Note: I learned a lot from writing this post and getting reactions along the way. Take a look and learn along with me. 

North Korea claims, fairly credibly, to have detonated an H-bomb. It has been testing and improving missiles with the goal of being able to make a nuclear attack against the continental United States. It has a history of violating any diplomatic agreement it makes. North Korea has a patron, China, protecting it, but seemingly unable to control it. And 25 million South Koreans, along with a large share of South Korea's economy, are within easy artillery range in the Seoul metropolitan area. What can the US do to protect itself?

First, let me stipulate that we should be willing to let China have whatever it wants in relation to the Korean peninsula on three conditions: 

1. What China asks for is acceptable to South Korea
2. China is willing to enforce a complete blockade of North Korea unless North Korea verifiably gives up nuclear weapons
3. China is willing to acquiesce in the military actions against North Korea spelled out below.   

The rise of China is something I worry about quite a bit (1, 2), but that challenge has to be faced regardless of what happens on the Korean peninsula. If China wants to keep North Korea in business as an increasingly powerful rogue state in order to get what it wants even beyond the Korean peninsula, it might not agree to the deal above; in that case, we have to go ahead trusting that despite bitter complaints from China, it will in the end acquiesce to the military actions spelled out below. 

Although diplomacy with China seems fruitful, diplomacy with North Korea does not. So the approach to North Korea consists of declaring unilateral actions the US will take. 

The immediate military action the US needs to take is to begin shooting down every North Korean missile launched, not with the Thaad system that is so much in the news, but with boost-phase interception. This requires explanation. Arthur Herman writes in the March 12, 2017 Wall Street Journal op-ed shown at the top of this post: 

To its credit, the Obama administration agreed to install land-based and sea-based missile technology like Thaad and Aegis to protect South Korea and Japan. The problem with these systems is that they can shoot down an incoming ballistic missile only toward the end of its flight, as it re-enters the atmosphere. That narrows the margin for error. ...

There’s a better alternative. Technology exists now for stopping a North Korean missile launch much earlier, in its boost phase. It’s called boost-phase intercept, or BPI, and the U.S. and Japan have the means to deploy it.

All large multistage rockets require high-thrust booster engines to push them out of the atmosphere, which then drop away when the missile achieves orbit. Destroying a missile at this early boost phase has many advantages. Since it’s the hottest stage of a ballistic-missile launch, it’s the easiest for infrared sensors to detect and identify. It’s also the slowest phase of the launch, so the missile loses any advantage it might have in speed in its later descent.

A BPI would be launched ... from an unmanned aerial vehicle waiting at 55,000 feet and equipped with infrared sensors that will detect missile launches from 350 miles outside North Korean airspace. ...

The aerial vehicle would be equipped with a conventional antimissile missile of 225 kilograms ...

There are already American-built unmanned aerial vehicles capable of carrying up to four interceptor missiles of this size, while conventional aircraft have successfully done BPI tests using missiles of this type. 

Boost-phase interception is not a mature technology, but neither is North Korean missile technology. We can get every bit as much practice in shooting down North Korean missiles as they get in launching missiles and countermeasures to boost phase interception. US resources are enough greater than North Korea's, we can win an arms race of missile vs. antimissile.

Initially, antimissiles will need a manned platform. That means US aircraft patrolling off the North Korean coast for quite some time until the unmanned aerial antimissile platforms are ready.

Crucially, if all or most North Korean missiles are shot down in boost phase, North Korea will get very little practice in dealing with the tough technical challenge of having a missile and its payload survive atmospheric reentry intact. Also, devoting technical resources to evading antimissiles may distract from other technical challenges like dealing with atmospheric reentry. 

Is shooting down all North Korean missiles soon after launch overly provocative? I don't think so. North Korea is loudly proclaiming that these missiles are intended as a threat to the United States. Shooting them down is simple self-defense. Not shooting down North Korean missiles practicing for an attack on the United States is appeasement. 

Shooting down North Korean missiles does not require attacking North Korean territory. And it is likely to be much more effective in hindering North Korea in sharpening its missile threat than attacking hardened North Korean missile sites or hardened nuclear development sites. Shooting down missiles is proportional to the threat. It is not escalation. 

China will hate the idea of the US further developing antimissile technology. China's distaste for our justifiably using North Korean missiles as target practice must not be allowed to stop us from doing so. If China wants to stop us from developing our antimissile technology further, it can work harder to stop North Korea from launching missiles. And if it does do everything it can, as laid out above, to restrain North Korea, it can have whatever it wants in relation to the Korean peninsula that is acceptable to South Korea, too. (South Korea is under enough threat itself, it is likely to be amenable to many things.)  

