Sounds like some basic housekeeping stuff has been omitted - like a duplicate batch check process that is SOP  to stop this 'human error' occuring... really the human and inexcusable error is not having a process to prevent the error happening.... This stuff is data processing 101 as our american colleagues would say....

Its a schoolboy error almost as bad as confusing the North and South Korean flags.... Big apologies there too!!!  

I can remember 20 plus years ago having checks on duplicate files (or tapes as we called them then).

If they contained the same number of records, and matched on a hash total of the monetary value, as a previously processed file, then they were undoubtedly a duplicate.

This worked fine until one month we had a file from one of the large Insurance companies, who hadn't managed to sell a single new policy in the month, nor have anyone cancel, and they hadn't changed any of the Premiums, so lo and behold it did match on both record count & financial amount, yet it wasn't a true duplicate.

Then we went back to the drawing board - stripping out a single record to get the control totals to differ so that we could get the file through.