Yes, [link|http://www.amnh.org/exhibitions/Fossil_Halls/cladistics.html|cladistics] relies far more on observation than genetic testing. But that is because observation is the simplest (and until recently the best) tool we had for studying the underlying evolutionary history.

Given that the evolutionary history is believed to have happened, and the tool is fairly effective, we would expect its conclusions to be generally backed up by any improved tools for analysis. And so it has proven with genetic clocks (the most widely used of which is mitochondrial DNA) generally agreeing with existing conclusions. (There are disagreements on details of course - but then again we never expected all of our best answers to be correct.)

However there always has been and will continue for the forseeable future to be some churn in the classifications. Part of that is the slow correction of historical mistakes. (One wonders when they will officially change the name of homo sapiens. We are a great ape, more closely related to the chimpanzee than either of us is to the gorilla.) But part of it is that the world resists convenient classifications.

What is generally not understood except by scientists - and is utterly rejected by uninformed creationists (sorry Wade, this is accurate) - is that the concept of a species is rather vague. People commonly think that two animals are the same species if and only if they can interbreed and have fertile offspring. With obvious caveats about gender. Yet they then turn around and accept that a Great Dane and a Chihuahua are both dogs despite obvious anatomical difficulties in interbreeding. (Let us not get into the troubles you have when there are other modes of reproduction, as happens with plants and bacteria.)

Before you stop and say that this is an artificial example, tell me what we should do with the [link|http://www.cs.colorado.edu/~lindsay/creation/ring_species.html|herring gull]. That isn't human interference. There are two species in England. Without the rest of the ring we would think there nothing interesting about that. But we have a complete living chain between these two species. Somewhere between them we have to decide what we are calling different subspecies, and species. But any such division is necessarily an artificial human imposition.

A similar situation exists with languages. It is obvious that French and Italian are different languages, and both are different from Latin. But both once were Latin. At what point do you say that the languages separated? At what point do you say that they stopped being Latin? Was there a moment where at one point people spoke Latin and the next French? Was there a moment where France and Italy were joined by a common language and then were divided? Of course not!

So the problem is this. We may agree that the existence of evening does not mean that day is the same as night. However knowing and agreeing to this does not assist us in agreeing where in the evening to draw the demarcation. And over time biologists will change their minds about what demarcations to draw where and why, and the exact classifications of species and subspecies will remain in flux if for no other reason than that we need to divide the placement of different labels when no clean division exists.

Cheers,
Ben