War in Taiwan: Could Emerging Tech Tip The Scales?

“Think about the resilience of your military systems at least as much as their lethality.”

We’re back for part 2 of our discussion with Mike O’Hanlon (Part one on the challenges of modeling a conflict here). Below we discuss:

• A systematic way to approach future predictions given the inevitability of technological progress;

• How and when military-civilian bureaucratic competition can promote cutting-edge research advancements;

• Why military-systems resilience is at least as important to deterrence as lethality;

• Mike’s book recommendations!

Emerging Tech Shenanigans

Let’s layer another level of complexity onto projecting US-China conflict: emerging technology. It’s one thing to go back forty years to the Falklands War and look at the present. But going into the future is another challenge that your book The Senkaku Paradox, which came out in 2019, did a really fun job of trying to explore.

Mike wrote a book in 2000 called Technological Change and the Future of Warfare where he picked a few dozen emerging technologies and tried to say, “In 2020, I think these are going to hit, and these are going to bust.” And then a few years ago you did a sequel to that where you graded yourself on how your 2000 predictions did, and tried to learn some lessons from that exercise in, looking forward another twenty years on, what the technologies you picked then, as well as a handful of new ones, would end up doing to the high-end battle space.

That was one of my favorite things I’ve read in the past few years: it had physics and intellectual honesty, and it spawned some great Googling — I didn’t even know about intercontinental railguns, for example.

What were some of the things that you ended up being right on and wrong on? And how did this personal experiment change your mental models of how technological change impacts militaries?

Mike O’Hanlon: That was very kind, and I appreciate the thoughtful reading of my work over that stretch. And to clarify, in case anybody’s at home worried they’re about to be taken down by somebody’s intercontinental railgun — the good news is that Jordan’s talking about technology development as opposed to technology deployment. I don’t think anybody actually has an intercontinental railgun right now, which is probably because we really wouldn’t know how to guide the projectile very well — it’s just a dumb piece of metal that’s accelerated through a long tube to a preposterous speed — and we found it’s actually better to let a projectile be less sophisticated and generally less fast but to have some kind of homing guidance.

The basic concept of what I tried to do, going back to 2000: I listened to the debate of the day about defense innovation, and people were getting very excited about things like computers, precision-strike weapons, laser-guided bombs and GPS-guided bombs — and they were starting to argue that this revolution of military affairs was in the offing, and that maybe war was going to become standoffish, antiseptic, autonomous, robotic. We weren’t talking that much about artificial intelligence back then, but we were talking a lot about cyber (The Terminator was two decades old by that point).

There was a lot of breathlessness, and we had seen amazing things happen in Desert Storm, especially in 1991, that made people think, “This is a moment in history where warfare could change radically, and if you don’t get ahead of that change, you’re going to be in trouble. Therefore the United States, even though it was the established superpower — and had just done very well in Desert Storm, as well as the Kosovo War — really needed to dismantle a lot of its existing capabilities and just put money into new stuff and new ideas.”

That’s fine, but it’s also a little breathless: when you dismantle things, you’re giving up real capability, and you better be confident that those real capabilities are no longer needed, and you better be confident that your new visions could actually be achieved. It’s one thing to talk about them beautifully, or put them on a PowerPoint or in a nice book; it’s something else to bring them to the test range, and then bring them to scale and incorporate them into modern military units. So I was convinced the whole debate was getting overexcited, that we were going to make some big mistakes by trying to change too fast.

I appreciated that the people who were talking about a revolution of military affairs wanted to shake up the system and make sure the services didn’t just keep buying incrementally better versions of the same kind of weapons they preferred to operate — say, fighter jets for the Air Force, etc.; there was good motivation behind the desire to shake up the debate. But it ran the risk of pushing us too far.

