American Scientist

To the Editors:

In Brian Hayes’s column “Connecting the Dots” (Computing Science, September–October), the first graphic, showing connections between suspects in the 2004 train bombings in Spain, has the foul odor of racial profiling.

It is no surprise that persons with apparent Middle Eastern names are going to telephone other persons with Middle Eastern names. Perhaps the persons ­indicated in green have been charged with crimes, and their names are public record in Spain. But what about the ones shown in red and their rights to privacy? Maybe your diagram will be used in a Spanish court to wrongfully convict someone who is guilty only of being of Middle Eastern descent and having contacts with others of similar ancestry.

I was in Oklahoma at the time of the tragic Oklahoma City bombing. The media immediately started discussing fundamentalist Islamic terrorists. This knee-jerk racial profiling was halted when white male Christians were accused and later convicted of mass murder.

Jim Lawson
Leonard, OK

Mr. Hayes responds:

Mr. Lawson implies that the diagram was constructed by starting with a core group of known conspirators and then adding acquaintances or associates, selected because they happen to have Arabic names. I agree that such a procedure would be reprehensible, but that’s not how the network diagram was created. All of the individuals mentioned had already been identified in the public record as persons believed to have been involved in the March 11 bombings in Spain or in related events elsewhere, such as the September 11 attacks in the United States. No one was included simply because of casual contact with a conspirator. Indeed, six individuals listed in the diagram have no indicated links to others, so they obviously can’t be there through guilt by association.

Still, the issue Mr. Lawson raises warrants serious concern. The diagram of the March 11 social network was constructed retrospectively, when at least the basic facts of the crime were already known. Even in these circumstances, errors are possible, and, as Mr. Lawson emphasizes, they can lead to grave injustice. Intelligence agencies are now reported to be applying similar tools prospectively, analyzing a pattern of links in order to detect illegal activity. The risks of this surveillance procedure are surely greater. Whether such “blind” network analysis is prudent (or even feasible) was one of the questions I wanted to address in my column.