I tried the difference and MP3 320kbs and ALAC sounded pretty close. There might be differences but it all is very close to my ears. One might see a difference in a graph maybe.
Five bucks says that they were using a track that was peaked out. The drum beats were probably clipping. I've found that certain MP3 encoders tend to shift the volume a little bit as they encode. iTunes generally encodes a little quieter than the original track, but there are some that boost the level a hair. If the track was normalized up to 100% (like almost all rock music is) and they used one of those encoders, it would push the loudest part of the track (the snare drum transients) up into clipping and make them sound crunchy.
If you see them again, take a track that is normalized down to 90% off the peak and see if it's still crunchy. I bet that fixes it.
They were playing tracks in a pro-tools session, and the waveforms did not look (or sound) clipped.
It is more noticeable when you are listening to isolated acoustic instruments (which audio engineers do quite often). But most fully mix downed tracks are so dense that it masks it.
I understand your hesitation to believe it, but these people are paid to hear those types of differences, and spend lots of cash on a stereo that can show the difference.
So it really depends on the situation. I always stick to the highest bit-rate when possible mainly out of principle. It is a little silly if you ask me to worry about hard drive space when you are inconveniencing yourself in so many ways already to attain maximum SQ.
It is definitely a waste time for portable listening though.