For example, some of the things that I was hearing people say back then that I instinctively thought were wrong were things like, “The oceans are going to become transparent to new kinds of sensors.” I remembered from graduate physics that electromagnetic radiation of any type does not penetrate seawater more than a few meters, or at most a few dozen meters, maybe 150 to 200 meters for blue and green when we go snorkeling in the Caribbean with our families. But that’s about all you’re going to get out of the oceans — except through sound waves, sonar. And we already had that deployed, so it wasn’t clear what was still to be done.

Another idea: the army wanted to build these tanks that were going to weigh only one-fourth of what tanks of the current generation — the M1, which we still have — and they were going to somehow have, what I call, the “Muhammad Ali” version of armor, which is to be so light that you fly like a butterfly, sting like a bee, and see the enemy shooting at you before the enemy gets a shot off, so you can dodge it or intercept it by having a tiny little anti-rocket rocket you can aim at the incoming weapon. And they were going to build these internal combustion engines that were ten times more efficient, and start having land vehicles drive 125 miles per hour on rugged terrain by the year 2010.

So what I said was, “This sounds wrong.” And I went back to my college physics — that was my major — and was pretty confident some of the allegations or presumptions or claims were incorrect. But then I said, “How can I analytically attack this in a more systematic way?”

And that’s when I came up with the thirty-five areas of technology. For each one, I tried to apply either core principles of physics or knowledge of the existing engineering literature to project what I thought was going to be realistic to achieve within the next one to two decades.

Consider big ships that go through the sea. What I was really struck by: our big ships in the Navy today typically go about thirty-five knots. And I kept reading this World War II history, about all these ships going twenty-eight or thirty knots. And I’m like, “Wow — we’ve had sixty years to get better, and basically our ships sail the same speed.”

I found the same thing with most aircraft. Yes, it’s true that our highest-performance aircraft can keep getting faster — but because of the efficiencies and inefficiencies of hydrodynamics through the atmosphere, you tend to want to fly an airplane at around 500 miles an hour; you don’t tend to want to fly it a lot faster. And we haven’t really come up with a big idea or a different design for an airplane that would change that. And with the internal combustion engine: yes, they keep getting better — one percent here, two percent there, making the way fuel is turned into an aerosol even more precise. But these are incremental changes in a technology that’s pretty mature.

The more I looked at a lot of technologies like that, I started to see that the laws of physics would get in the way of big breakthroughs — making them either impossible or requiring a whole different concept (the oceans, for example, were not going to become transparent) — or, I found that technology was really impressive, but it had been impressive in that area for quite a while, and it was not improving in transformative and revolutionary ways.

So I wound up concluding that, for the most part, I was a skeptic on the “revolution in military affairs” hypothesis: I did not necessarily see this sweeping potential for breakthrough across all these different tech sectors. Yes, there was potential in computers and also a couple other areas — maybe biological weapons, maybe certain kinds of directed energy — but the ecosystem as a whole was not changing incredibly fast.

Jordan Schneider: Before we get back to the computing revolution and AI’s potential impact, let’s talk about the other limiting factor (aside from the physics): bureaucracies.

You say that it’s most difficult to predict how multiple technologies might be combined, as well as how military organizations might respond to new technological opportunities that required multi-step processes to reach their potentials. Any thoughts on that and, in particular, thoughts on forecasting how well bureaucracies can evolve to take advantage of potential technological breakthroughs?

Mike O’Hanlon: One way is competition. That’s why I actually don’t mind seeing multiple military services sometimes overlapping in their interests, their portfolios.

With drone technology, for example, the Air Force was investigating and to some extent deploying unmanned aircraft in the 1990s, but didn’t really think of these platforms as ways to deliver ordnance. They had cruise missiles — which are unmanned aircraft in and of themselves — but they really didn’t want to start using Predator or Reaper or Global Hawk or what have you; they preferred to do the higher-end work, especially the weapons delivery, with manned airplanes — and that was partly a cultural thing.

