Nevermind they were just talking about transcription factor binding sites. None of this is new stuff.
No, you were right. The ENCODE group published a paper last year claiming that virtually all of the genome is transcribed (ENCODE Project Consortium in Nature, 2007). While it's an interesting observation that is probably true, the question is what does that really mean? Unlike translation which is a deterministic process (e.g. the codon AAA will always code for the amino acid lysine), transcription is a relatively sloppy probabilistic process where the DNA signals to start and stop don't always work properly, so you end up with things like aborted and run-off transcripts. However, the cell has control measures in place to degrade abnormal protein and RNA transcripts. So if you take a snapshot of the cell at a particular moment, you'll find all of these artifacts that are destined for the genetic dustbin and will never be translated into proteins. Add onto that, that these are relatively rare events, so if you find 1 of these long run-off transcripts in a sea of 50,000 normal ones, does that really matter?