When I've heard a change due to break in, not swapping, it wasn't subtle and it wasn't huge. This was not black and white, big or small, dramatic or ordinary.
My system can easily discern a cable change, a CD mat insertion, a change of footers, IC or SP or PC swap, a tube swap (when I used tubes). Some of those changes were big, some small, but they all had one thing in common. One area in particular stood out at first, only to reveal more once the initial impact wore off.
It took different manner of music to bring out the more subtle improvements as one type of music, let alone one song, or one part played over and over, can, in no way, help to ascertain the benefits. Obviously, one song or a part of it, can not convey the magnitude of improvement since that particular piece only covers a mere, few facets of the overall performance that can manifest itself on ones system. It will hit big time at first, then slowly reveal more of itself over time and with different music.
The changes I've heard during break in usually take me by surprise. I'm not expecting it, it just happens. My brain is in its usual auto mode when I hear the difference. It's when I'm listening for listenings sake. Nothing scientific here. This is with recordings I'm very familiar with. It's like I'm being tapped on the shoulder (ear). Subtle, yes, but noticeable.
That's when I focus my attention. That's when I break out other familiar recordings to see what else parallels what I'm hearing, or compliments with other areas of improvement. This is usually followed by one or two more levels of improvement, over time, and then no more.
It's something I've come to expect but not anticipate. One allows the natural event, the other forces an outcome that can delude. I think that is why double blind tests aren't valid due to the very nature of their being: they force anticipation, instead of allowing something natural, which can seriously skewer a test subjects ability to judge. With their sanity, prestige, and reputation on line, all manner of perception can be tainted.
The same can be said for the well intentioned performing their own DB tests. Being of same mind, or at the very least, friendly and of a peer, they can negatively affect the results due to the very nature of their clique.
All the best,
Nonoise