Split for What
Two field guides for the same region can disagree about how many species of a given frog live in it, one saying four, another saying nine. The genetics haven't changed between printings. Nobody found a new gene function. Same animals, same DNA, and a factor of two in the species count. The instinct is to treat this as a research gap: sequence more individuals, run another phylogenetic tree, and the number will settle. It won't, not because the data is bad, but because the two guides are not measuring the same thing and never were.
A species boundary is a decision about where to draw a line through continuous variation, and there is more than one legitimate place to draw it depending on what the line is for. A taxonomist reconstructing evolutionary history wants every distinct lineage named, because the tree itself is the object of study; missing a branch is missing the finding. A conservation body assembling a protected list wants units that map onto something a government can act on: a population with a range, a threat, a funding line. A field guide wants units a person standing in a marsh with binoculars can actually tell apart. None of these are wrong methods for finding out what a frog is. They are different jobs a name has to do, and a taxonomy is a tool built to do one of them.
This is why more data doesn't converge the two lists. Feed both authorities the same complete genome and the phylogeneticist finds four genetically distinct clusters worth naming as species, because clusters are what the method is built to find. The conservation body looks at the same four clusters and asks whether splitting them into four legally protected units, each with its own recovery plan and its own threshold for extinction risk, produces four fragile lists instead of one list durable enough to survive a bad funding year. That's not a disagreement about the frogs. It's a disagreement about whether the downstream machine, a court, a budget, a field survey crew, works better fed one category or four. The data never settled it because the data was never the disagreement.
The same pattern shows up whenever two catalogs disagree about the same object and everyone assumes the fix is more precision. A hospital's billing code for a condition and a researcher's diagnostic category for the same condition split and lump patients differently, not because one team missed a symptom, but because a billing code has to map to a reimbursement decision and a diagnostic category has to map to a treatment protocol, and those two decisions don't always want the same boundary. A library's genre shelving and a publisher's marketing category for the same book diverge for an identical reason: one is built for someone browsing, the other for someone selling. Nobody involved is confused about the book. They are answering different questions with the same word.
The mistake is treating classification as a photograph of the world when it is closer to a filing decision made for a purpose. A photograph can be checked against the world and corrected. A filing decision can only be checked against the purpose it serves, and two purposes can both be legitimate and still produce incompatible files. Calling in more evidence to referee that kind of disagreement is a category error dressed up as diligence. It looks like rigor. It's actually a way of avoiding the harder conversation, which is naming what each list is for and admitting that they might both be right for their own job and still never agree.
The next time two authorities split the same organism differently, the question worth asking isn't which one is correct. It's what each one needed the split to do.