Given that I've never yet used an infinite scroll implementation that had reliably-working 'back' button behavior, I'm skeptical of the claim that yours worked. I'd put some money on the hypothesis that it worked under ideal conditions, or worked in testing, but failed an irritating fraction of the time in actual real-world usage.

EDIT: In fact, even if by some miracle it didn't, it probably still triggered a twitch of stress and irritation in users who assumed it would.