(Editor’s note: Readers also might be interested in an accompanying piece, also on Peter Bergen’s book, by David Lapan.)

“We train our boys to be killing machines,” President Donald Trump declared on Twitter in October 2019. He meant it as a compliment—and, as Peter Bergen shows in his powerful new book, Trump and His Generals: The Cost of Chaos, Trump’s sense of military men and women as “killing machines” was what drew Trump to populate his cabinet early on with generals like Michael Flynn, John Kelly, James Mattis, and H.R. McMaster. Yet, in some cases sooner and in some cases later, Trump soured on all of them—and they soured on Trump. (The possible exception is Flynn, but he at least soured enough to cooperate extensively with Special Counsel Robert Mueller’s investigation.)  So, what did Trump think he would get from these generals, and what went wrong?

It’s all there in Trump’s October 2019 tweet. Trump’s view of generals—and of the military more broadly—was and remains a caricature, cartoonish to its core. Trump views the armed forces and those who serve in them as “killing machines” who want to use force, who simply want to leave after using force, and who want to treat foreign partners who help with the use of force in purely transactional terms—just as Trump expected his real estate deals to be carried out. In turn, when Trump’s generals defied Trump’s ill-founded expectations, he moved from surprise to disappointment to anger. That repeated cycle emerges in exquisite detail in Bergen’s rich narrative. (Full disclosure: Bergen is a Vice President of New America, where I’m an International Security Program Fellow.)

Start with the use of force. Trump took office beholden to the notion that the U.S. military wanted to be unleashed—freed from perceived shackles imposed by President Barack Obama and allowed to use force with fewer constraints and less oversight. That left Trump startled when his generals resisted the major use of force he’s considered in his first three years in office: against Iran. Bergen documents how Secretary of Defense Mattis, when asked to produce options for striking Iran, “simply ignored the president’s directive.” Mattis did so precisely because he realized that a U.S. strike against Iran might well unleash a regional war and, if it drew in Russia and others, perhaps even a wider conflagration. This was similar to Mattis’ refusal to provide options, also as directed by the Trump White House, for Russian and Iranian targets in Syria, precisely because, “[f]or Mattis, a military operation that spiraled out of control in Syria could spark a wider conflict.” These were just two instances of Trump’s generals proving what anyone who’s had the honor to work with the U.S. military knows: that those who must actually bear the burden of conflict are often the least inclined to relish it and the most inclined to consider it carefully, even skeptically, before letting slip the dogs of war.

But Trump’s generals didn’t just surprise Trump by not wanting to use force when imprudent. They also surprised him by not wanting to “bring the boys home” after using force, if the ultimate mission remained unfinished. Here, the key case is that of Syria and America’s counter-ISIS efforts there. Bergen recounts how Trump, in a December 2018 phone call with Turkish President Erdogan, abruptly concluded that America’s fight against ISIS in Syria was over and decided to get American troops out, telling Erdogan: “It’s all yours. We are done.” Mattis, backed by two other top generals—General Joseph Dunford, then Chairman of the Joint Chiefs of Staff, and General Joseph Votel, then Commander of U.S. Central Command—seemed to startle Trump by their impassioned resistance to a hasty departure. For Trump, it was time to go: enough of the enemy had been killed, so the mission was accomplished. But here, too, Trump’s impression of the military was facile. Killing the enemy—whether ISIS or others—is not the end but the means; and, with respect to Syria, the actual mission objective was ensuring that ISIS would no longer pose a threat to Americans. That objective—the real objective—remained unfulfilled, which meant that, for Trump’s generals, abrupt withdrawal was highly premature and downright dangerous.

And Trump misgauged his generals yet again when it came to America’s foreign military partners. Trump’s approach to NATO provides the clearest window here. For Trump, partners—even allies—are as good for the United States as whatever they’ve done lately for us. But Trump’s generals knew otherwise—that the whole idea of alliances and partnerships is to forge bonds that redound to mutual interests in ways that transcend the particularities of the moment. Bergen describes how Kelly, Mattis, and McMaster—who disagreed on much by the end—were united in their support for NATO as a bedrock of American national security.

Trump never saw what they saw in NATO: allies who fought and died alongside Americans, even in military actions undertaken in response to attacks primarily against Americans, such as the U.S.-led campaign in Afghanistan after 9/11, as well as holders of shared values like the commitment to democracy and human rights. Where Trump expected “killing machines,” his generals proved much more interested in preserving those with whom Americans could live and serve side by side with pride and comfort. Those generals understood those exquisite lyrics from Hamilton: “Dying is easy . . . living is harder.” When, as Bergen documents, Trump criticized his generals’ approach to America’s burden sharing with South Korea by claiming, “My generals don’t know anything about business,” Trump revealed his complete failure to recognize that those generals don’t treat military partners as Trump treated business partners—precisely because the military isn’t looking to make a quick buck but, instead, to protect America over the long haul.

Ultimately, what drew Trump to his generals—indeed, to generals more broadly, and to the military writ large—wasn’t any real understanding of how they thought or why they fought. It was, instead, a caricature of who they were—what Bergen rightly describes as “a gruff, tough-guy persona of the type that appealed to Trump.” Moreover, Trump was drawn to what he saw as the generals’ powerful toys—what Bergen calls Trump’s “schoolboy fascination with military hardware.” This wasn’t an understanding of the military that led Trump to embrace it: rather, it was a cheap caricature.

That caricature lay at the heart of the contradiction Trump promised as a candidate in 2016. As Bergen notes, Trump simultaneously (1) “called for a greatly expanded military and an unconstrained war against terrorists” and (2) “railed against America’s seemingly endless wars in the Middle East.” Only in Trump’s cartoonish view of the military could this contradiction be reconciled: Trump wanted to help his generals kill, then rush home. But that’s not a strategy any actual general would advise. And so here we are, three years into Trump’s presidency. The generals with whom Trump surrounded himself are gone; and Trump marches on, exulting by tweet in “killing machines” and lacking a coherent strategy for handling Iran, ISIS, or NATO.

IMAGE: U.S. President Donald Trump reaches out to touch the arms of Chairman of the Joint Chiefs of Staff Joseph Dunford (R) and U.S. Defense Secretary Jim Mattis (C) while delivering remarks during a meeting with military leaders in the Cabinet Room on October 23, 2018 in Washington, DC. (Photo by Win McNamee/Getty Images)