No experts, as far as I can tell, believe that Donald Trump can 1) get Congress to approve funding for the construction of a wall that runs the length of the border with Mexico, 2) get the wall built to a height of some 50 feet across terrain and along boundaries that are often tough to negotiate and 3) get the Mexican government to reimburse us for our costs. Any one of those three things, by itself, would be nearly impossible to achieve, much less all three in concert. Yet, as recently as last week, Trump insisted that he stood by the plan as formulated. A big wall. The whole border. Mexico pays.

On Monday, some of Trump allies had started to shift the way in which they talked about the wall, moving from a 50-plus foot wall made of prefabricated concrete slabs to a "virtual" wall — a wall not in the sense of the walls of your house, but the walls created by an ankle bracelet when you're under house arrest. A series of sensors and cameras that could be placed along the border to provide a system allowing for better monitoring and enforcement. Setting aside the political challenge of such a massive shift in Trump's rhetoric, it's worth assessing how feasible such a wall would be. And it's easy to do that, because the Department of Homeland Security spent hundreds of millions of dollars trying to build a virtual wall within the past decade.

It didn't work.

In February 2010, Boeing's in-house magazine, Boeing Frontiers, profiled the efforts of the company's work on the virtual fence, a project called SBInet (where SBI stood for "Secure Border Initiative"). Boeing was four years into its contract with the government to build the network, which involved the construction of  80-foot towers in the Arizona desert topped with sensors and cameras.

The general idea worked something like this illustration published by

SBInet focused on facilitating communication and information flow to border patrol agents, detecting intruders, evaluating the risk they posed and dispatching law enforcement as needed. The first phase of the project involved building nine towers along a 28-mile stretch in Arizona, with each tower including surveillance and target-acquisition radar arrays and an electrooptical/infrared camera. Between the towers, ground sensors would be embedded out of sight, relaying information back to the network. A wall — but virtual.

Boeing's public relations team explained how the 240-person SBInet team worked. (Some of the SBInet work was done by subcontractors, including Ross Perot's Perot Systems.)

Work included placement by cranes of tower foundation “wafers,” piecing tower sections together, and hoisting microwave dishes and radar equipment to the tops of the 80-foot-high (24-meter-high) towers. Teams from across Boeing conducted integration and testing, working with the project’s Tucson field office to network components, troubleshoot software, align the system and meet requirements spelled out by Customs and Border Protection.
“This is not your typical factory,” said Scot Magill, Tucson deployment production manager, who spent 29 years on the Apache helicopter program before joining SBInet.

At the beginning of the article was an odd sentence, though. "The Department of Homeland Security recently announced that the SBInet program will undergo a department-wide reassessment that will include an evaluation of the right balance of SBInet technology, physical fence and Border Patrol agents required to provide control along the southern border."

In fact, only a month after that issue of the magazine was released, the Department of Homeland Security halted work on the SBInet project. In January 2011, it was killed entirely. About $1 billion had been spent on the project.

What went wrong? In part, the project was poorly managed. The Government Accountability Office started raising concerns in 2007, a year after the contract was given to Boeing. (At that point, the first phase of the project was expected to be completed by the end of 2008.) Guidelines and time frames set by the government were vague, forcing Boeing to make things up as they went along. Testing was managed poorly and bugs cropped up faster than they were addressed. In the GAO's words, "[a]bout 1,300 SBInet defects had been found from March 2008 through July 2009, with the number of new defects identified during this time generally increasing faster than the number being fixed — a trend that is not indicative of a system that is maturing and ready for deployment." Had the money spent on SBInet been spent instead on building a wall on the border, it could have covered 73 miles, another GAO report noted. After the project was expanded from the original 28 miles, SBInet covered 53 miles in total.

The problems weren't only administrative. As The Washington Post reported when the project was tabled in 2010, SBInet was the third attempt to solve the border problem with technology. "Between 1998 and 2005," our Spencer Hsu wrote, the government "spent $429 million on earlier surveillance initiatives that were so unreliable that only 1 percent of alarms led to arrests."

An executive with a firm that specializes in long-range object detection explained the problems with the plan to SecurityInfoWatch in 2010. "One of the challenges is that while ground-based radar is great for wide-open, level areas like airport tarmacs, it's not right for that border environment, where you might have tall bushes and uneven terrain," SightLogix's John Romanowich said. "The system was also creating way too many nuisance alarms."

When CBS's "60 Minutes" looked at progress on SBInet in early 2010, it found the same problem.

"You know, when Boeing first got the contract back in 2006, they made promises that they would be able to apprehend, at least detect and apprehend 95 percent, plus or minus five percent, of all the incursions," [the GAO's Richard] Stana told [CBS News' Steve] Kroft.
Asked if that has happened, Stana told Kroft, "No. They promised camera ranges of 10 miles. They promised radar ranges without clutter."
But that didn't happen either, according to Stana.

SightLogix had been approached by a company bidding on the SBInet project, but Romanowich turned them away. "We can't do it; we're bound by the laws of physics," Romanowich said, "so you'll have to find some other company that isn't bound by the laws of physics."

The death of SBInet didn't kill interest in technological protections along the border, as the Center for Investigative Reporting noted in 2011. Other companies that made products similar to components of SBInet got new contracts even as SBInet was failing.

It's easy to see why. Border protection that doesn't require flooding the region with agents (though that's one tactic that President Obama endorsed) and which avoids the ecological and logistical problems of a wall sounds just about perfect. In practice, though, it's proven tricky to turn into reality, even within Obama's own administration.

But then, compared to getting Mexico to pay for a 50-foot wall that runs thousands of miles, SBInet sounds downright simple.