It was the CIA that decided to start slinging weapons underneath these unmanned aircraft, because the CIA needed to get certain very specific things done, especially after 9/11, and also didn’t have huge budgets to build a fleet of stealth aircraft to go do it the old-fashioned way. So the competition in this case — not even involving another military service, but another part of the US government altogether — became crucial. Of course, now the Air Force is in the game, too: they can’t afford, bureaucratically or otherwise, to be the second-best at anything involving aerial attack. Once it has been demonstrated that there is a way to do this pretty effectively with uncured systems, the Air Force is going to feel some of the pressure. So that’s one-way healthy competition.

Another way is, of course, that we need civilian leadership that challenges the services. It’s usually not possible for the civilian leadership to completely overrule a military service on a matter that’s core to that its culture. But you can challenge. In fact, it can go both ways: a lot of times, good ideas come first and foremost from within the uniform military. But you want to empower debate: these bureaucracies are fairly old-fashioned in some ways, and people who want to keep doing things the same way — especially once it reflects the culture of that organization and becomes almost a key part of their sense of identity and purpose — it’s very hard to challenge.

So the job of somebody who’s looking at it from the outside or coming in as a civilian is to empower the iconoclasts — but, not let them carry the day either. Like I said, I think there were too many iconoclasts in the 1990s who believed that a revolution of military affairs could be sweeping and comprehensive, and a lot of their ideas, I think, were bad.

So really, you want the debate itself. You can’t presuppose that either the traditionalists or the reformers are going to be correct on any given issue. You have to fight it out on the merits of the case. That’s the real job of trying to push these bureaucracies forward.

Control and Command: the Crux of Deterrence

Jordan Schneider: How much harder is it to project deterrence in a time of rapid technological change?

Mike O’Hanlon: Start with the basics — and one basic is that, if you have Americans forward-stationed in another country on its territory in significant enough numbers, we would be directly implicated in any war from day one. That tends to help regardless of technology. The fact that we’re in Japan, Korea, Germany, Poland, I hope the Baltic states increasingly in the future, and other key exposed locations in Central and Eastern Europe and Kuwait, Qatar, Bahrain — that signifies an American commitment. It’s pretty hard for an adversary — regardless of the technology they employ — to think they could launch a large-scale aggression against a country like that and not have the United States intervene as well, partly because we’re going to have our people getting hurt or killed on day one in any comprehensive attack.

So, just because technology is getting more advanced, bear in mind that we have not ever really had an ally attacked when we already had large-standing US forces on its territory. And they don’t have to be huge, but they have to be substantially credible.

Jordan Schneider: I Googled this today: there are forty American military personnel in Taiwan at the moment — which is not a lot; I think there are rumblings in Congress to do more or less with that.

But it really comes back to our first half of this conversation: the deterrence is, “We’re going to run the model in the first place.” Because I think that’s the only way you really have confidence: you think you can win a war without actually having to fight it.

Mike O’Hanlon: It’s a good point. The presence on Taiwan is complicated — I’m not completely in favor of it. I’m certainly not all that happy that it has become public information, and there are not combat-capable numbers of Americans — far less than the number of Americans who live in Taiwan just on a day-to-day basis for civilian or personal reasons. In that sense I’m not sure it reaches the threshold that I was talking about — I’m thinking about 40,000 US troops in Japan, 30,000 in Korea, close to 10,000 in Poland, 30,000 in Germany, and usually in the broader Middle East, a few thousand per major presence as in Kuwait, Qatar, Bahrain. If you have only a few trainers out in the field, that may not send the same message or constitute the same kind of deterrent.

So the Taiwan case is one to keep a careful eye on. And you’re right, some people want to just unambiguously make our presence there so clear that the Chinese would know they were picking a fight with the United States if they ever attacked Taiwan. And others would say, “No, that is just a provocation beyond the pale — and as soon as China realizes we’re moving in that direction, it’s going to add to their incentive to attack in the first place. Old rules of deterrents be damned, this is Chinese territory; we’re not going to let the Americans out-maneuver us by sneaking a few boots ashore and then thinking that leaves Taiwan somehow impregnable or unattackable.”

