Of course, I don't want to be too reassuring. Several folks pointed out that not all herniated discs retreat back to their home between the vertebra. Indeed, probably most don't. So, how about some real data, some hard numbers on how herniations change over time? I give you: Kjaer 2016, “the first study to investigate changes in the size of lumbar disc herniations” over a long period. See the bibliography item for full details, but the upshot was: 65% did not change, 17.5% resolved, 5% fluctuated (weird), and only 12.5% got worse. Those numbers are not awesome numbers — obviously herniations do not all magically go away — but I do think they are different and much less discouraging numbers than most people have in their heads, I think.
Not that herniation severity correlates well with pain or predicts recovery in the first place.
Of course, I have now cited Kjaer et al in my low back pain tutorial.