Many of the CRS were developed in the 60s by IBM on the TPF (Transaction Processing Facility) mainframe operating system. The system's characteristic is short, high volume and high speed transactions. Even "rebooting" the system takes very little time (less than 1 minute).

Everything had to be small and quick considering how expensive CPU power and disk storage was back then. Thus, all these codes.

Your record locator is actually a quick index / hash into the system's disks to find your PNR. Someone told me that it is actually a disk address but I am not sure about that.

So, as PNRs are purged (usually 1-3 days after the last segment is flown), spaces free up. They do run overnight clean jobs to find free space and relink the pointers in the system (sort of like defrag).

Disclaimer: I am not a TPF programmer but I have worked with some and learned these things from them over the years. If anyone has more information and literature on this, I would love to read up and learn the mystery.