I was please to speak with Lt. Gen. David Deptula (ret.) on the Acquisition Talk podcast. He is currently Dean at the Mitchell Institute for Aerospace Studies and, as I learned in the episode, he was the only Air Force pilot to be mission ready to fly an F-15 all the way from Lieutenant to Lt. General. He also helped plan the air campaign in Desert Storm and was the first Deputy Chief of Staff for ISR including unmanned aircraft.
The episode starts with a discussion on his study of mosaic warfare, or the concept of disaggregating functions into simpler systems which are tied together in a broader operational network. These “tiles” will work alongside complex multi-mission platforms already fielded, complementing their capabilities with lower-cost solutions that are attritable. Dave dispels myths about mosaic warfare, such as resilience problems stemming from the need to be connected all the time in a single architecture. Mosaic is about building in adaptation and resiliency.
The conversation moves onto Dr. William Roper’s vision for a digital century series aircraft program, reminiscent of the experimentation of several diverse fighter aircraft during the 1950s. Dave finds the allusion unfortunate because he doesn’t expect the Air Force to procure large batches of different aircraft. However, he agrees with the principles of incrementally prototyping several designs and continuously maturing them, even if they aren’t procured. He also agrees with Roper’s decision to raise the Next Generation Air Dominance (NGAD) program into a Program Executive Office (PEO) that can pursue multiple concepts with gradual integration, and to this end supports a mission-funded budget account.
The episode also features discussions on digital engineering, the combat cloud, why the military is the as conservative as the catholic church, design as a process of discovery, how tactical decisions centralized between Desert Storm and Afghanistan, the questions of mobilization and resourcing the strategy, and ways to think about the F-15EX purchase.
Podcast annotations.
Here’s a story about the Analysis of Alternatives (AoA) process that’s hard to understate in terms of importance:
They’re not analyses of alternative effects they’re analyses of alternative widgets. I’ll tell you a quick story. A decade an a half ago, I attempted to initiate an Analysis of Alternatives for GMTI, in other words the ability to detect, track, and relate moving ground targets from an airborne platform. JSTARS is the conventional way we do that, but there are many alternatives. Sensors from space, netted sensors, fractionated systems from a variety of different platforms in the air, large scale sensors in subspace.
Make a long story short, I’m aiming to look at the different alternatives to achieve the effect of being able to detect moving surface targets. So I retired, left the position, and the powers that be changed title of it to the JSTARS AoA. Well how do you have an Analysis of Alternatives named after a solution of the system you’re trying to replace? To the people that did that, they said, ‘OK we need the evolution of JSTARS.’ No you don’t…
Here we are a decade and a half later, and the Air Force is finally got it, because they’re looking at Advanced Battle Management System, which is more of a description of a set of capabilities and outcomes than it is a particular way to do it.
ABMS will start being a large dollar program only in the FY 2021 budget. After all, if all the stars align for a program in 2019, the fast-track budget process will get you money starting fiscal year 2021. But the bigger point is Dave’s emphasis on cost-per-desired-effect rather than cost-per-widget. As the JSTARS AoA exemplifies, the current process seems to fixate people’s attention on logical follow-ons to a current system rather than really exploring and evolving concepts.
Dave says the AoA is a “set piece process” ahead of program approval and funding requests. The AoA happens first, then a program development is initiated with procurement and operations fully planned for. The information necessary to decide is assumed to be apparent up front. There is little room for continuous feedback of information and prototyping back into the process. And so, with most of our information based on legacy platforms, the logical conclusion is sustaining advancements on what has been done before.
Realization of this issue is in part why the Air Force changed the F-35 follow-on NGAD program into a PEO that may pursue a variety of prototype and component designs. Yet the program office and not the PEO is the fundamental unit for budget requests. This means there is still a lack of flexibility to shift resources and evolve concepts.
I recommend a mission-funded account that would allow different components or systems to be started, ramped up, or canceled based on updated information. That would help bring some nonlinearity into the system and avoid locking in decisions based on preconceptions. Dave agreed. “I love your idea of a mission funded account.” He explained:
In the past, you’ve got to design the whole enchilada from the beginning. Generally, what happens and why we end up with delays is that there is discovery as one proceeds with the development of program. And then people get taken to task because ‘wait a second, you said you’re going to do this, and then you did this, and there was a delay here.’ Why was there a delay? Because there was discovery. So what you lay out is something that needs to be explored. Maybe we don’t design the whole enchilada up front.
One of the hardest parts about aircraft design is not so much in the physical characteristics of airframes and engines, but in the integration with software. Certainly integrating electronics and software into aircraft has been a primary problem for at least 50 years. The F-16’s design could only have worked with the fly-by-wire controls. Robert Perry noted that while European firms could develop airframes much more cost-effectively than the United States back in the 1960s, they struggled just as much integrated complex electronics. Actually, electronic systems were a major part of the rationale for requiring fully integrated designs upfront, as historian Elliot Converse III found:
The standard practice continued to be that an aircraft’s subsystems were developed independently and, if necessary, modified for compatibility with the airframe. But engines and other electronically driven subsystems such as fire control and navigation were becoming increasingly complex and normally required considerably more time to develop than the basic airframe.
However, the advancement of modular design and standard interfaces will hopefully permit a movement back to the older practice of combinatorial innovation, which is now observed in the boom of APIs in the commercial economy.
In my mind, a key enabler is likely to be enterprise tools. Despite tremendous increases in computing power for digital engineering, enabling technologies like wind tunnels will still be necessary to determining proper specifications particularly in the transsonic and hypersonic regimes. For software, cloud and automated testing tools will be important. These enterprise tools don’t fit well in current DOD practices that cater to programs with custom-built solutions.
I’d like to thank Lt. Gen David Deptula (ret.) for joining me. See more of his great work at the Mitchell Institute, including force sizing methods in The Force We Need and a great empirical paper on USAF inventory from 1950-2017, Arsenal of Airpower. He also writes relatively frequently at Breaking Defense, including an interesting piece that argues low bomber availability rates are a sign of high demand more than poor management. Here is Dave on the NGAD program. And here is Dave speaking with Vago Muridian about Nuclear C2 and F-15s. Be sure to listen to him on the Chrome 360 podcast discussing his experience in Desert Storm, highly recommended.
Leave a Reply