Towers, Sensors and Scanners, The Technology Half of Border Security

The government once promised to build a high-tech "virtual fence" across the Arizona desert. It was going to use towers, radar, cameras and a command network so advanced that a crossing would be spotted, identified and answered almost before it began. The contract went to one of the biggest defense and aerospace companies in the country. The program had a catchy name, a big budget and a hopeful press release. Then it ran into the oldest enemy of every government technology project: the real world, where dust gets on the lens, hills block the radar, and the software does not do what the brochure said.

That story is a perfect place to start, because the loudest voices in the border debate talk about only one thing: steel in the ground. Walls make good television. Software and sensors do not. But the border is as much an information problem as a physical one, and anybody who wants to win the argument for border security needs to know the technology half of the system, including the embarrassments. Conservatives believe in accountability for government spending. That means we should be the first to ask what the gadgets really did.

The failure: SBInet

In September 2006 the Department of Homeland Security awarded the prime contract for a program called SBInet, the Secure Border Initiative Network, to Boeing. The plan was to string together towers carrying radar and cameras, along with sensors and a common command system, to give agents a picture of the border.

A prototype section called Project 28 covered about 28 miles of the Arizona border near Sasabe. It had real trouble. Reports from the Government Accountability Office and the Department's own reviewers described problems with the radar, with the cameras and with the software that was meant to tie everything together. The project dragged, costs grew, and the system fell short of what the agents who would use it needed.

In 2010 Secretary of Homeland Security Janet Napolitano ordered a reassessment and froze funding for the program beyond the first sections, and on January 14, 2011, she announced that SBInet was canceled. DHS replaced it with a plan that matched technology to each stretch of terrain, using proven, off-the-shelf equipment where it fit.

Here is the lesson, and I would draw it for any administration of either party. When the government tries to buy a giant integrated system in one lump, with the vendor in charge of the vision, the result is often a very expensive disappointment. When it buys proven tools, tests them in the field and expands what works, the result tends to be better. That is not a partisan insight. It is the same lesson you learn watching a failed software rollout at a large company.

The quiet successes: towers, sensors and aircraft

After the cancellation, the border technology effort did not vanish. It got more modest and more practical, and in the places where it was used with agents nearby, it earned its keep.

Fixed surveillance towers. Towers with cameras and radar can watch a long stretch of terrain around the clock. They do not arrest anyone. Their job is to see, classify and hand a location to an agent. A tower extends the reach of every agent on the line, because one operator can watch many miles of ground.

Ground sensors. These are small devices, seismic or magnetic, placed along trails and roads. They respond to movement or vehicles and tell a station that something is crossing a certain spot. They are cheap compared with a tower, and they help most in remote areas where a tower cannot reach.

Aircraft and unmanned aircraft. Border Patrol and CBP have flown aircraft along the border for decades and began using unmanned aircraft systems in the mid-2000s. In open country, a view from above covers ground quickly and shows what a vehicle on the road cannot see.

Mobile equipment. Vehicle-mounted cameras and radar let agents move technology to where traffic shifts, which matters because crossings move when enforcement presses somewhere else.

Every one of these has the same limit, and it is the most important sentence in this post. A sensor sees, and a person responds. A tower that detects a crossing is worth nothing if no agent can get there. Sensors also produce false alarms from animals, weather and passing vehicles, so someone has to review the alerts. The technology multiplies what agents can do, but it does not replace them.

The ports: scanners and the problem of volume

Between the ports of entry, the issue is spotting people and vehicles in open ground. At the ports, the problem is volume. Millions of vehicles and enormous quantities of cargo cross each year, and most of it is legitimate trade that a family-run business depends on. The officers must find the small part that is not, without strangling the legal flow.

The main tool is what CBP calls non-intrusive inspection, or NII. These are large imaging machines, using X-rays or gamma rays, that let an officer look inside a truck, a trailer or a passenger vehicle without taking it apart. Radiation portal monitors check vehicles for nuclear and radiological material. Officers also use canine teams, license plate readers and advance information about cargo to decide what to look at.

Scanners are powerful but not magic. They cost money, they need trained operators, they slow traffic if misused, and they do not catch everything. A dense load of legal goods can hide contraband, and image interpretation takes skill. The core trade-off is a matter of throughput versus certainty: scan everything and you choke trade, scan too little and you miss threats. That is a policy choice with real consequences for farmers and manufacturers who ship across the border every day, and a conservative who cares about commerce ought to take it seriously. I come back to that tension in They Say a Wall Can't Stop Fentanyl. Here Is the Whole Answer.

One alert, start to finish (a hypothetical)

To see how the pieces fit, follow a made-up alert from beginning to end. Picture a remote valley with a ground sensor on a trail, a tower three miles away, and one agent on patrol nine miles off by dirt road.

The sensor trips at night. That alone tells a control room only that something moved. The tower swings its camera toward the sensor's location, and an operator watches a thermal image. Here the technology earns its first dollar: the operator can tell a group of people from a herd of cattle in seconds, which saves the agent a long drive for nothing. The operator radios the agent with a location, a count and a direction of travel. The agent drives to intercept.

Now ask where the plan can fail. The sensor battery might be dead. The tower might be down for weather. The operator might be watching forty other screens. The agent might be on another call, or the road might be washed out. Every link has to work for the system to work, and the chain is only as strong as the weakest one. Add a barrier in the middle of that valley and the chain gets more time to work, which is the connection between this post and the one on barriers.

That story also shows what the critics of "technology instead of walls" miss, and what the critics of "walls instead of technology" miss. Detection without delay gives you a good video of people walking away. Delay without detection gives you a barrier nobody is watching. You need both, plus a human being at the end of the road.

