First, let me state my baises: I'm much more familiar with the
information processing side of things that the nuts-and-bolts molecular
biology, so I'm counting on you to keep me honest. I also realize that
to a person with a hammer everything looks like a nail. That said...
1) Since nature endures little that is not immediately useful
(and usually efficient), "junk" DNA is probably not much like "junk"
mail (CONGRATULATIONS! YOU MAY ALREADY BE A WEINER!). It has been
suggested that the "quartenary" or structural use has been for
positioning the strand. This is attractive, but seems weakened by the
notion that the unused strand, in fact the major portion of the
would have to maintained with the same vigor as the "live" part. The
overhead would be high for that kind of return.
2) On the other hand, the proximity of used sequences might make
the strand vulnerable to mechanical breaking. Inert portions could
the events, and, since the bonding is both lateral (base pairs) and
longitudinal (along the ribose spine) a cabling effect would occur.
3) There has been speculation that the unused portions may be
inactive, older versions of modern code. If this is the case
(but probably not structurally, see "2"), there might be some
at work. It might even permit some form of intra-genetic competition,
reducing the shock effects of some forms of mutation.
I eagerly look forward to feedback on these speculations.