge sense. (In the less bound/grounded statistics sense, they're closer to the same concept. But it seems Steve means the science/knowledge sense.) And when we explain things this way (by allowing some flex and slop in some of the terms of the model so someone from another domain can do the mapping themselves), we're relying on the audience to have a bushy *context* so they can/could bind all the terms as concretely (definitely) as we've done in the source domain. If the 2 contexts (person modeling in the source domain & person modeling in the target domain) aren't equivalently rich, then "explanation" fails.
> 
> And this is where Dave's wrong about multiscale modeling. The context at the large scale can be wildly different from the context at the meso- or micro-scales, similar between meso- and micro-scales. It hinges on whatever is meant by "narrow", of course. But multi-modal modeling not only exists, but is fairly common. There are even toolkits for doing it without giving it too much thought. All that's needed is to define (or even loosely describe) couplings between the modes so that they can sync up in time and space. Within the components, anything goes.