A word about privacy and data

Cameras that watch the border also watch the people who live near it, and that includes American citizens, ranchers, and small towns in border counties. A fair conservative position takes this seriously. The Fourth Amendment protects people in their persons, houses, papers and effects against unreasonable searches, and a tower that records everything within its range collects information about everyone in view, not only people crossing illegally.

Congress and the agencies have to answer the plain questions: what is recorded, how long is it kept, who can look at it, and what rules limit its use? Those answers belong in public statutes and agency policy, not only in internal manuals. A security measure that the public trusts will last, and one that looks like a dragnet will draw lawsuits and lose funding. If you want the technology half of border security to survive the next change in administration, it has to come with limits that people can read. You can check how your own representative voted on border technology and privacy provisions through the votes page.

The checkpoints and the law

Not all border enforcement happens at the line. The Border Patrol runs checkpoints on roads inside the country, and the Supreme Court has addressed them directly. In United States v. Martinez-Fuerte (1976), the Court upheld fixed immigration checkpoints, including the one near San Clemente, California, ruling that brief stops for questioning of vehicle occupants at a permanent checkpoint did not require individualized suspicion. The Court reasoned that the public interest in stopping the flow of unlawful entrants was great, and that the intrusion of a short stop at a known, regular location was limited.

The Court drew a sharp line between fixed checkpoints and roving patrols. In United States v. Brignoni-Ponce (1975), it held that roving patrols near the border must have reasonable suspicion before they stop a vehicle, and that a person's apparent ethnic background alone is not enough. And in Almeida-Sanchez v. United States (1973), the Court held that a roving patrol search of a car without probable cause or consent, miles from the border, violated the Fourth Amendment.

That is the structure: more room for fixed, predictable checkpoints, less for random stops. It is a good model for a country that wants security and also wants the Constitution to mean something.

The 100-mile rule

You may have heard that Border Patrol agents have special power within 100 miles of the border. The truth is narrower than the rumor and wider than the Constitution's text might suggest.

Section 287(a)(3) of the Immigration and Nationality Act, codified at 8 U.S.C. 1357(a)(3), lets immigration officers board and search for aliens on vehicles, trains and vessels, and enter private land (but not dwellings) without a warrant, "within a reasonable distance from any external boundary of the United States." A regulation, 8 CFR 287.1, defines that reasonable distance as 100 air miles from the border.

Here is what that does and does not mean.

1. It lets agents operate in that zone, set up checkpoints, and board certain conveyances to look for people unlawfully present.
2. It does not suspend the Fourth Amendment. Agents still need reasonable suspicion for roving stops, and the Court's decisions above still apply.
3. It does not allow entry into a home without a warrant or consent. The statute itself carves out dwellings.
4. It covers a large share of the American population, because many big cities and several entire states lie within 100 miles of a border or coastline, and that fact makes people uneasy. The unease is reasonable. The fix, if you want one, is for Congress to revise the statute, and that is a debate worth having openly.

Conservatives who care about both security and the Fourth Amendment should hold both positions without embarrassment. We can back Border Patrol checkpoints that the Supreme Court has upheld and still demand that agents obey the limits that Court set. A government that respects its own limits is one that earns the public's help.

What each tool can and cannot do

Here is the plain scorecard.

• Towers and radar detect and classify movement across wide terrain. They cannot apprehend, and they need agents and a road.
• Ground sensors give cheap coverage in remote areas. They produce false alarms and need review.
• Aircraft give speed and perspective. They depend on weather, crews and funding.
• NII scanners see inside vehicles and cargo at ports. They cost a lot, need trained operators, and cannot scan everything without slowing trade.
• Checkpoints catch people and contraband moving inland. They are legal when fixed and brief, and a poor tool when they turn into random stops.
• Barriers delay and channel, as I discuss in What a Border Barrier Actually Does, and What Honest Critics Get Right.

Notice that none of these works alone. That is by design, and it is also the reason the border debate frustrates people: the right plan is a stack of imperfect tools, each covering the weakness of another. For the broader map of how people and goods actually enter the country, see The Four Doors Into America and Who Holds the Keys to Each.

The conservative test for technology spending

Because the SBInet story is a warning, I would apply four tests to any technology request, whichever party makes it.

1. Does it have a field test with real agents? Pilot first, then buy. A prototype that has not met the dust and the heat of a real border is a promise, not a product.
2. Who responds? A detection tool with no agents and no road behind it is a camera pointed at a problem.
3. What does it cost to run and maintain? The purchase price is the beginning of the bill. Ask for ten-year operating costs.
4. Who is accountable? Look for a contract with milestones the government can enforce, with real consequences when they are missed. The Government Accountability Office has issued many reports on these programs, and you can read them yourself on gao.gov.

Those are ordinary rules from the family budget: try before you buy, count the upkeep, and keep the receipt. The same people who would never buy a pickup without a test drive somehow accept billion-dollar systems on faith. Let us not do that.

The bottom line

• Border security is a stack of tools: barriers, roads, towers, sensors, aircraft, scanners, checkpoints and agents. No single one does the job, and each covers a weakness of the others.
• SBInet, the "virtual fence," was canceled by DHS on January 14, 2011. The lesson is to buy proven tools, test them in the field and demand accountability for the money.
• Fixed checkpoints are lawful under United States v. Martinez-Fuerte (1976), but roving stops need reasonable suspicion, and the 100-mile rule does not suspend the Fourth Amendment.
• Ask three questions of any proposal: does it work in the field, who responds when it sounds the alarm, and what does it cost to run for ten years?
• Read the Government Accountability Office's border technology reports at gao.gov before you repeat a claim about what a system can do.