This article analyzes the value of splitting. To quote XKCD, “really we’re both just categorization pendants” (https://xkcd.com/2518/). I feel both lumping and splitting have tremendous value, as well as scientific relevance (as that is touted in the article). Perhaps lumping is easier because brains pattern match well; but “enhance, enhance, enhance...” on a single pixel is just a hallucination (a single enhance using multiple video frames as samples can be somewhat meaningful, by which point the true signal is exhausted). More information is required, which itself is pattern matched, and splitting is a natural consequence of having more to hang in the tree data structure.
Splitting is good. So is lumping. I’m not seeing a strong case made here other than IBM’s “think”.
I might instead advance the claim that producing a minimal sufficient knowledge tree (or, a map of the right resolution, except that it has variable scale to detail interesting/relevant parts better, so the extended analogy mis-maps (heh) a bit) for the questions of interest is a better goal. This likely requires having the data available to make those splits (or knowing how and being able to go get it if time is available), but that doesn’t change the goal of producing a minimally sufficient map to minimize cognitive burden.
But—perhaps the point is “the map is not the territory” (lump more) is considered well-enough known and this is intended to push back against too much lumping, “the fallacies of compression”. Still I think I prefer an integrated view.
Action and creation is usually deliberate, derived from one or more purposes. Sometimes a purpose is specifically to elicit a desired mental framing (not strictly reaction, rather internal state) in others. Two people can take the same actions for different reasons—a face punch with the purpose to disable someone differs from an identical face punch to create a mental framing such as an emotion of fear—in such cases the purpose often needs to be overtly communicated to clarify the action was done for different purposes than might be normally inferred. We have a term for actions and creations with a strong overt element of eliciting a desired mental framing. Although it may often seem useless, there is nevertheless social value in creating those experiences, demonstrated by the pay that is made for ordinary examples such as entertainment experiences and products which are functional but also elicit mental framings. Many who create such artifacts or performances do it as much for the desired mental state they wish to experience, as for the hope that others enjoy the same and pay them.
ChatGPT was able to infer what I was talking about (“Playing taboo: what is the concept I am referring to here?”), so it’s at least comprehensible, though it could likely be much more terse. (With appreciation to the LTUE keynote speaker who set an ordinary water bottle on a table in an ordinary way and declared it an instance of this concept, then made the point that by so declaring it was no longer possible to claim it not such an instance, but only to decide if it was a “good” or “bad” instance of the concept).
The point I wish to make by this is there are no words that can’t be defined.