So there, the psychology is complicated — it’s not as simple as the Japan, Korea, Germany, or Poland cases.

Jordan Schneider: Aside from the basics, what else can countries do to show that they have capabilities that should scare their adversaries in technologies that they haven’t tested in decades?

Mike O’Hanlon: I think that the number-one thing I would counsel in response to your question is to think about the resilience of your military systems at least as much as their lethality. I have a hard time believing anybody’s going to challenge us because they don’t think we have enough weapons or firepower to ultimately bounce the rubble the last time in an all-out fight. They’re going to be more inclined to say, “The United States has all these cyber systems that are vulnerable, that were built with Bill Gates’s software that was designed to almost fail, and then Microsoft would come fix it for you (and make a nice penny in the process of fixing it) until they put out the next upgrade, which you got to buy because you got to keep up with the Joneses…” — the whole dependence on commercial software.

As much as I admire the innovation and ingenuity at a place like Microsoft, there was a certain element of a scam, in my humble opinion, regarding the idea of these upgrades every two years. On the one hand, that’s the pace at which things were moving with Moore’s Law, and so I guess you can’t really blame Microsoft for trying to give us all more and more capability, and a lot of people wanted that. But on the other hand, it led to a lot of software that had a lot of bugs, and we use a lot of that same software in our military systems.

The Japanese thought in 1941 that the best way to beat the United States was not to launch an attack on our homeland, but to launch an attack on our Pacific fleet, put us on the mat long enough that maybe we wouldn’t think it was worth the effort to re-arm and come back and get them — and meanwhile, they could consolidate their positions and create bastions in the Western Pacific. Their goal was not to defeat us in detail or in attrition warfare; it was to just give us a bloody nose or a knock long enough that we would reassess the political desirability of getting fully into the fight. It was unsuccessful, obviously, but it was not crazy given their options — because certainly what would’ve been a lot crazier is to try to get ready to defeat us in an all-out fight. Their idea was a bad idea, but it was not the worst possible bad idea I could think of.

Whereas, if you’re thinking now about the modern era, the main vulnerabilities we have are largely around our command-and-control communications computer infrastructure, and our dependence on that, and the habits we’ve developed in the last decades of assuming that would always be resilient and available because we fought lower-tech powers.

So, the number-one thing you should try to do against the United States is attack its command-and-control systems. And for us, the number-one thing we need to do to project deterrence is to show and demonstrate the resilience of those same systems.

Yes, I want to have firepower and lethality and enough munition stockpiles to have a protracted fight if necessary — but most of all, I don’t want anybody to have a theory of rapid victory against the United States where, by virtue of somehow taking down our infrastructure — whether it’s command and control, or railroads and ports and airfields that allow us to deploy and resupply forces — they can find an easy win, a first-round knockout punch, or at least a first-round knockdown punch that puts us on our back long enough that they can go achieve their desired goals in whatever theater they want to achieve them. Like a Chinese takeover of Taiwan before we can get back up and respond — that’s the number-one concern I would have for preserving and projecting deterrence today.

The number-two concern is Secretary Austin’s concept of integrated deterrence; we need to be able to wage economic war. The idea is that — especially in situations where they’re attacking a country that’s not our core ally, or where the scale of the fighting has not reached a certain threshold — we have to be able to respond resolutely and punitively without having to draw first blood against them or without having to disproportionately and therefore not credibly exceed the actual magnitude of the interest that’s being threatened. We don’t want to have to risk the destruction of New York to go save one of the Senkakus, a tiny island that’s not even occupied by anyone. You have to have the ability to be proportionate — which means you also need tools of economic warfare to complement those of military campaigns.

Jordan Schneider: Let’s talk about some of the bull cases for technologies that could have a real impact over a twenty-year horizon. Which should you pick, why, and how do you think they could play out in a US-China conflict?

The conversation continues with

• Mike’s predictions for revolutionary technologies in the coming decade

• His read on the impact of AI and quantum relative to cyber

• His favorite military history reads