THANKS largely to Kim Jong Un (aka “Little Rocket Man”), missile defence of the American homeland is a hot topic. Next month the Trump administration is expected to publish a review of the nation’s defences against ballistic-missile attack. Funding for the Missile Defence Agency (MDA) is likely to exceed $11bn for 2018, over $3bn more than the president’s original request (assuming Congress can come up with a deal on the overall budget). An emergency request of nearly $5bn for additional “missile defence and defeat” funding was made in November.

The intelligence agencies had assured Mr Trump when he took office that not until 2020, possibly even 2022, would Mr Kim have a reliable intercontinental ballistic missile (ICBM). That relatively comforting assessment was blown apart in July when North Korea successfully tested two missiles with the range to hit cities in the continental United States, and, in September, when it conducted an underground explosion of what appears to have been a thermonuclear device.

Since then, the priority has been to reassure Americans that they can be protected from Mr Kim—whom Mr Trump’s national security adviser, H.R. McMaster, has alarmingly (and without evidence) described as “undeterrable”. In October, Mr Trump boasted to Sean Hannity of Fox News: “We have missiles that can knock out a missile in the air 97% of the time, and if you send two of them, it’s going to get knocked out.” Most missile experts were horrified by the president’s blithe confidence in the effectiveness of the only missile-defence system, known as ground-based mid-course defence (GMD), that is intended to shield the United States from a limited ballistic-missile attack. They fear Mr Trump may persuade himself that a pre-emptive attack on North Korea would be risk-free, at least for America.

As far back as the wildly over-ambitious Reagan-era strategic defence initiative (“Star Wars”), missile defence has been as much an aspiration as a policy, according to Michael Elleman, a former missile engineer now at the International Institute for Strategic Studies. Local and regional systems, such as Patriot missiles, the ship-based Aegis and THAAD (Terminal High Altitude Area Defence), have become effective. But GMD, which has to hit a much faster moving target from further away, has not progressed to the same extent.

In the wake of the September 11th 2001 attacks, the Bush administration needed a response to the growing threat of ballistic-missile technology proliferating to “rogue” regimes, such as Iran, Iraq and North Korea. Consequently, the GMD was quickly cobbled together with a mixture of old and new technology and hurriedly commissioned in 2004. Today’s GMD and its associated systems span 15 time zones, comprise seven different types of sensors (on land, at sea and in space) and 44 interceptors, each costing $75m, deployed at military bases in Alaska and California. GMD is designed to track, intercept and destroy an incoming nuclear warhead outside the earth’s atmosphere through the force of the collision alone.

Yet even now, after $40bn has been invested in it, GMD still has the hallmarks of an immature system. Tom Karako, a missile-defence analyst at the Centre for Strategic and International Studies, concedes that many of the improvements that were planned and expected have not yet come to pass. GMD’s interceptors have been tested 18 times, succeeding on ten of them. In May last year, a successful intercept was carried out for the first time against an intercontinental ballistic missile of the kind Mr Kim would need to reach the west coast. But three out of four previous tests had ended in failure.

Mr Trump’s 97%-success-rate claim appears to be based on a misunderstanding of the MDA’s arithmetic. The actual “single shot probability of kill” of GMD interceptors is 56%. The MDA has taken 60% as its benchmark and calculated that if four interceptors were launched at one warhead, the kill probability would rise to 97%. However, James Acton, who works on nuclear policy at the Carnegie Endowment for International Peace, questions this. “If one interceptor fails because of a design or manufacturing flaw,” he points out, “the others may fail too because they have the same problem.”

Moreover, test successes have been under ideal conditions. With just a few minutes’ reaction time and faced with several incoming missiles, each equipped with multiple decoys, some “leakage” is almost inevitable, thinks Mr Elleman.

What should the missile-defence review recommend? Mr Acton thinks it may go for a big increase in the number of interceptors based in Alaska—perhaps up to 100. But he questions whether that would really improve capability, because of the system’s inherent flaws.

Another possibility is speeding up the deployment of the Multi-Object Kill Vehicle, which would give each interceptor missile multiple shots at incoming warheads and is due to be ready in 2025. A more radical option would be to develop solid-state lasers small enough to be carried by drones, which would fly close to an enemy country and kill missiles in their vulnerable boost phase. That may be a decade away. In the meantime, someone should explain to Mr Trump that, at least for the foreseeable future, there is no certain defence of the homeland against even a fairly limited ballistic-missile attack.