Jonathan Ernst / Reuters

President Trump’s May 8 announcement that he was withdrawing from the Iran nuclear deal should not have come as a surprise. He’d spent years railing against the plan—“the worst deal ever,” he dubbed it—and had promised to rip it up. And yet up to the moment when the president made the final call, there was still some suspense about what he would say.

This was not merely wishful thinking by the deal’s backers, though it was partly that. It was not only that members of Trump’s team, most notably Defense Secretary James Mattis, had voiced support for the deal. It wasn’t even just that Trump relishes taking the press (and even his own advisers) by surprise.

No, the other big reason that no one could be sure was that Trump almost always folds. Faced with a tough decision, the president has consistently blinked, giving in to his opponents. Trump has mocked former Secretary of State John Kerry for his supposedly poor negotiation powers, argued that multiple past international agreements suffered from weak-kneed diplomacy, and criticized British Prime Minister Theresa May, a close U.S. ally, for her conduct of the discussion on the U.K.’s departure from the European Union.

The president talks a goodor at least aggressivegame, but he doesn’t always walk it. The Iran deal is one of the few caseswhich also include ditching the Paris climate agreement, leaving the Trans-Pacific Partnership, and moving the U.S. embassy in Israel to Jerusalemwhere Trump has stuck to the hard-nosed approach he promised during the campaign. But he twice flinched on Iran, too. In October 2017, his first chance to walk away from the deal, he made threatening noises but kicked the deal over to Congress and didn’t withdraw. He passed up another chance in January.

Foreign leaders were among the first to notice Trump was a pushover. In the first weeks of his presidency, they angled for face-to-face meetings with the new president, realizing that in personal meetings, especially when subjected to flattery, he could be swayed. The efforts paid off: Trump shared sensitive intelligence with the Russian foreign minister and ambassador. He backed down from threats to brand China a currency manipulator, and said that a 10-minute lesson from Xi Jinping on Chinese–Korean relations had convinced him Beijing had less sway over Kim Jong Un than he had believed. That same spring, Trump called the leaders of Mexico and Canada to announce he was going to pull out of NAFTA. The two men convinced Trump to agree to renegotiation instead, a significant climbdown.

The pattern has persisted ever since, with two prominent examples in the last three days alone. One was a high-stakes meeting at the White House, centered on the FBI’s reported used of an informant on the Trump campaign. Ahead of the meetings, there was speculation that Trump, who had tweeted furiously about the informant, would demand an unusual inquiry into the Justice Department or force Deputy Attorney General Rod Rosenstein out. Instead, the meeting ended with a plan for future meetings, and the Justice Department’s committing to do more or less what it had already committed to doing.

Meanwhile, negotiations with China intended to forestall a trade war, which Trump previously said would be “easy to win,” wrapped up. The U.S. agreed to postpone tariffs on Chinese imports. The Chinese made some vague promises about narrowing the U.S. trade deficit, but as Keith Bradsher writes, “China gave up little in return, spurning the administration’s nudges for a concrete commitment to buy more goods from the United States, and avoiding limits on its efforts to build new high-tech Chinese industries.”

This is a repeated motif on trade deals. Trump makes a splashy announcement, but when it comes to the details, he pulls his punches. China is not the only target of suggested Trump tariffs. In March, he abruptly announced tariffs on imported steel and aluminum, saying foreign products were hurting U.S. producers. But he has postponed tariffs on the European Union after objections from American allies there, and also postponed tariffs on steel from Canada, the single largest exporter of steel to the U.S. in 2017, because of the ongoing NAFTA negotiations.

How are those NAFTA talks going, by the way? Though Trump was nearly as negative about NAFTA as he was about the Iran nuclear agreement, he shied away from immediately withdrawing last spring, and talks have slowly dragged on. After the administration missed a deadline that Speaker Paul Ryan set for congressional approval of a deal, Treasury Secretary Steven Mnuchin said this week that the process could slide into 2019.

He even has some regrets about withdrawing from the TPP. The Wall Street Journal reported in April that the president had told lawmakers that he had “deputized Robert Lighthizer, the U.S. trade representative, and Larry Kudlow, the director of the National Economic Council, to study the possibility of re-entering the TPP if the terms were favorable.”

Trump is just as apt to fold on other matters. On a June 2017 call with Saudi Arabia’s King Salman, according to The Washington Post, Trump was supposed to scold the Saudis for their offensive against Qatar. After Salman delivered a flattering series of compliments, the president dropped his objections.

In December, when the U.S. announced it would recognize Jerusalem as the capital of Israel, the UN General Assembly voted overwhelmingly to condemn the move. Trump said the U.S. would withhold aid from those who voted for the resolution. “Let them vote against us,” he said. “We’ll save a lot. We don’t care. But this isn’t like it used to be where they could vote against you and then you pay them hundreds of millions of dollars.” Yet when the White House released its budget for 2019, “not a single country lost funding on the basis of voting against the U.S. at the UN,” John Hudson reported. More recently, he made a similar threat to countries that don’t back a joint North American bid for the 2026 World Cup, raising the question of why any country should take the threat seriously.

