So... who can prove it? Like, with a blind test?
Look at hydrogenaudio, for example. Except for rare cases ("problem samples" etc.) the consensus there is (afaik) that with current codecs even 128 is in most cases transparent.
For me, 128 isn't tranparent. Even on a crummy built-in sound card and a CMOY amp I can ABX 128 vs. Flac easily. I didn't notice artifacts, but the 128 is flat, less air or soundstage. ( Now, 192 vs. Flac is much harder for me - I'll have to try that one again later... )
[Edit, I'd better qualify that, 128 CBR is easy. 128 VBR not nearly so easy, but still discernable.. I don't think I want to even try 198 VBR now.]
It bothers me that hydrogen audio dosn't take into account the type of equipment being used. Claiming transparency based on tests with "stock" sound card setup isn't the same as if it was acheived with high quality DAC and headphones.