In addition to being valuable in its own right, shooting down every North Korean missile soon after launch will make other warning the US must make credible. In particular, the US must be able to deter North Korean development of a nuclear missile submarine threat. The ability to shoot down submarine-launched missiles in boost phase will help, but depending on technological developments, it may be necessary at some point in the future for the US to destroy North Korea's submarine capability. Fortunately, other countries can probably be dissuaded from providing North Korea with submarine bases, so North Korea's submarine bases will be on the North Korean coast. Because of the nature of ports, it is a lot harder to harden a submarine base than it is to harden a missile site or nuclear development facility. So if needed five or ten years from now, the US should be able to carry out an effective attack on a North Korean submarine base before North Korea gets a fully operational submarine-launched nuclear-missile capability. We need to announce our intention to destroy North Korean submarine bases in the event North Korea gets close to a submarine-launched nuclear-missile capability long in advance so that China gets used to the idea. (There might be some effect on more protective measures for the submarine base, but there is only so much that can be done to protect a submarine base against US firepower.) Obviously, we don't want to be put in the position of having to attack a North Korean submarine base, so it is crucial to build up our credibility that we would if necessary, precisely so that we don't have to. Making every effort to shoot down all North Korean missiles soon after launch is one of the best ways to build up that credibility.  

It may be reasonable to pursue Chinese economic sanctions for about two more months (long enough to get past the October 18, 2017 Party Congress in China) before beginning to shoot down North Korean missiles. But it would be folly to wait much longer than that if the situation then is similar to the situation now, but with several more North Korean missile launches having taken place. Those two months should be devoted to putting US boost-phase interception technology development into high gear.   

Technical Issues. If Arthur Herman (the op-ed writer above) is overoptimistic about the state of boost-phase interception technology, it will make sense to pursue the economic sanctions route for longer before trying to shoot down North Korean missiles. Jason Smith raises technical issues about the readiness of boost-phase interception technology in this storified Twitter exchange. One issue raised is that North Korea and our other enemies will learn more about our antimissile technology if we use it. To me this does not seem a strong argument. Surely we learn as much by getting experience with our antimissile technology as they (and other potential adversaries) do by seeing where we are so far. In order to avoid losing face, the US needs to avoid overpromising how well its antimissile technology will work in the short run. We don't need to be able to connect with a North Korean missile every time at first; shooting down a North Korean missile at a rate far below 100% can still help the US as long as the expectations are set low.  

Jason emphasizes the difficulties created by North Korea's mountains. The highest mountain is Mount Paektu (or Baekdu), 2744 meters high. At 10 Gs, 98 meters per second squared straight up, a missile would reach an altitude of 3136 meters after 8 seconds. So after 10 seconds or so, even at some angle, a missile will be above all the North Korean mountains and the mountainous terrain can no longer hide the missile. (However the problem of seeing around the curve of the Earth is now relative to the tops of the mountains, so it would require the detection systems to be that much higher.) Since it will only be traveling at a speed of about one kilometer per second at that point, it needs a longer boost phase than that. (Even straight up, 1 kilometer per second can be counteracted by Earth's gravity in just 102 seconds.) Arthur Herman claims:

Once the launch is detected, operators of the BPI system would have nearly a minute to decide whether a launch is genuine or not, and then to initiate the intercept—more than enough time to prevent a mistake.

If Arthur is even in the right ballpark, shaving 10 seconds off of that minute because of mountainous terrain still leaves 50 seconds. 

Jason also raises the issue of false alarms. Given its threats, North Korea has no right to a space program, which in North Korea's case would be little more than an advertisement of its missile prowess in any case. Being willing to risk shooting down a space launch may be essential in order to respond quickly enough to a North Korean missile. Alternatively, in the quite unlikely event that they want to pursue a space program for entirely peaceful reasons, the North Koreans can arrange for a pre-inspection of their satellite-launch rocket and monitoring its time of launch so that it can be distinguished from a missile. 

It is important to distinguish between missiles and supersonic jets. That should be feasible. To the extent it is difficult, North Korea can be warned about the types of extreme behavior of supersonic jets that would cause them to be targeted as if they were missiles.

A key principle we need to insist on to the international community is that given the provocations of the North Korean regime, shooting down a North Korean missile is not an act of war, it is antiwar. If one asks as a technical matter of international law why North Korean can be treated differently than other countries in this regard, its repeated violation of UN sanctions is salient. The UN is not good at managing all the details of international affairs, nor should we expect it to do so, but sometimes it is able to identify a rogue state. 

I am old enough to remember the ridicule heaped on Ronald Reagan's interest in antimissile technology—so called "Star Wars." (The Soviets took it much more seriously than most American journalists did.) Thirty years have passed. Antimissile technology is not in the same place now it was back then. If it is not as good as it should be, that may owe something to partisan opposition to antimissile technology at times during those 30 years, and overoptimism that economic sanctions would take care of the North Korean threat. (I confess that I thought the North Korean nuclear threat would have been dealt with by now, so I understand this miscalculation.) If boost-phase interception technology is not ready to do what I am positing it should do, someone should be held to account for not getting it to the point we can use it by now when we desperately need it. And we need to rectify any deficiency in our antimissile technology as soon as possible. 

One final objection to antimissile technology is that it might destabilize strategic deterrence between the US and Russia or between the US and China. But this is an argument that antimissile technology is too good, quite inconsistent with the argument that antimissile technology is not good enough to use against North Korea. There is no time in the foreseeable future when antimissile technology will be good enough to stop Russia or China from wreaking massive nuclear destruction on the US should they choose to do so. It is quite possible for antimissile technology to be good enough to help greatly in dealing with North Korea without disturbing the doctrine of "mutually assured destruction" between the US and Russia or between the US and China.