The latest "junk" DNA finding is that pyknons have been found in the much studied plant Arabidopsis thaliana. Pyknons are short (about 20 nucleotides) DNA sequence patterns that are common, and show up in both genes and the "junk" regions of DNA (introns and intergenic).
In addition to their dual presence as (i) evolutionary "junk" (which increasingly is being found to be useful in spite of evolutionary expectations) and (ii) within genes, pyknons are also similar to regulatory RNA sequences, and have some interesting correlations and patterns regarding their chromosomal positioning.
So why are pyknons so prevalent in these "junk" DNA regions? The researchers think they are a by-product of the action of RNA gene silencing. In other words, it's just junk after all. If the history of evolutionary expectations is a useful guide, you can expect that this will be yet another one gone wrong.