Notes from Will Roper's talk to students: Tons of great insights!

Here are some notes of mine from watching Air Force acquisition chief Will Roper’s talk given to Steve Blank, Joe Felter, and Raj Shah’s class at Stanford (highly recommended!):

Top notes:

Digital engineering is a harbinger of a new type of acquisition that can save DoD. Build fast, cheaper, and less risk. T-7 hit quality curve on first unit that normally would take 100 units – allows to get way down learning curve without the same cumulative quantities. If build faster and more iteratively, don’t need mass production. Opens door to shift more cash flow into design, rather than O&S. This is at core of how US will compete against China. Digital engineering first thing Roper’s seen that allows profit and cash flow opportunities back into design, rather than having that in production and O&S. “DoD is like a hospital with a very small pediatric wing, and very large geriatric wing.”
- (Eric: in this digital world of engineering, blockchain would be a huge market/transaction cost leverage by having canonical designs, combinatorial innovation with automated payments, completely new incentive and organizational structures, etc.)
Roper humble on his ability to understand how important or implications of tech like synthetic biology. DoD is not prepared for a biological competition. Whose job is that in government? DoD builds its military around Newton, Kepler, Maxwell, Schrodinger’s equations.
“Bureaucracies try to stabilize their policies, while tech tries to disrupt them.”
How to get new companies to scale and new tech to replace legacy platforms, get away from small contracts and get a program of record? “Steve, that’s not a $1 million question but a $100 billion question.” Something changed in 1970s, that slowed down Air Force acquisition. Something didn’t break, but a change in tech. In late 1970s, aircraft and other systems changed from a system to a system of systems. It was about integration. With high performance radar, stealth, weapons, etc. Integration of getting computers on subsystems was difficult. In order to get past that, economies of scale and more of the same platform, became the paradigm. DoD still stuck in that, where tech changes occur infrequently, and DoD tries to buy a lot of the same platform. Mass production is a bad way to drive down cost in a program, because if you spend all your money on one platform, you keep it and incrementally upgrade. 70% of funding is in O&S. AF has oldest fleet in history, over 23 years old on average (Eric: CSBA finds older right? 35 for fighters, 53 for tankers… But this seems to be Roper’s thesis on why progress in weapons slowed down.)

Other notes:

Starts with supposition that the future can be predicted. Not written down, but foundational to Pentagon process. When you say, you don’t know what will happen in 2030 for tech or threats, how do you get programs? Some say its AI, other say quantum, other biotech, and more still. Its exciting and there’s no telling what tech will lead to the next industrial revolution (maybe they all will have S-curves).
How to use National Guard and Reserve to keep connected to commercial talent. Also need a revolving door. “Revolving door is a pejorative term in Washington, and we have to make it superlative. Revolving door in how we will win.
Pentagon in time of crisis, trying to figure out its role where it isn’t major funder of R&D. Pentagon challenge is to reboot itself, get rid of cold war model and need to start bringing in tech and instantiating it. Build partnerships as a central paradigm for moving tech forward.
Tech is a battlefield, and Pentagon is not ready for that.
The acquisition system must be agile. Speed requires a change in value and culture. The harder part is how military works in broader ecosystem, bc DoD always tries to predict future, plan, and then send contracts to incumbent primes. Landscape has changes, DoD hasn’t. To use a separate set of rules so DoD can do business with 4/5^th of R&D going on in commercial is exceptionally hard.
It is in the national interest to accelerate tech companies as an end in itself, not just deploying capability to warfighter. If a market doesn’t happen in the US first, it will start elsewhere, and US might not have access.
DoD has process problem with architectures and data standards (i.e., ABMS). Chain of command breaks things up into divisions and doesn’t have a good way of interoperability. (Eric: the DoD budget is stovepiped into platforms as well, not sure COCOMS are like that though.)
“We’ve become a software company in our own right very quickly.” Using enterprise tools like cloud and platform to start building military internet of things. DoD is far behind the times, but can catch up.
DoD is good at building legacy hardware, but haven’t learned digital/software tech.
US can’t beat China at scale, only speed and agility. Need urgency in organization.
DoD didn’t stabilize drone market and lost it. Drones that started in US moved out, DJI became big player and Roper had to sign off on DJI purchases every day, and now drones disrupting the US. DoD should have leaned in, and help certify new eVTOLs and accelerate the market. Benefits to industry boomerang back to DoD. How does DoD, which is like a mini country in the US, bring its assets to bear to support/accelerate experimentation in new tech.
It’s better to make the enemy react to you when you can’t predict future threats, so whoever is more agile (OODA loop) wins. Agility is the only delineating advantage in this era. An enemy can copy whatever we do. We have to be faster. Whenever Roper shows operators something new, they immediately have a requirement for it, but they didn’t know it beforehand. Need to get away from linear process that looks more like a cycle or loop.
Education reform is Roper’s litmus test. (Eric: like Rickover, who really focused on education in 50s and beyond).
Q: What role do the traditional primes play? Military needs to go more heavily into deep tech, take risks that private often wont. With the primes, they can play same role using their small business funds. So many things that plug into defense programs that aren’t defense-unqiue. Primes’ ecosystem is less defense-unique than they are to govt. Roper wants to incentivize making a defense program its own tech ecosystem. (Eric: but how to maintain dual-use?)
Large bets in STRATFI is something the size of a Series A. (Eric: I suppose they’re targeting about $10M… where’s the $30M and $50+M?) Doesn’t want DoD to have equity in companies, or does he? Helping Army clone AFVentures. “The learning comes in the doing, so you got to start doing it.”
Need to dispel rumor that VCs/startups don’t want to work government. 80 percent of investment goes to software because lower risk and faster to market. Government can be more in patient capital. Government contract to a company should be a signal of a quality company, give private capital confidence they can invest alongside. In 2020, AF Ventures had 4:1 matching VC:AF, better than 3:1 last year.
(Eric: I think WRT ABMS/internet of things, Roper’s talking about STITCHES model of interoperability. Need a map/model of network interfaces, then can autogenerate code.)
“If you have any inclination, follow it. You can make a huge difference. Even in four years.”