Least Concern Is a Default, Not a Finding
"Least Concern" sounds like a conclusion. Somebody looked at the population trend, the range, the threats, and decided: this one's fine. For a lot of species carrying that label, nobody looked at anything.
The IUCN Red List has seven main categories, running from Critically Endangered down through Vulnerable and Near Threatened to Least Concern, with Data Deficient and Not Evaluated sitting off to the side as admissions of ignorance. The system is built to distinguish "we checked and it's fine" from "we don't know." But in practice, a huge number of assessments never clear the bar for genuine data collection before landing on Least Concern anyway. A species gets a wide-looking range on a map, no documented population crash, no red flags in the literature that exists, and it drops into the safe bucket by elimination rather than by evidence.
Take a fish that lives at depth, surfaces in trawl surveys a handful of times a decade, and has never had a dedicated population study run on it. There's no data suggesting decline. There's also no data suggesting stability, because there's barely any data. Data Deficient would be the honest category, but Data Deficient carries its own cost: it reads as a shrug, it doesn't count toward conservation planning targets, and assessors under pressure to produce a complete-looking list have an incentive to round down to something that looks like an answer. Least Concern is the box that absorbs the uncertainty and dresses it as a finding.
The word choice is what does the damage. "Least Concern" isn't a neutral label like "Category 4" or "Grade C." It's a phrase that asserts direction: somebody was concerned, checked, and concluded there was little to worry about. That's exactly what happens for the species with real population data behind the label. For the species with none, the phrase performs the same reassurance while describing a completely different epistemic event, absence of alarm standing in for presence of evidence.
This isn't a quirk specific to conservation taxonomy. Any system that offers a clean-sounding default state for the unremarkable case will collapse two different things into it: the case that was checked and came back fine, and the case that was never checked at all. A security scan that reports "no known vulnerabilities" means one thing when it ran a full audit against a current CVE database, and something else entirely when the scanner simply didn't have signatures for the software in question. Both produce the identical green checkmark. A background check that comes back "no record found" means one thing when the search actually queried the relevant jurisdictions, and something else when the request silently failed to cover the county where the person actually lived for a decade. The report reads the same either way, because the report only has one field for the outcome and no field for how hard anyone looked.
Once you notice the pattern, it shows up anywhere a process has to output a status but doesn't have to output its own confidence about that status. The label absorbs both the checked-and-clean case and the never-checked case into one bucket, and readers, downstream systems, funding committees, anyone consuming the label rather than the process behind it, treat every instance of that bucket as if it came from the strong case. Nobody re-derives which kind of "fine" they're looking at, because the whole point of a category system is that you're not supposed to have to.
Call it default innocence: a status meaning "not shown to be a problem" gets read as "shown not to be a problem," and the label gives you no way to tell which one produced it without going back to the underlying work. The two states aren't adjacent risks, they're different kinds of claim entirely, one about the world and one about the absence of an investigation into the world. Collapsing them into a single reassuring word doesn't just lose information. It actively converts an unanswered question into an answer, and hands out the answer for free to anyone who never asks how it was reached.
The fix isn't complicated in principle: report confidence and finding as two separate fields, always. It's just that a category system exists partly to let people stop asking the second question, and default innocence is what fills the space where that question used to be.