http://www.ted.com/talks/jack_horner_shape_shifting_dinosaur...
It's also pretty entertaining to watch
A species is a particular genetic type which does not routinely interbreed with other related types it encounters, and retains some physiological differences from them, as a forked, internally consistent genetic branch.
We're a hundred million years late to demonstrate that the sauropods were or weren't interbreeding. We don't have that behavior to observe. We don't have any DNA, a fairly accurate way of demonstrating breeding pools, to work with. We only have the occasional left third rib, and equations derived from the most similar complete skeletons we know of. Even when we do find a complete skeleton, we can only date it to a certain degree of precision, and while understanding that there is an evolutionary tree that is constantly sprouting and changing branches, we have nothing to really demarcate when one species "becomes" another - we are speculating about two specimens' reproductive compatibility with millions of years between them.
Could you breed with a proto-human? Would you? How far back? Would the kids come out alright? Would you call them a new subspecies or pick one of the parents' taxonomy? On top of that, to represent the last several millions of years of subtle changes, new branches, & trimmed branches in hominid genetics, what if you only have 6 specimens closer than chimpanzees, mostly age/sex-indeterminate, only one of which is a complete skeleton? How do you draw an org chart of the hominids then?
There are size variations corresponding to age and gender, heritable variations within a breeding pool, variations corresponding to differential nutrition & the good times vs the bad times for a population, variations corresponding to individual sickness and injury, variations corresponding to the entire population getting steadily larger or smaller for a thousand or thirty thousand or a million years, variations corresponding to sudden changes in habit associated with adaptations to rapid changes in environment, variations corresponding to drift from a parent population associated with permanent reproductive incompatibility as well as reversible reproductive isolation.
The historical paleontological record is thousands of times richer in micro-scale detail than we have the data to describe, even if we did have good words to describe it with that the popular media could understand. But they want to talk in terms of species, because that's what people are familiar with.
Ex: Sabertooth cat and http://en.wikipedia.org/wiki/Narwhal
The scientists that study this kind of thing try to check for similarities and classify them as the same or different when appropriate.
They probably classify them as different if the time they lived (based on the surrounding rock they were found in) or their bone structures is sufficiently different from anything previously discovered.
It's also a lot easier to merge two classifications rather than split a group into two.