The brief sketch: credit reporting agencies have Seriously Hard Problems (TM) in determining which identity to associate with an incoming piece of data. For example, suppose you have a file on a Patrick J. McKenzie who once lived in Chicago. You get a report from a hospital that a P.J. MacKenzie in an unspecified town in Illinois (not the actual guy) stiffed them on a $250 medical collection. They sold the debt to a debt collector who, having run a skip trace, determined that I am probably him, and they've reported the debt against my information directly. Do you merge those two identities? By the way, he's also delinquent on $100,000 in other debts.
This sort of thing happens all the time in credit reporting. See my blog post regarding "names are hard."
I would love to hear more about this. Do you have a blog post or other online posting that describes what happened in this case?