There are truths both ways here.
Back in the 90's, Pepsi ran a marketing campaign called the Pepsi Taste Challenge. Time after time, on blind tests, random test subjects consistently chose Pepsi over Coke; so much so that Coke introduced a new version flavour. The "New Coke" flopped. The reason? What tastes good with one sip, may not be enjoyable by the time you're halfway through a can.
Coke followed up the flop of "New Coke" by doing whole can tests - giving people the full serve rather than just sips. Overwhelmingly, people preferred Coke by the time they drank a significant amount.
I don't want to start a Coke vs Pepsi war here, that's not the point of the story. The point of the story is that what is enjoyable when we flick between sources (sound, taste, texture, any sensory input) may be quite different to what is enjoyable when we stick with that same source for a period of time.
In auditioning, this is also true. If you listen to a high quality mid-focussed speaker/headphone and then switch to a high quality bright, sparkly speaker/headphone, you will probably find one seems more enjoyable to you (based on your tastes), but if you stick with one or the other for a period of time, you will likely learn to enjoy what it offers. In other words, A/B testing is very valuable to determine differences, but it may not be a good/reliable reference for determining true preferences. (E.g. a bright, sparkly headphone may seem amazingly revealing and resolving in a quick A/B, but may become more fatiguing with prolonged listening)
To me, it's all about balancing what tests you use for what ends. Lossless vs lossy is definitely about differences, but it's also about storage capacity for many people. Long term testing will show you if you can still enjoy lossy formats in "real life" even though you may know (from A/B testing) that there is a difference.
Edited by Loquah - 3/13/12 at 4:19pm