The Bunker, delivered to our subscribers Wednesdays at 7 a.m., is a newsletter from the desk of National Security Analyst Mark Thompson. Sign up here to receive it first thing, or check back Wednesday afternoon for the online version.
We’re going to pretend we’re like the Pentagon this week. First, we’re going to hype a shiny new technology, then we’ll dive into our inability to keep what we already have shipshape.
COMPUTER—5, AIR FORCE PILOT—0
Savvy supersonic silicon
The lucky service winner this week is the Air Force. That’s true even though its pilot “lost” in virtual dogfights staged by the Pentagon eggheads at the Defense Advanced Research Projects Agency against an AI competitor. AI—artificial intelligence—is, along with hypersonic weapons, one of the Pentagon’s hottest technologies. They could give the U.S. military a vital edge against prospective foes, or they could be a money pit. If history is any guide, they’ll probably muddle up somewhere in the middle.
DARPA’s AlphaDogfight Trials finale took place August 20 when an AI algorithm (and no, “algorithm” is not named for the former vice president, despite his pioneering political role in developing the Internet) beat an F-16 pilot sitting in a simulator in all five dogfight scenarios. “It’s a giant leap,” said Lieutenant Colonel Justin Mock, an Air Force F-16 pilot working with DARPA.
The victor was AI developer Heron Systems; the loser was callsign “Banger,” an otherwise unnamed Air Force pilot with more than 2,000 hours of F-16 flying time, and a recent graduate of the Air Force Weapons School’s F-16 Weapons Instructor Course. In other words, one of the military’s best.
Banger cited flight restrictions drilled into Air Force pilots that the AI system ignored. “The standard things we do as fighter pilots aren’t working,” Banger acknowledged. The AI craft, callsign “Falco,” was able to react more quickly than his human enemy, whose brain cells can’t process John Boyd’s famous OODA loop—observe, orient, decide, act—as speedily as AI silicon wafers can.
There’s good news and bad news associated with the test. The good news is that such AI software could someday be the “brains” of unmanned warplanes, reducing their dependence on human ground operators. But “a fully autonomous Heron flying the entire airplane is still quite a ways off,” said Air Force Colonel Dan Javorsek, chief of DARPA’s Air Combat Evolution program. It also could be used to handle tactical decision-making aboard warplanes with human pilots, freeing them up to make strategic decisions.
The bad news is that the aerial dogfights are a relic of a time before long-range radars and missiles and are of scant military utility. They may also represent a misplaced focus when the U.S. military has been unable to prevail against insurgents on the ground who don’t fly. But the Military-Industrial Complex has never let such “Top Gun” concerns stand in the way of the bottom line.
THE NAVY’S SHIPYARD WOES
The service’s problems aren’t confined to the waves
Back in 2017, the Navy’s skimpy training of its sailors led to a pair of at-sea collisions that killed 17 of its sailors. Last week, a grim Government Accountability Office report made clear that similar problems affect the service’s shipyards. Maintenance delays are rampant, despite a $2.8 billion investment over the past five years to improve the performance of the yards in Hawaii, Maine, Virginia, and Washington state.
“The Navy’s four shipyards completed 38 of 51 (75 percent) maintenance periods late for aircraft carriers and submarines with planned completion dates in fiscal years 2015 through 2019, for a combined total of 7,424 days of maintenance delay,” the GAO found (PDF). “For each maintenance period completed late, the shipyards averaged 113 days late for aircraft carriers and 225 days late for submarines.” Idle time—like your car sitting in your mechanic’s garage because he’s too busy with other customers—soared for attack subs from 100 days in 2015 to 1,019 days last year—a 919% increase.
The report cited a lack of shipyard workers and the resulting “excessive use of overtime” to try to make up for “unplanned work that is identified after maintenance planning is finished,” the GAO said. Production shops at all four shipyards are working beyond their capacity, Navy officials acknowledge. “Working overtime can make staff less efficient,” the agency added. “Shipyard officials stated that too much overtime can also increase the likelihood of accidents and rework, among other things.”
Part of the problem is the military’s can-do culture. When it comes to the Navy, sometimes admirals have to simply say “no.” In the kind of information that rarely makes its way into public reports, the GAO said (PDF ) five years ago that the service was able to meet only 44% (PDF) “of the requests from combatant commanders around the world for Navy forces to support ongoing operations and theater security cooperation efforts” in 2015. Cutting such requests in half would be a good place to start.
But that would make too much sense. The Navy prefers to spend billions developing the next generation of fighter aircraft to be used aboard aircraft carriers, even as too many of the flattops are sidelined for repairs.
After years of reading how the U.S. military has been developing so-called “exoskeletons” to turn individual grunts into super-soldiers, it’s nice to read the truth at last: “Why Military Exoskeletons Will Remain Science Fiction,” Vikram Mittal wrote August 17 at Forbes.com. “The military industrial complex has a massive graveyard of exoskeleton projects,” he notes. “Time and time again, the defense community has failed to produce functional exoskeletons due to technical challenges.”
You can almost hear the grousing from inside the Pentagon’s silver-bullet shoppes: “What an amateur! He’s only a combat vet with a BS in aeronautics from Caltech, an MS in engineering sciences from Oxford, [after 50 years The Bunker takes pride in a true Oxford comma] and a PhD in mechanical engineering from MIT. Sure, he now teaches engineering at West Point. But what does he know?”
One thing he knows is that he’s unlikely to find employment in the defense industry. Now, if he’d only train his sights on laser weapons, battlefield nuclear reactors, and national missile defenses…
Click here for the full story, on the POGO website.