Has anyone ever transcoded music from a high-quality aac to an mp3 (or vice-versa). The internet is full of people who say this should never be done, but apart from the theoretical standpoint that you can only lose information, does it matter in practise?

• is the difference perceivable, except on studio-equipment?
• does the re-encoding actually lose much information? If, p.e., high frequences are chopped away by the initial compression, those frequencies aren't there anymore, so this part of the compression-algorithm won't touch the data during the second compression. Am i wrong?