I thought the test flawed, because I found the difference too obvious (done only 2 test, but I didn't hesitate at all).
They didn't specify which encoder was used, and I'm thinking there's perhaps a volume imbalance too.
I know I reach relatively easily transparency, by using lame encoder.
Unless using a high end headphone makes a difference (the guys at hydregenaudio, claim it doesn't matter).
That's strange, I have a pretty good ear and I don't notice much volume imbalance with it. The way I did the tests was to click all 3 immediately to get everything buffered up. Listened at normal volume.
The first test was at work on low quality headphones and bare bones built in dell audio. The second test was at home through an Audioengine DAC into DT 990's.
I use LAME to do all of my encoding and it works pretty well but I do it at 320 and no joint stereo. I know a lot of folks use Itunes and that is a lower quality encoder. I believe the site stated they used exact audio copy and then LAME.
If anyone cares to devote the time I'm sure you could load the two samples in something like Audacity and diff them.