Daniel W. Drezner is a professor of international politics at the Fletcher School of Law and Diplomacy at Tufts University and a regular contributor to PostEverything.

Korean People’s Army soldiers carrying packs marked with a radioactive symbol take part in a military parade in Pyongyang on July 27, 2013. North Korea appeared to carry out a sixth nuclear test on Sept. 3, 2017, possibly of a hydrogen bomb more powerful than any device it has previously detonated, presenting President Donald Trump with an unprecedented challenge. (Ed Jones/Agence France-Presse via Getty Images)

This past weekend, President Trump put in a busy week of golf, bestowing golf trophies, and tweeting insults at Puerto Ricans devastated by Hurricane Maria. But he made sure to take at least five minutes of his time to undercut his secretary of state, Rex Tillerson:

I hope that the only thing these tweets accomplished was that they kneecapped Tillerson. Because if Pyongyang takes them seriously, then everyone is in trouble.

Please understand that Trump’s tweets are actually worse than the rhetoric he deployed in his demented U.N. General Assembly speech, in which he said, “Rocket Man is on a suicide mission for himself and for his regime.” For all of the recklessness of that speech, however, those threats to North Korea were contingent: “If it is forced to defend itself or its allies, we will have no choice but to totally destroy North Korea.” As bad as that language is, the message was that the military option would be a response to further DPRK provocation.

There is no contingency in Trump’s tweets over the weekend, and that is a serious problem. It signals to North Korea that nothing sort of abject acquiescence will stop the militarization of the crisis. Kori Schake explained why this is problematic a few weeks ago in the Atlantic:

McMaster argued that military action must be on the table if diplomacy cannot keep North Korea from attaining the ability to attack the United States with nuclear weapons, supposedly because Kim Jong Un cannot be deterred. This is exceptionally sloppy thinking. North Korea may not have slowed its development of nuclear weapons and delivery vehicles, but it has been deterred from attacking South Korea and the United States since 1953.

The White House is conflating the possession of nuclear weapons with their use; its policy would be less likely to result in nuclear war or a humiliating climb down by the president if it distinguished between the having and the using. By predicating its policy on preventing acquisition, the administration has dramatically increased the value to North Korea of its nuclear and ICBM programs. A shrewder course would be underscoring that nuclear weapons make no difference, because any conventional or nuclear attack by North Korea on America or its allies would—as has been the case since 1953—result in the end of the Kim regime. That approach diminishes rather than accentuates the political gain to North Korea of becoming a nuclear possessor.

North Korea has not been deterred from developing nuclear weapons, but it has been deterred from the use of large-scale force. The more Trump signals that the time for talk is over, the more North Korea must fear a preemptive attack from the United States. And if the North Korean regime becomes convinced that such an attack is likely, it will inevitably choose to unleash its nuclear and conventional arsenal on Japan and South Korea. Which would be devastating for everyone involved.

This spiral model of military escalation rests on a crucial assumption: North Korea’s leaders take Trump’s “fire and fury” rhetoric seriously. Clearly Trump wants that to be the case. As Jonathan Swan reports in Axios, Trump likes the idea of being perceived as the madman of world politics:

Plenty of world leaders think the president is crazy — and he seems to view that madman reputation as an asset. The downsides are obvious: the rhetoric can unnerve allies and has the potential to provoke enemies into needless, unintended war. But Trump keeps using the tactic, with varying degrees of success.

Here I am going to disagree with Swan: I have seen zero evidence that Trump’s madman gambit has yielded anything in the way of policy gains. Indeed, when push has come to shove on NATO, NAFTA, North Korea, or any of his domestic showdowns, Trump has folded like a piñata. Indeed, the crux of Swan’s story is an anecdote about threatening to leave KORUS, and Trump folded on that threat as well. As far back as February I warned that he was becoming more predictable and less credible in his rhetoric.

The Axios story just makes things worse. It publicly confirms that Trump wants to use the madman gambit in international negotiations. But the first rule of the madman gambit is that you cannot talk about the madman gambit because then it stops working. And as North Korea seeks outside analysis to understand Trump, I guarantee you that this story will come up in conversation.

Consider that in 2017, the best hope for war not breaking out in Northeast Asia is that Kim Jong Un decides that he does not need to take the president of the United States seriously. If Kim decides that Trump is bluffing, then he will not feel compelled to strike first.

That outcome might be good in the short run, but it is spectacularly awful in the long run. The more Trump squanders American credibility, the greater the likelihood of the United States having its threats not believed in world politics. Which means that, at some point in the future, Trump will have to act on his bellicosity.

The hard-working staff here at Spoiler Alerts is still not worried too much about the world ending. I strongly suspect that the rest of the world looks at Trump as a blowhard rather than a madman. But over the long run, I am terribly worried that the president of the United States is an idiot who is too stupid to realize that he is an idiot.