In January, Trump met with Senate Minority Leader Chuck Schumer with a government shutdown looming. In exchange for Democrats backing funding for increased border security and possibly a wall, Trump agreed to back a deal to replace DACA, the executive order creating a legal status for unauthorized immigrants brought to the United States as children. Hard-liners in the White House thought the deal was such a giveaway, in fact, that Chief of Staff John Kelly reportedly called Schumer and torpedoed it.

In February, Trump held a meeting with members of Congress to discuss gun control following the school shooting in Parkland, Florida. He ridiculed lawmakers for being too timid in confronting the NRA and asked them to take up proposals opposed by the group. Days later, he met with the NRA, and when he unveiled his slate of ideas to combat gun violence, it included none of the proposals the NRA opposed.

In April, Trump demanded an immediate withdrawal of American troops from Syria. Military advisers worked to talk Trump out of it, bargaining the president down to speeding up the timeline. Less than two weeks later, he launched new airstrikes against the Syrian regime, following chemical attacks. On Monday, the commander of Central Command told the Tampa Bay Times it hadn’t taken much pushback for Trump to concede on withdrawal.

This is not to say that Trump has no core convictions. On a select set of policies, especially those he discussed most frequently on the campaign, the president has been extremely tenacious, as I wrote in May. He has worked to find a way to implement a travel ban and build a border well—not always effectively, but unceasingly. He single-handedly kept Obamacare repeal alive long past when other Republicans were ready to give in. He drove through tax cuts over the predictions of most analysts. It is only at the negotiating table, supposedly his natural habitat, where his nerve seems to fail him.

Why is Trump so apt to fold? One reason is that he is surrounded by advisers who often feel that his ideas are bad and try to talk him out of them. The Iran-deal withdrawal came only after National-Security Adviser H.R. McMaster, who supported the deal, was pushed out and replaced with the more hawkish John Bolton. Beyond that, however, Trump brings two big weaknesses to the table. First, he’s not as skilled a negotiator as he likes to pretend, and second, he seldom puts in the time to learn the facts about whatever he’s handling.

In April, Jonathan Swan had an illuminating post about Trump’s one-dimensional approach to negotiating:

President Trump tells people he keeps the world guessing with his wild unpredictability. But those who work most closely with him say he’s a one-trick pony in negotiations.

The trick: Threaten the outrageous, ratchet up the tension, amplify it with tweets and taunts, and then compromise on fairly conventional middle ground.

“​His ultimate gamble is: ‘You don’t have as big of stones as I do,’” a source close to Trump told me. “‘You’re going to feel too uncomfortable where I go. The stakes are too high. This is too far outside your comfort zone.’”

This turns out to be an effective tool, because not many people have Trump’s appetite for awkwardness. But if it doesn’t work, the president has few other tools at his disposal, and tends to back down.

Trump is also at a disadvantage because he doesn’t bother to do his homework. The president has strong instinctive views, but little tolerance for studying facts. During the campaign, he said he’d figure out what he needed to know once he won, but since taking office he hasn’t bothered. That means he’s already ceded the debate over reality to his opponents, as when China’s Xi laid out his history lesson on Chinese–Korean relations.

The president’s single negotiating trick and lack of interest in facts threaten to come together to dangerous effect in his upcoming summit with North Korea’s Kim. That is, assuming it happens: The Times reported on Sunday that Trump was starting to get cold feet about it, and he said on Tuesday that the meeting might be delayed beyond the scheduled June 12 date.

Trump has portrayed even the fact of this meeting as a major breakthrough, and evidence that he can get done what other presidents could not. He managed it in part, however, by backing down. In November 2017, he said that North Korea must fully denuclearize before he would consider talks. Pyongyang made no such commitment before Trump announced the summit, though.

Should the meeting come together, however, Trump will be in a high-stakes conversation with Kim. This is risky because, as past experience has shown, Trump is easily susceptible to flattery and other forms of persuasion when meeting face-to-face with foreign leaders. He will also face a disadvantage on the facts. The nuclear program has been central to North Korea’s national policy for years, while for an American president it is just one of scores of important issues, and Trump’s aides say he has made little effort to cram.

“Going into the North Korea meeting, senior administration officials say, the president has been almost singularly focused on the pageantry of the summit—including the suspenseful roll-out of details,” the Associated Press reported Tuesday. “He has not been deeply engaged in briefing materials on North Korea’s nuclear program, said three people with knowledge of the White House efforts.”

This doesn’t mean that Trump will necessarily fold when he meets Kim in Singapore. What makes a summit like this so interesting, and risky, is how unusual and volatile it is. It matches up two leaders who are unusual and volatile themselves. Both have a long history of ostentatious bluster, and spotty records of follow-through. When they’re eyeball to eyeball, it’s an open question which one will blink first, but history suggests it’s likeliest to be Trump.

We want to hear what you think about this article. Submit a letter to the editor or write to letters@theatlantic.com.