Plenty of people so far have commented on how the recent Nightline debate was inexcusably abominable, and they've said it better than I, so I won't bother to give my own rehashing of the catastrophe.
Instead, I wanted to comment on one point that the atheist side would have done well to have had in its pocket - and which we all would do well to keep in mind when debating the ignorant on the subject of evolution.
Cameron and pal seemed to have a difficult time understanding the concept of an "intermediate form" of an organism - citing the absurdity of freakish chimeratic hybrids (argumentum ad lapidem) as proof that one species couldn't turn into another. Given the source, I can understand and almost excuse his ignorance for reasons I will explain below. But what bothered me most of all was that he was not alone in this shortcoming: the moderator didn't understand the concept either (and I hope beyond hope he was simply playing devil's advocate), and our valiant defender (Brian Sapient) did a rather terrible job of setting things right by not attacking the misconception on its most fundamental level.
The reason the theists (and, to be fair, people in general) have difficulty with this concept is due in part to humanity's excessive faith in its systems of categorization. Any linguist or anthropologist (since Saussure, anyway) will tell you that all human systems of symbolic organization and classification are completely arbitrary. That is: there is no intrinsic link between an object or group of objects and the words or symbols used to describe them. Amazing studies have been done on cross-cultural variation in color words (the classic crayola set of seven colors is far from universal: many cultures combining blue and green, for example, or having only words for black, white, and red), emotions, and so forth. We talk about and perceive the world according to arbitrary systems of classification that work through common consensus.
The same is true for Biology, in which discipline this system is called taxonomy - the organizational structure by which we classify all living beings, the finest resolution being the level of species. Although rigorous science now has a less-arbitrary way of determining speciation (I believe that to be of different species, two animals must not be able to successfully breed, or to produce potent offspring by their union), this logic is a relatively recent innovation. Far more representative of human thinking as a whole, I think, was Cameron's revealing comment that Horses, Mules, Donkeys, and Zebras were all "horses" - meaning that they were of the same species. But by what criteria does he make this assertion? I doubt a horse breeder would agree. I wonder whether he would include Camels as well, being similar in at least some respects? But because the distinctions among these species are not relevant to Cameron, his brain feels perfectly justified in lumping them all together as a category.
The point here is that, for the sake of clarity and brevity, we only have a functioning vocabulary of words large enough to get by - and again, that vocabulary is entirely arbitrary.
So where does this leave us? If we can explain that our classification system is not black and white, then we can begin to tackle the mistaken image that the intermediary species between a crocodile and a duck would be a crocoduck. Sapient hit on it when he said that every organism that has ever lived is an intermediate form: we can only make the distinction between species by looking synchronically at every living being and drawing meaningful and useful distinctions between them as they presently appear. But the creationists seem to believe that this kind of rigid, functional taxonomy should also be visible diachronically as well - when in fact the process of change is so arduously slow and incremental that changes visible to the human eye only happen on an order of thousands of generations.
Take the breeding of dogs as a smaller-scale example. You would say that a Wolf and a Poodle are not the same species (I assume this is so - if merely for the fact that trying to get a wolf and poodle to mate would be ugly). Let's assume, now, that we have a complete genealogy for every wolf and every dog that has ever lived. Now, take the wolf from which all dogs supposedly derive (a tautology in itself) in the one hand, and an actual, living poodle in the other (any one will do). We can now draw an unbroken line of succession (patriarchal or matriarchal - doesn't matter) from primordial wolf to modern poodle. Now, compare the bookends to the organism directly in the middle: chances are you can tell the difference easily, and if you had occasion to you would come up with a name for this critter. But now disregard one half of the line and do the same to the other, examining the center point against its bookends. Now do it again, and again, and again. The more often you split and compare, the less drastic will be the degree of difference among the three examples. If you carried this process on indefinitely, you would ultimately end up with a group comprising only three generations: great-grandparent, grandparent, and parent. Clearly each of these is also different (we humans might not be able to tell, but the canines in question certainly could - another distinction between useful and non-useful information), but we would not call them different species.
At what point, then, do we draw the line? The point is that we can't. Our taxonomy is purely qualitative, and we only have so many levels of qualitative resolution. We can distinguish between individuals of our species (usually), and we can distinguish between individuals of different species. But when we try to resolve an order of magnitude between these two, we do not have the conceptual or linguistic equipment to draw a distinction without beforehand agreeing on a rigid set of objective standards based on homology or chemistry rather than the "squishy bit" logic of our brains.
What, then, would the name be for the missing link between species X and species Y? If we haven't found one yet, we don't have a name for it precisely because we've never yet had cause to refer to it. When we do find it, though, we will probably know it by a Latin version of its discoverer's name - because those are our arbitrary cultural rules for naming unnamed species.
Taxonomy, and language more generally, is merely a convention that allows us to create meaningful symbolic representations for both subdivisions of an unbroken spectrum on the one hand, and agglomerations of discrete instances on the other.