The Software that Cried Wolf?

Here is one of those stories that one expects to get from the BBC but which they actually harvested from their Associated Press feed:

New York's police chief has delivered a cheesecake to an elderly couple in Brooklyn, to apologise for dozens of mistaken police visits to their home.

A computer glitch had led officers to Walter and Rose Martin's home 50 times in the past eight years, police said.

The latest intrusion came on Tuesday, with officers pounding on the front and back doors, shouting "Police, open up!"

Thursday's visit - cheesecake in hand - went well. The Martins, aged 82 and 83, shared pictures of their grandchildren.

New York Police Commissioner Raymond Kelly visited the couple's modest home on Friday to "apologise and explain" for the mix-up, NYPD Deputy Commissioner Paul Browne told the Associated Press.

The problem started in 2002, when the Martin's home address was used as "test data" for a new computer crime-tracking system.

The couple complained about the harrowing visits in 2007, but the data remained in the system despite efforts to "purge the records", Deputy Commissioner Browne said.

He added that the Martin's address had now been flagged with alerts "barring officers" from questioning the octogenarians.

This is a really nice little story about how the NYPD can put its best foot forward towards community relations.

As a story about software systems, however, the positive twist of the tale is, as they say, not so much. Indeed, it is yet another incident of a case where those with the power to do so could not update a database to repair inaccurate contents. There is also a question of how the matter was resolved. What if there is a criminal problem at the Martin's address? What procedures are in place regarding how that "flag" is handled to determine whether the report is valid or an artifact of this latent error? Furthermore, assuming those procedures exist (which, as far as I am concerned, may be a bit of a stretch), how will they impact the efficiency of police responsiveness in general, since, presumably, the database was initially set up to improve both the efficiency and the effectiveness of such responsiveness? Only when we have good answers to these questions shall we know whether or not this narrative can be applied as a lessons-learned case study; and only if the narrative can be applied will there be even the slightest possibility that the next time it occurs (and there will be a next time), the outcome will not deteriorate into farce.