I'm not sure where I encountered this hypothesis but I find it compelling. As noted by many, junk DNA, acquired from viruses and mutations and genome shuffling, is quite a puzzle. Why does it persist? It takes energy to copy, and misreading it can cause fatal or maladaptive mutations. From that perspective, it shouldn't persist (with slowly accumulating drift) for billions of years, as some shared junk sequences have across species. But it does.

Obviously, because it isn't junk; it is of value to the organism. Even if it's not of any use right now, even if it's completely biologically inactive at present. Because it is still extremely high entropy information. They're remnants of solutions other living systems once used, at some point, to solve the problem of staying alive.

If I were going to try and exploit genetic mutation to produce novel solutions to biological problems, I would start from an existing genome. In fact, I'd start with as much data, from as many organisms, as I could get my hands on and store. Perhaps we carry junk DNA because mutations in existing coded sequences, even mutated, currently useless ones, are far more likely to be functional, and so potentially a useful adaptation, than literal randomness. It's life's portfolio of solutions, badly photocopied little snippets accumulated over the years, and we all carry it around for future generations that might live in an environment where it's useful.