Category Creep

TL;DR: I've noticed a tendency to expand definitions too broadly, which dilutes their clarity and utility. I make the case that effective language, in my eyes, requires precise boundaries that slice concepts into hierarchical, mutually exclusive, and collectively exhaustive categories, preserving clear distinctions. By respecting these boundaries, we avoid what I call “category creep” and enhance language's ability to facilitate clear communication and thought.

At dinner last night, our conversation about religion unfolded into claims that in some sense capitalism is a religion, humanism is a religion, and so on—raising the question, what actually is a religion? Through the conversation, the group expanded into wider and wider interpretations of what could qualify as a religion: sports fandoms, the cult of Elon Musk, companies, and so on. By the end, it felt like we were calling everything a religion and had diluted the word into meaninglessness. I've seen this type of cognitive failure mode before, and I'd like a precise term for it.

Take, for instance, the “Church of the Flying Spaghetti Monster”—which, though in the eyes of the tax code might be legally recognized as a religion, is generally understood to be not actually a “real religion” but rather a humorous, subversive riff on true religion. This suggests that it lies outside the conceptual boundary of “religion” in some crucial way, even if certain institutions recognize it as such.

Whenever debates arise around the definitions of words, I'm reminded of the classic question, “If a tree falls in a forest, does it make a sound?” People get remarkably heated about this, despite often actually agreeing about the true underlying reality. The word “sound” here is overloaded: we agree that the tree creates a sound qua “acoustic vibration in the air” but disagree on whether it produces a sound qua “auditory experience” if no one hears it. The disagreement isn't over base reality but over the shape and boundary of the word sound within concept-space.

Resolving the Church of the Flying Spaghetti Monster ambiguity, if we separate the concept religion into the more specific categories of “a U.S. tax code recognized legal entity” and “an authentic belief in and worship of a higher power”, nearly everyone would agree that the CoFSM is the former but not the latter, putting the debate to rest. Notably, this might not just be a case of “sign confusion” but potentially rather of “symbolic theft” wherein terms are intentionally appropriated, in this case either to delegitimize real religion or for financial gain.

To illustrate why precise boundaries are essential, consider the value of dividing a space into mutually exclusive yet collectively exhaustive subspaces, like we did in childhood math problems. This type of structure—where each possibility is accounted for exactly once, without overlap—is useful across domains. For example, a zoologist benefits from a clear taxonomy, organizing the animal kingdom into distinct, hierarchical categories like order, family, genus, and species. Language, when used effectively, functions in much the same way by organizing concept-space into well-defined, non-overlapping boundaries, preserving clarity and avoiding category creep.

I've noticed a cognitive failure mode where someone, swinging too freely through analogies, overly word-associates themselves into incorrect conclusions. This became clear to me while listening to a recent debate between Richard Dawkins and Jordan Peterson, wherein Dawkins accused Peterson of being “drunk on signs”. The common structure of this failure mode goes something like this: Thing 1 is like thing 2, thing 2 shares qualities with thing 3, and so the person concludes that thing 1 must also share properties with thing 3—or worse, that thing 1 is thing 3. In Peterson's case, he claimed that because a dragon is (the imagistic equivalent of) a predator, and because a lion is a (type of) predator, that therefore a lion is a dragon. The disagreement came from Peterson's interpretation of “is” as a loose conceptual equivalence, baffling Dawkins because a lion is not actually a dragon. The issue is that “is” here is overloaded, creating ambiguity by blurring metaphor with literal truth. Under Peterson's fuzzy analogistic logic, a lion is like a type of dragon in a figurative sense, but to Dawkins a lion isn't literally the same as a dragon. In this case, in my eyes, Peterson suffers from “category creep” as his overly-metaphorical reasoning flattens important distinctions, reducing “lion” and “dragon” to interchangeable terms. Words, after all, are conceptual boundaries—and when the boundaries get stretched too far, their overlaps reduce their usefulness as categorical distinctions. Though boundaries always have both an inside and an outside, when we think of words as conceptual boundaries, we often focus far more on describing what lies inside the boundary than what lies outside. While lions and dragons have similarities, lions do not fly, lions do not breathe fire, and lions are not mythological. Sometimes, it's just as helpful—if not essential—to clearly define what a word is not.

Overextending categories—like labeling sports fandoms as religions or lions as dragons—dilutes the specificity and utility of language. This in my eyes is a common and under-discussed cognitive failure mode. In the spirit of this discussion, a label we can assign to this sort of semantic overreach could be “category creep”. Ultimately, we as a culture continue to refine linguistic boundaries, and the ideal of pursuing hierarchical MECE categories through precise definitions would help facilitate more accurate communication and thought.