_Britannica.png "Hunt-Lenox Globe (Kattigara - CC 3.0 Unported)")
It is an unfortunate circumstance that ENCODE publicity decided to declare “junk DNA” dead, again. It’s not a totally unique position. Creationists and John Mattick have argued that there is no useless DNA for ages.
The demise of “junk DNA” is a fait accompli of the way “functional” is defined. It is not a definition of “functional” most of us would recognize. Ewan Birney, who should know, explains that when ENCODE says “functional” they mean “not biochemically inert in at least one of our many assays*”. As Mike has noted from his own research experience, many totally random DNA sequences synthesized in a tube are “not biochemically inert” nor are they biologically “functional”.
The fact is, if you only think of “junk DNA” as a problem, you aren’t seeing the forest for the trees – and you certainly are lacking a touch of poetry in your bleak soul.
For starters, there is little mystery surrounding the existence of most of the “junk DNA” and has not been for some time.
There’s a saying attributed to Sydney Brenner that finely parses the distinction between “garbage” and “junk” – garbage is the stuff you throw away, junk is the stuff you keep (try to find a reliable quote of this if you dare). Seems to me Sir Sydney was working hard to defend a term (coined by Susumu Ohno) that does not reflect our current understanding very well (science is littered with such unhelpful, linguistic artifacts).
My position is that variation at the vast majority of base pairs in the genome has somewhere between no effect and almost no effect on organismal fitness, but that does not make it junk. Junk undersells those seemingly useless base pairs. Junk is clutter. It gets in the way, detracts from your existence. Junk DNA is much more.
Most “junk DNA” is derived from transposable elements. It’s the fallen bodies from our genome’s history of evolutionary warfare with these parasites (we lost most of the battles). It’s the jungle where parasitic transposable elements threaten to launch themselves at the few functional bits of DNA we have left with the potential to maim us or become our newest exon. The evolutionary history of our species is written in junk.
Introns allow our paltry few protein genes to generate an enormous diversity of proteins through the regulation of alternative splicing (a topic near and dear to my heart). Most intronic sequences are the “junk” that provides a sea of possible binding sites for splicing regulators. They are the haystack that the splicing machinery searches for the needle that will allow functional proteins to eventually be created from mRNA templates. Worthless.
The variation that makes you you is mostly junk. The highly variable repetitive elements that are assayed in DNA identity testing are junk. Your junk is more unique than the sequences that make you you.
It’s the undiscovered country. The complete human genome sequence is incomplete. Around 10% of our genome stubbornly refuses to be assembled, and it’s because it is highly repetitive sequences, like all that other junk.
Somewhere in that junk is the code for:
The world is a more exciting place with dragons.
*Credit where credit is due. ENCODE’s reproducible assays are an amazing triumph of technology, skill, and coordination.
The Finch and Pea 
I usually prefer the more specific term selfish DNA, which does not include everything that has been called junk.
And (I’m not putting words in your mouth), we need to be careful to distinguish between DNA that has served as a reservoir for new function and “repurposing” (Sean Eddy’s term), and DNA that has been retained because it could serve as a reservoir for new function and repurposing.
The first idea is unquestionably true, while the second one is probably false. The parsimonious explanation is that we no mechanism or selective pressure to get rid of excess DNA.
I can say without reservation that I agree with Sean & you. DNA need not be meant to be used for that purpose to provide capability that DNA that is “supposed” to be there cannot. Ali elements don’t mimic exons (sort of) to help protein evolution. That characteristic is an artifact of their evolutionary history and their selfish interest.
I think you mean selfisher DNA. As we learned from Dawkins, who is not an evolutionary biologist, all DNA is selfish.
Doolittle and Sapienza actually cite Dawkins, but the Nature editors probably didn’t let them use the word ‘selfisher.’ People hate the term junk DNA, but just think about much more fun we would have had if ‘selfish DNA’ was the more common term, with all the angry references to Dawkins that would result.
I think we should take a page from the microbiome people and suggest that the “junk DNA” genome is the real genome (after all there is more of it) and the “functional” genome is just the bit it uses to carry it around.