When Gen. John Campbell, the Army's vice chief of staff, appeared last year at a budget hearing on Capitol Hill, he cited his son's experiences as a soldier in Afghanistan to answer a senator's tough questions about a troubled intelligence technology system.
This week, after an inquiry by The Associated Press, the Army acknowledged that Campbell misspoke about his son's unit, omitting some key facts as he sought to defend a $4 billion system that critics say has not worked as promised.
Campbell faces another Senate hearing Thursday morning, this one on his nomination to lead U.S. forces in Afghanistan. He may get another chance to discuss the Distributed Common Ground System, known as DCGS-A (pronounced DEE-cigs-ay). The network of crash-prone software, sensors and databases was supposed to allow troops to process and integrate intelligence from a variety of sources, from electronic intercepts to overhead imagery to spy reports.
The need is greater than ever, since gathering and making sense of intelligence in Afghanistan will remain a priority even as U.S. troops draw down.
Army leaders, including Campbell and his boss, Army chief of staff Gen. Ray Odierno, have circled their wagons around DCGS-A despite a series of independent government reports that have pointed to significant weaknesses.
When Rep. Duncan Hunter, R-Calif., a member of the House Armed Services Committee, visited troops in eastern Afghanistan last year, "DSGS was shut down in the corner, piled with books and papers," he said.
Equally troubling to critics is how the Army has made it difficult for commanders to use an off-the-shelf commercial product that soldiers say is more workable and user-friendly than DCGS-A, even though the commercial system has been embraced by the Marines, special operations forces, the CIA and other government agencies.
"DCGS folks promised a solution three years ago, and they have yet to deliver," said Col. Peter Newell, who retired last year after heading the Army's Rapid Equipping Force.
Army officials acknowledge problems with DCGS-A. In a statement, spokesman Matthew Bourke said the Army is working to improve the system in its next generation, which is being put out for bids next year.
DCGS-A was first developed a decade ago, but the spotlight on its shortcomings grew brighter in 2010, when Lt. Gen. Michael Flynn, then the top military intelligence officer in Afghanistan, said in a memo that "intelligence analysts in theater currently do not have the tools required to fully analyze the tremendous amounts of information currently available."
Flynn made an urgent request for a "theater-wide, Web-based analytical platform" that sounded a lot like a product offered by a Silicon Valley startup called Palantir, which grew out of antifraud technology developed by PayPal and was valued in December at $9 billion.
Yet over the last four years, records show, Army leaders have made it difficult for some commanders to purchase Palantir.
Army units that have managed to obtain Palantir report that it has saved lives in Afghanistan by helping to map insurgent activity and bomb networks in ways the Army system could not. It is also far cheaper: A 2013 Government Accountability Office report estimated that the Pentagon had spent about $35 million in recent years to equip the Marines and some Army units with Palantir, compared with $4 billion for DCGS-A.
Palantir can merge disparate data sets - cellphone calls, fingerprint and DNA records, photos, bomb incident reports - and array them on a map in seconds. DCGS-A's workstations employ a mapping program that is much more difficult to master, in a system that does not allow seamless data fusion. When soldiers update a file in Palantir, that file becomes visible to every Army Palantir user, which often is not the case across the DCGS-A network.
Last April, Sen. Claire McCaskill, D-Mo., confronted Campbell with DCGS-A's litany of troubles, including that the Army's testing lab in 2012 found the system "not operationally effective, not operationally suitable and not survivable."
Campbell said Palantir does only a small portion of what DCGS-A is supposed to do, though he acknowledged that it is easier to work with. The Army's system, he said, "saves lives" and has access to more intelligence than Palantir's software does.
He added: "My son is a soldier in the 82nd. He's a specialist. He deployed to Afghanistan." Campbell said his son was in "one of the units that asked for DCGS - or his brigade did, not himself."
Emails obtained by the AP show, however, that the younger Campbell's unit - the 82nd Airborne Division's 1st Brigade Combat Team - found the Army system inadequate and requested Palantir after six men died in two roadside bombs in April and May 2012.
On May 12, 2012, Col. Mark Stock, the brigade commander, signed an "urgent needs" request for Palantir, citing "major capability gaps in the division's existing intelligence software architecture."
The political climate was not good. Two weeks later, in an email responding to a different unit's request for Palantir, Newell wrote, "While I don't disagree with your need, I cannot buy Palantir anymore without involving the senior leadership of the Army, and they are very resistant."
In an interview, Newell said Army leaders did not prevent him from equipping units with Palantir, but they made it difficult. "I still don't know what the big threat of Palantir was. It was baffling to me to see the lengths they were willing to go."
In a statement, the Army said Campbell "misspoke," and meant to say that his son's unit had requested Palantir, not DCGS-A.
The unit did not get Palantir before it departed in September. The Army blamed logistical hurdles. Newell said the request came too late.
In January, Campbell's son's brigade - he was no longer in it by then - deployed on a training exercise in Fort Polk, Louisiana, and again experienced difficulties with the Army's intelligence system.