I think we might further define two types of lumper-splitter divides: analytical, and operational. This article explicitly restricts itself to analytical issues, though I'd say it strays into operational territory in the "software modeling" section (since it refers to models made with the intention of immediately implementing them).
The most obvious examples of operational lumper/splitter problems occur in geopolitics. I'm something of an operational lumper myself, and would love to have been born in a time when one could claim to be a world citizen. Not gonna happen anytime soon, though.
However, it appears that I'm an analytical splitter.
As a lumper, I'd like to relate this somehow to the "lumped element" model view of a system (e.g. circuit with discrete components connected by ideal wires) versus the "distributed" model. The splitters will chew me out on this, though, I'm afraid; the analogy cannot hold because of the differences.
E.g. "all red wine tastes the same to me, but differently branded 12AX7 vacuum tubes sound totally different".
"These three Lisp dialects are totally different languages, but everything else is an amorphous Blub."