Many US states, airports and agencies have begun using facial-recognition technology despite concerns among law enforcement authorities such as the Tampa Police Department, which abandoned its system because it had not helped catch a single criminal. Chicago police recently announced plans to install thousands of cameras around the city that track unusual movements by individuals, even though this "content analysis" surveillance technology has yet to be proven.
"Facial recognition got a lot of hype after 9/11, but it has problems," said Steven Gish, senior research analyst at Roth Capital Partners. "They don't have one that can do one-to-many matches. It is really good at doing one-to-one matching -- when you are at a counter to get a ticket -- but not picking a face out of a crowd."
The government's withdrawal of the "Total Information Awareness" project, which would have linked databases to compile composite "signature" behaviour of terrorists, was a significant setback for the large-scale use of such security technologies. Groups such as the Association of Computing Machinery told the US Congress that the massive system risked opening the door to identity theft or generating "false positives" from imperfect analytical tools.
Identifying the threat
Before the US government can decide which technologies are valuable to security, officials must define the threat they are working against. To date, the White House has described domestic defence goals in only general terms in the seven strategic reports it has issued over the last two years.
"We found there was no commonly accepted set of characteristics used for an effective national strategy," wrote the authors of a February report by the Government Accountability Office (GAO), the investigative arm of Congress. "The seven national strategies related to homeland security and combating terrorism vary considerably in the extent to which they address the desirable characteristics that we identified."
In this absence of clear direction, US homeland security officials approached their mission by identifying key landmarks and other potential targets for attack. That strategy quickly proved problematic when the list ballooned to more than 33,000 sites as every state and region lobbied to include specific buildings, bridges, stadiums, monuments and other structures, federal officials say. The list has been cut to about 1,700 sites, sources say, but it is still too long for any meaningful planning.
"You protect a bridge against what? An airplane, a boat, a bomb? It's not possible to plan for every possible contingency," said Randall Yim, formerly the GAO's managing director for homeland security issues and newly named director of the federally funded Homeland Security Institute. "If we have information about a specific threat, such as one against a particular financial institution, that's one thing. But it doesn't work if you don't have that kind of information, which is most of the time."
