rationalskepticism.org

Yep. The issue is not that DNA wrapping around nucleosomes into chromatin isn't protecting the DNA, the issue is that you don't need large swathes of neutrally evolving transposable elements to prevent accidental transcription. Simply increasing the size of the transcription factor binding site would mean you could make it much less likely for two such sites to be similar in sequence, which means you would avoid a lot more nonspecific binding resulting in accidental transcripts.

And again, the significant variations in genome size even between very closely related species is evidence against this kind of "protective junk" hypothesis.