- Refer to a taxon.
- Form the first part of the names of all species within that taxon.
They do #1 poorly because they're defined typologically. The definition for a genus is just, "Some taxon that includes the type species." But they could do this task well if they were given phylogenetic definitions instead.
But that doesn't work, either, because it conflicts with #2. Taxa defined by phylogenetic definitions may overlap, or be empty. For #2 to work, every single species has to be part of one genus (and only one genus). Phylogenetically-defined taxa don't really work that way.
So genus names are stupid. But we have to use them, because there's no other system for naming species.
Because they refer to taxa poorly, different disciplines often have wildly different ways of using genus names. In entomology, a genus may have hundreds of species. But, increasingly in dinosaur paleontology, each genus gets one species. Nearly every single Mesozoic dinosaur genus is monotypic.
This is a pattern we see over and over in recent years:
- A new dinosaur species is discovered.
- Researchers do a cladistic analysis and determine that it is the sister group to another species, already named, Originalgenus oldschoolensis.
- At this point, most researchers in other fields would name the new species something like Originalgenus noobius. But, no, even though it's barely different from O. oldschoolensis, it gets a new genus, so it's Newguy noobius.
Today's researchers do have an excuse prepared for #3. It goes like this:
- "Sure, this analysis shows it as the sister group of Originalgenus oldschoolensis. But what if a future analysis shifts it a bit so that they no longer form a clade? Cladistic taxonomies may require it to be placed it in a new genus."
- "We sure as hell aren't going to let anyone else name that genus; not after all the work we did describing it!"
Ignoring the mild egomania in #2, this sounds reasonable enough. But this way of thinking has given us a huge number of completely redundant names, as well as pushing dinosaur paleontology into an extreme corner of the "splitter vs. lumper" debate. Isn't there a better way?
There Is a Better Way
Just give your species a new subgenus!
Genus: Originalgenus Original Author 1900
Subgenus: Newguy subgen. nov.
Species: Originalgenus noobius sp. nov.
Now, as long as O. noobius continues to be regarded as the sister group (or otherwise "close enough") to O. oldschoolensis, you just keep the status quo. But if things get shaken up and O. noobius requires a different genus name, by the rules of the ICZN, it has to be Newguy. And you still get the credit!
I know, it's stupid ... but it works!