Have you read this article? It's in the Economist, so it's not universally trustworthy, but the core of the article is correct: we have "jumping genes", called "transposons". These are pieces of genetic code which magically move around in our DNA. They are largely thought of as genetic parasites, and they tend to do pretty severe damage when we notice them.
"Retroposon" is the class of "being", including both retroviruses and transposons. However, not all transposons are retroposons. Retroposons copy themselves into RNA and then utilize reverse transcriptase to insert themselves in a new location. The second part is the important part: they inscribe themselves into the DNA wherever they want. There are other kinds of transposons, as described here.
"Type II" is interesting because it "buffers" itself from the rest of the DNA, and can hijack other segments' transposase as needed. Type III ("MITEs") are incredibly interesting because they are so mysterious. Tiny little repeating segments found all over the place. Retroposons are interesting because they are actively a part of the DNA's own replicating and "unfolding" process, whereas the other two are obviously hijackers utilizing other methods to force themselves into the DNA.
A clear example of what they can do is found in corn, which you can read about here. "Indian Corn" has weird coloration - that coloration is caused by transposons. Transposons usually inhibit or "turn off" any gene they land in. Without those transposons, Indian corn would be universally reddish-purple instead of mottled and striped. On the other hand, if they were permanently settled on the genes, rather than busy copying themselves back and forth, the corn would be roughly the same color as corn-field corn.
A far more critical example is bacterial transposons. These transposons are carried by bacteria, and when created in a plasmid (a fairy tale for another day, children, but in short: mini-DNA) they can migrate to other cells. This allows bacteria to share genetic segments, such as resistance to antibiotics, from "adult" to "adult", instead of just passing this kind of immunity to "children". This is not necessarily limited to the same kind of bacteria, or even to bacteria, although we usually think of viruses being the ones to edit other race's genetic codes.
In fact, some viruses are essentially just retroposons. Like, say, AIDS. These are usually called "retroviruses".
Pretty cool, huh? Back to the tame variety.
If you've read the long, boring transposon page, you've seen he says only 50 LINE-1 elements are functional. That isn't precisely true - in certain situations, a lot of L1s "wake up", apparently.
LINE-1 segments are found to be active in, most notably, brain production. As you are aware, if you've read the Economists' article, making a brain is a horribly wasteful process in which at least half the would-be brain cells die before the brain is even fully formed. After that, you continue to lose brain cells rapidly. Wheee!
LINE-1 transposons shut off or alter genes, radically changing how the cells mature. This creates a huge diversity of cells.
There are specific chemicals which control the activities of L1 genes, and here's where I start going off into la-la land:
I don't find it at all hard to believe that the different portions of the brain are formed by cells which are specifically affected to be a certain way, thanks to a spray of a specific chemical. This could be a small change - perhaps the neurons in one portion of the brain have a shorter fire/rest cycle - but that's really all that is needed to make the neuron more suitable for a given kind of pattern recognition.
I wonder if I could think of a similar effect for permuting computer algorithm "atoms". Of course, these systems evolved over billions of years and they still don't work quite right, so I doubt my curiosity will solve the mysteries of the universe... but I'll be happy to think about it.
And now you know about ten times more about transposons than anyone outside of the field will ever need to know. :)