Richard Danzig joined me on a joint episode of the Acquisition Talk and China Talk podcasts to discuss US-China relations and military innovation. Richard is a Senior Fellow at Johns Hopkins APL, a former Secretary of the Navy, and much more than that. We traverse a number of subjects, including:
- How the risk of war with China is reflected in trade policy
- The problems regulators face in high-tech industries
- Views on growing the US Navy to 500 ships
- How US prime contracts differ from state-owned enterprises
- Whether the Chinese are more risk-tolerant than the US
Moving institutions
One of the most difficult aspects of innovation is when an old institution has to be brought along with it. It’s not just the customs and preferences of the people who have obtained success with the old methods, but the rules and regulations written for a time long past. In the market, new companies can come in and test the boundaries of what’s socially and legally acceptable, like Uber. However, the “Uber” of the military would’ve needed prior approval from dozens of offices, making a daring foray into innovation all the harder. Richard as a couple of interesting stories related to that. Here’s one:
If you look at the commissioning ceremony of the Navy’s unmanned ship, it’s an impressive event. I looked at the picture a fair while and called up the program manager and said to them… “What is that circular donut shape thing I see hanging in the middle there is that a life preserver to which he said yes. And I said, why does an unmanned ship have a life preserve?”
His answer was, “Yes. We were aware that this was strange, but there’s a coast guard regulation that requires that every ship have a life preserver.” And I said, “did you try and get a waiver?” He said, “Yes, we did. But we quickly concluded it’d be easier to hang a life preserver on the ship and get a waiver.” So I think that story is indicative of the challenges.
Are the Chinese more tolerant of innovation?
While the US Government continues to struggle to fund AI/ML projects, the Chinese government has already been investing at scale in that area for years, not to mention quantum, blockchain, 5G, and so forth. Looking over the last 30 years, China has gone through a transformation unrivaled in history. This perhaps allows their leaders to be far more risk tolerant of the disruption that comes along with these technologies.
In the end, my belief is that a free society will outpace a closed one in terms of progress. Richard cautiously agrees, but qualifies that’s “because it’s what I want to believe — it’s so deep in our own culture.” But one aspect of centralization is that it is short-term efficient, and so my fear is that China may outpace the US in critical technologies in the next decade or two. As Keynes said, “In the long run, we’re all dead.”
Both Richard and Jordan push back on me. ” I encourage you not to idealize the Chinese.” I think those are wise words that should be dropped more often during great power competition debates. Listen to the podcast for a lot more on this.
How adversaries reflect each other
One aspect of this idealization for people in the US government — focusing on China’s strengths and not its weaknesses — is that they want to look more like China. I was at an MIT talk on cybersecurity, and a nice woman from the government said that the US is vulnerable because it isn’t centralized like China. She had some sweeping plan to bring all controls into one place. I pushed back using obvious economic arguments, but she didn’t want to hear it. “Back in WWII,” she said, “we only won because Roosevelt got rid of markets and centrally planned all production.” I suppose we all rewrite history to fit our narratives — myself included.
The point of this section is that adversaries often come to look like each other. For example, in the 1950s the DoD created a “Missile Czar” position. It got rid of democratic board/committee structures and moved to single executives (e.g., R&D Board to DDR&E). In the 1960s, the DoD literally adopted Soviet 5-year planning methods in the PPBS, which we still have today! And what happened when the DoD felt behind on AI/ML? Ash Carter wanted to stand up an “AI Czar,” which is today’s JAIC.
Here is Richard pointing to this phenomenon:
One of the interesting things is how, adversaries convert each other in their own images. they become more so we adopted the Soviet five-year plans, et cetera. There are many examples of that, but the Chinese military is becoming obviously more like us and we may become more like the Chinese and are being more directive, developing industrial policies, et cetera.
Long-term planning fails
Speaking of the DoD’s central planning processes, here’s Richard:
Somebody said to me, have you ever tried to write fiction yourself and my answer to that most? Yes, of course. The Navy five-year plan.
That 5-year plan in part reflects the Navy’s 30-year shipbuilding plan! The nice thing about these long range plans is that it creates a solid baseline which oversight agencies can measure against. And moreover, the baseline is agnostic to context — APB cost growth is measured exactly the same for aircraft, ships, IT systems, etc. Observers don’t need to know all the complications of technology, CONOPS, alternatives, etc., that were baked into the approved plan. Here’s Richard on the PPBS baselines:
And we’ll measure programs against that. And that as a manager has big attraction because now I have some kind of anvil against which I can hammer out programs or some measuring stick by which I can assess competing things well, but it is illusory.
Richard explains that when it comes to innovation, we are “driving in the dark.” Prediction is very hard, especially about the future. So all that long-term planning is either wasted effort or locks in projects riddled with errors. Richard argues we should have a high discount rate (perhaps even a hyperbolic discount rate?) because our predicted benefits years into the future are not only less useful than if we had them now, but in fact they may never materialize because of uncertainty!
Building for the near term
I’m much more inclined to put an emphasis on near term stuff, but to build it in ways that are adaptable have maximum kind of adaptability to different circumstances. Let me build submarines where I can adapt the firing to in ways that, can be used for conventional weapons or nuclear weapons.
I’d like to push back on Richard here a bit. I do not prefer adding multi-mission flexibility into one system. Indeed, I think this is the status quo. Create a hugely expensive and elegant platform that “does it all,” allowing us to retire several systems and replace it with one.
The Littoral Combat Ship was an example of multi-mission modularity that failed. I think the problem with that is that the design process rapidly increases in complexity. LCS didn’t use existing standards that worked, but rather tried to invent them into a system of systems. I see modularity and standards are part of an evolutionary process rather than a “big bang” design process.
In my mind, the better way to build for the near term is to disaggregate missions into smaller or single mission platforms in a proliferated networked architecture (e.g., something like mosaic warfare, but I think it reflects an evolution of the diversity of systems the US military had in WWII and the Korean War). This strategy has the promise of dramatically reducing production and sustainment costs, which take up the vast bulk of lifecycle costs.
I doubt Richard and I are as far away on this matter as I make it appear. For example, the Valkyrie “loyal wingman” drone is part of that disaggregated/networked architecture, but it will also has modular payloads.
R&D as a reason not to adopt
Jordan and I often come back to this idea of basic vs applied research, and where on the margin US policy should move. Richard says any balanced diet needs both aspects, but the US diet has way too much R&D and not enough application. Here he is:
I’ve increasingly come to the view that the R&D can be a substitute for the real change. It’s a good thing to do. I like it. But what people tend to do is if I’m in the military, it’s easy for me to say, we ought to have more R and D we just throw money at this.
That doesn’t force me to change, to give up my platform or whatever somebody is out there working on it, and it’s measurable. And there’s an R and D establishment that likes it because they’re doing it. They’re funded by it. And meantime, I’m not facing the hard question, which is. How much do I alter my methods of operation and really, for example, embrace unmanned ships.
By the way, appropriators cut the Navy’s FY21 budget request for demonstrations of small and medium unmanned surface vessels from $464 million down to $149 million.
Thanks Richard!
I’d like to thank Richard Danzig for joining Jordan and myself on our podcasts. Please read his excellent paper, Driving in the Dark: Ten Propositions About Prediction and National Security. You can read several of his pieces over at Muck Rack and CNAS, including “Artificial Intelligence, International Competition, and the Balance of Power.” Listen to interviews with Richard on CyberLaw podcast, Public Policy podcast, Silver Bullet Security podcast, and Defense & Aerospace report.
Leave a Reply