Get PJ Media on your Apple

Belmont Club

Spin versus spool

December 28th, 2009 - 7:04 am

Janet Napolitano’s talking points emphasizing that “the system worked” to prevent a terrorist attack seem to be fizzling just as badly as Mutallab’s bomb.  Even CNN is ripping into it, as their video shows.  It’s becoming a running joke. As the previous post argued, treating this problem with upbeat spin isn’t going to work. It should never have even been tried.

While the solution to the public ridicule is likely to be improved spin, it doesn’t begin to address the real problems . One of them was casually revealed by the statement that there are over 500,000 names on an extended terror suspect’s list.  A list with that many records is of limited utility unless it can winnowed down to a tractable set in a given situation.  You really want to be able to use this data to answer specific questions like: is this man related to such and such an event or should this person fly? Simply saying that the terrorist is in the list is like saying the needle is in the haystack. It wasn’t of much use in keeping him off the plane.

So what does the architecture of the data look like? Can you run queries with joins across different agencies or across from the goverment to the private sector? Can you take one of the names on this list of a half million and find out what calls he made on such and such a date or where he’s been? What kind of metadata is in there?

Nobody is going to answer these questions in open source, but its a good bet that the difficulties to getting the data to yield actionable information will run afoul of two contraints. One is technical, but that can be solved. The other is bureaucratic and legal. That is probably a harder nut to crack.  The significance of information in a wartime setting is radically different from its import in the criminal justice system. In war information is mostly used to affect events in the future.  In criminal justice it is used to interpret events in the past. Treating information as intelligence or as evidence creates radically different organizational cultures and incentive structures. It also creates different architectures in information management.  Deciding to treat terrorism as a law enforcement problem will probably result in information strategies that are optimal for convicting people, but sub-optimal for preventing terrorist attacks.  A justice system is by definition in the business of punishing crime, not forestalling pre-crime.

None of those defects are going to be addressed by shifting from one set of talking points to the other. Newspaper articles, appearances on talk shows and assurances in print will have no effect upon solving these information problems. And they are manifest on the face of things. If Janet Napolitano actually thinks the “system worked” then she may not even be cognizant of the problem. Maybe someone can explain it to her.

That’s probably the job of those who are conducting what the Administration says is a review of the Homeland Security system. If so, here are a few suggestions which may make it more effective. First, exclude all the lawyers from the review and get a small group of experts to understand where everything is. To understand where all the marbles are kept. Next, get these experts to figure out how all the dots can potentially be connected across the whole system.  Get them to answer the question: if you could legally write a query to connect all the dots, how would you write it — never mind how many joins, how many system boundaries it crosses. That’s the simple part. Then after you’ve answered these questions, bring in the lawyers. Now comes the hard part. What you’re left with afterward — what the lawyers let you do — represents the technical potential of the system minus its bureaucratic cost. Hopefully you’ll be left with something. Then sell it to the political leadership. That’s the hardest part.

Well hope springs eternal.


Tip Jar or Subscribe for $5

Click here to view the 85 legacy comments

Comments are closed.

One Trackback to “Spin versus spool”