This week President Donald Trump got angry. With his arms folded across his chest, his eyes darting around the room, you can feel the emotion behind his words: “North Korea best not make any more threats to the United States. They will be met with fire and fury like the world has never seen.” The president’s remarks were clear evidence of a danger that must be addressed.

Not the danger of this particular crisis—there is little we can do about that. For now, US law puts no impediments—no checks or balances—on the president’s ability to launch a nuclear war. Instead, what the president’s words highlighted was the inevitable failure, over the long run, of nuclear deterrence.

For decades, US security policy has relied on a theory, an idea about how human beings are likely to behave. And we, as a nation, have agreed to run risks based on this idea that the threat of mass destruction can prevent attacks. Nuclear deterrence seems sensible enough. Who, after all, would be crazy enough to start a nuclear war? And, for the most part, our presidents seemed to confirm that nuclear weapons were secure in American hands.

As former UN secretary general Ban Ki-moon is fond of saying, there are no right hands for nuclear weapons. The problem is the need for perfection. A single slip-up could lead to catastrophe. The instruments of deterrence are inherently fallible. I’m not talking about the computers that control our arsenal and its early-warning systems, though Eric Schlosser’s Command and Control does an admirable job of explaining the risks inherent in keeping thousands of nuclear weapons on hair-trigger alert. But the machines are not the main problem—we are.

Nuclear deterrence is not a computer that purrs quietly in a corner on its own. Human beings are intimately involved at every step. As the president is so vividly demonstrating, people lose their tempers, overreact, and get overwhelmed by emotion. People can lose their sanity—raving and acting at random.

Nuclear advocates have said for decades that nuclear weapons can’t be gotten rid of, because they “can’t be disinvented.” This is undeniably true, but also entirely specious. No technology is ever disinvented. Who disinvented the PalmPilot? Who disinvented black and white TV? Who disinvented the Hiller VZ-1—a flying platform designed to lift a single soldier 10 to 20 feet up into the air? These technologies weren’t “disinvented,” they were abandoned, either because better technology came along (as with PalmPilots and black and white TVs), or because people simply realized the original technologies weren’t all that useful (like the Hiller VZ-1—why would you put a soldier in a position where the person’s both especially noticeable and entirely vulnerable?) Nuclear weapons fall into the second category. They’re just not very good for anything, except slaughtering civilians en masse.

Of course we can get rid of nuclear weapons—if they’re stupid technology. Imagine you bought a new kind of stove that (you heard later) blew up on a regular basis and, it turned out, couldn’t even boil water. Why would anyone keep technology that is both dangerous and virtually useless?

Eliminating nuclear weapons used to be considered pie-in-the-sky utopianism. But since 2008, when four cold war hawks of considerable standing—Secretary of State George Shultz, Secretary of Defense William Perry, Secretary of State Henry Kissinger, and Senator Sam Nunn—came out in favor of eliminating nuclear weapons, the debate has shifted.

Nuclear-weapons advocates imagine all sorts of exaggerated powers for nuclear weapons: They protect us, cement our alliances, even uphold the world order that sustains our prosperity. In their obsessed minds, nuclear weapons are essential. But their beliefs are based on misperception and wishful thinking, not reality.

People point to the fact that nuclear weapons haven’t been used in 70 years as proof that they are awesome, portentous weapons, too powerful to use. And the fact of their disuse is suggestive; it suggests they are lousy weapons. It’s possible no one has used nuclear weapons for 70 years not because there is a kind of holy dread and wonder that surrounds them but because no one has been able to find a situation in which the weapons would actually be useful.

If you want to judge the question of whether nuclear weapons are essential or not, a far more telling piece of evidence comes from when George H.W. Bush retired almost all of the tactical nuclear weapons in the US arsenal. What was telling was not what was said, but rather what wasn’t said. No one demanded their nuclear weapons back. No military officers went to Congress, sat at the witness table, and demanded the return of their tactical nuclear weapons. No one pounded the table, shouting, “Those weapons are essential for the safety and security of this great nation!” Their silence speaks volumes about how military professionals judge the military utility of nuclear weapons.

We need to dispel the nuclear believers’ dark fever dream of awe and power, and insist on hard, cold reality: Nuclear weapons are risky, blundering weapons whose only real use—deterrence—will lead to catastrophe.

Trump’s threat signals the end of the delusion that nuclear deterrence can be safe. If even stable, mature democracies can elect leaders who can’t be trusted with nuclear weapons, then there is no way to justify keeping them.