Borderland: The Failure of the Virtual Fence

by Benjamin Domenech on 11:51 pm March 18, 2010

Not a virtual fence

[tweetmeme]

The announcement this week that the feds have frozen funding for the much-maligned SBINet project, the Boeing-managed program launched in 2005, shouldn’t come as much of a surprise. Costs have already doubled beyond what was originally anticipated, and it still isn’t working.

So far, only a 28-mile prototype of the virtual fence in Arizona has been delivered to the government, and not without snags. Previous GAO reports described cameras with limited ranges that failed in the desert heat and sensors that couldn’t identify nonthreatening movements caused by animals or the wind… Christopher Bronk, a fellow at the James A. Baker III Institute for Public Policy at Rice University who has followed the virtual fence, said a key problem was that the project’s installations were too prominent, with highly visible stationary towers and bulky propane tanks that would-be crossers can spot from far away and therefore avoid.

This is a classic example of a good idea being completely mucked up by government restriction, contractors over-promising on delivering technology, and federal COTS policies. It’s time to start over from scratch.

I talked today with a friend who works in this field personally, and he cited a series of fundamental problems plaguing the Boeing approach to this project. This report from last fall details some of the issues: a basic inability of the technology being used to distinguish between coyote, tumble weed, and homo sapien, vulnerability to wind and rain, obvious rigs which are easy for crossers to avoid, etc. The commercial off the shelf approach may appeal to congressmen and DHS Sec. Janet Napolitano, and it’s the sort of thing contractors are good about promising. But the effect in this case has been a project with dozens of different technologies, from different companies, all of which need to talk to each other.

The upshot: after more than $800 million, Boeing’s been able to cover fewer than 30 miles out of roughly 2,000. The problem isn’t so much the concept as the application:

Still, a video presented by Rep. Michael McCaul (R-Texas) of SBInet’s video capabilities seemed to impress everyone. Recorded in February, the video shows six trespassers into the United States effectively tracked and stopped by US Border Patrol agents. Using the system, agents in a Tucson command and control center where able to guide agents in the field and inform them of possible threats before they were physically encountered.

The idea of a virtual fence isn’t a bad one. Setting aside the social ramifications of the borderland policies — I’ll be writing next week about an interview with Michael Lind about his suggestions for the president’s immigration reforms — a virtual fence could prove to be more effective and efficient than a real-life one in the long term. But it requires a dedication to a higher level of product refinement, comprehensive testing, and use of higher grade products that work together efficiently — and those things will all cost more money, not less. And thus far, the millions in taxpayer dollars haven’t achieved results much better or more consistent than border minutemen whose personal setups from Fry’s Electronics and eBay.

Follow Ben Domenech on Twitter.

Previous post:

Next post: