It’s 5:31 pm on November 27, 2003. A US plane touches down at Baghdad International Airport—hardly a unique occurrence at the time. The two US Air Force pilots slow the aircraft and eventually bring it to a halt. This one, though, doesn’t sport the same dull gray paint job as the seemingly endless numbers of cargo aircraft ferrying supplies to the tens of thousands of US and coalition troops in the country and the materiel to fuel the war effort. This aircraft bears the distinctive, blue and white livery of Air Force One. President George W. Bush is onboard. He’s visiting US servicemembers for Thanksgiving.

Few observers at the time knew what a precedent Bush was setting. He would visit troops in both Iraq and Afghanistan several more times before leaving office. His successors, Barack Obama and Donald Trump, would, as well. Nobody on the plane that day likely expected that US servicemembers would be deployed to America’s post-9/11 wars for another seventeen Thanksgivings. The Iraq War was eight months old, and while signs of the emerging insurgency were there, conditions hadn’t descended to the brink of full-fledged sectarian civil war, as they soon would. Perhaps a few people predicted that the US military would be in Iraq in substantial force for the better part of a decade (with follow-on troops in the country to support the fight against ISIS for another handful of years), but US officials certainly didn’t publicly expect such a long-term commitment. Meanwhile, the Afghanistan War existed on the comparative strategic periphery. The total number of US servicemembers there in late 2003 was about a tenth of the number in Iraq, casualties were relatively low, and the fighting was handled disproportionately by special operations forces. And yet, sixteen years after President Bush stepped off of Air Force One in Baghdad to share a turkey dinner with troops, President Donald Trump did the same thing with US forces in Afghanistan.

The first stop during Bush’s two-and-a-half-hour visit in 2003 was a dining facility, where about six hundred soldiers from the 82nd Airborne Division and 1st Armored Division were gathered for dinner. If he had returned for Thanksgiving the following year, they would have been different soldiers from different units. The year after that, different ones again. This was a rotational war—a fact that would have lasting consequences on its conduct and results.

Sixty years earlier, in November 1943, if President Franklin D. Roosevelt wanted to share Thanksgiving dinner with the troops, he might have done so with soldiers wearing the same 82nd Airborne and 1st Armored Division patches that adorned the sleeves of Bush’s dinner companions in 2003. He would have found many of them along the Gustav Line in Italy. If he wanted to do so again the following year, he could have found the same units, with many of the same soldiers, still fighting in Europe—the 1st Armored Division in northern Italy’s Po River valley and the 82nd Airborne now in France.

The simplified narrative contrasting these two models holds that during World War II, American servicemembers knew they wouldn’t be going home until the war was won, while those who served in the wars in Iraq and Afghanistan only needed to look at a patch chart—essentially the master plan forecasting which units would deploy, when, and where—to see when they would be leaving and who would be replacing them. The reality is slightly more complicated, to be sure, but the general idea is accurate. After 9/11, America would fight its wars on a rotational model. Two decades of war. Two decades of new units arriving and experienced units leaving. Two decades of Thanksgiving dinners eaten by different men and women wearing different unit patches.

Why does this matter? As an observation, it’s hardly remarkable. It’s just another way of saying what many have already said: that the United States didn’t fight a twenty-year war in Afghanistan, but rather twenty one-year wars. This description has become so common that it’s difficult to pin down who said it first (although whoever did so likely cribbed from John Paul Vann’s description of the Vietnam War). But that’s the problem. It’s used as if its meaning is self-evident. It’s used so often that it risks losing whatever intrinsic explanatory power it once had, inching ever closer to an empty platitude that everyone accepts, nobody questions, and—worst—relieves of us our professional obligation to probe deeply into the reasons that twenty years of war, trillions of dollars, and thousands of fatalities and many more left wounded led to such middling strategic outcomes.

So, what does it actually mean that the United States fought twenty one-year wars in Afghanistan (and another succession of such wars in Iraq)? Is it inherently problematic? After all, four runners on a 4×100-meter relay team will typically finish a lap of the track before a single runner can do so alone. Fresh legs and a lower heart rate take over, superior to the single runner’s increasing heart rate and tiring muscles. Why can’t a brigade combat team do everything it can and hand the mission over to a new unit before it exhausts its resources and combat power, with the fresh unit carrying the progress forward and building on it? In theory, it can. But the US experience over the past twenty years shows just how far practice can diverge from theory.

The rotational model led, with varying degrees of directness, to several key features that would define the conduct of the wars in Iraq and Afghanistan. First, it created a situation in which one unit would replace another, but with a vastly different approach. For the most part, a maneuver unit had only so much leeway in terms of selecting its objectives and the areas that it would prioritize. These typically were enshrined in the lines of effort defined by higher headquarters. But a commander who was wholly committed to the counterinsurgency model laid out in US doctrine from 2006 onward would lead a unit through a deployment that looked very different from one who held steadfast to the idea that US combat forces’ job was to close with and destroy the enemy. And here’s the thing: those two archetypes existed in both wars. The result was too many swings between kinetic and nonkinetic operations, key leader engagements and firefights, funding projects and launching raids.

Second, even where this wasn’t the case, even when lines of effort were consistent and their relative prioritization was generally unchanged despite the turnover in units, there were still challenges. For one, the US military largely wrote its doctrine for these wars on the fly. In many ways, doing so was a remarkable feat, but it also meant that many of the finer points—the details that don’t get captured in doctrine but the importance of which becomes extraordinarily apparent on the battlefield—lagged. This is why, for example, the Army was still producing intelligence officers in 2007 who were well prepared for a Fulda Gap scenario against a Soviet attack but decidedly less so for the social network mapping, pattern-of-life analysis, and both lethal and nonlethal targeting they would have to learn as soon as they deployed, where those were the capabilities that mattered. Battlefield problems were being solved with battlefield solutions, developed by units in contact with both the enemy and noncombatants, and much of that ad hoc knowledge was lost during the transition period between outgoing and incoming units. This issue was only compounded by a failure to digitally maintain hard-earned lessons that later units could easily access and learn from. Even if the same unit deployed again (which most did, and soon), they would be starting fresh.

Third, units became reactive in a way that didn’t account for broader contexts. What does a pair of bombs hidden in fruit carts in Lashkar Gah mean to a unit in the middle of its six-, nine-, or twelve- month deployment there? In my experience, it means a sudden fixation on fruit carts. What does it mean if you look at the entire history of the US-led war in Afghanistan and find an extraordinarily small number of bombs planted in fruit carts? Something very different. The US military entered a forest in Afghanistan in 2001, and another one in Iraq in 2003. And then we spent too many years in each myopically focusing our microscopes on a succession of individual trees.

Fourth, the rotational model encouraged a focus on short-term goals. The cynical explanation for this is highlighted by a document that was making the rounds among Army officers in Iraq during my first deployment there. In it, an anonymous junior officer described his or her reasons for getting out of the Army. Among them were the differences between what the document defined as “GWOT officers” (those who commissioned and had served during the post-9/11 wars) and “‘90s officers” (those whose service predated those wars). The ‘90s officers, the junior officer lamented, viewed the wars as the proving grounds where promotions were earned—a deployment was the same opportunity for officers to demonstrate their capabilities as rotations at combat training centers had been previously. Why, then, would any officer undertake to achieve something that could not be completed during his or her deployment? Why give your replacement the credit? Even a more generous view of motives, though, yields an outcome that is not that dissimilar. Leaders who truly wanted to do something good during the wars in Iraq and Afghanistan wanted to leave knowing they had done so. How best to do that than to set goals that are achievable before redeployment? Existing in the middle of a long-term effort that already has poor metrics by which to gauge its success, without being able to pinpoint how you advanced the ball down the field, is hard.

There are certainly other negative impacts of the decision to deploy US forces in the manner that we did for the past twenty years. There might also be benefits. It’s hard to imagine, for example, a US populace that would have accepted long wars fought by soldiers and units deployed indefinitely, so the rotational model arguably gave US leaders the time to pursue long-term political objectives—even if that pursuit didn’t achieve much.

Of course, this model and its World War II counterpart weren’t the only options on the menu. In the Vietnam War, the individual replacement policy kept Army units largely in place in the country, with those units manned by a perpetual churn of new soldiers. Perhaps a model that similarly enabled institutional knowledge to be preserved, while also leveraging the comparative benefits of an all-volunteer force over a conscripted one, would have achieved more. In retrospect, though, the rotational model looks almost inevitable. Units were initially deployed to both Afghanistan and Iraq for short periods of time because the specified objectives were presumed to be achievable in short periods of time. When they proved elusive—and when those objectives changed to incorporate what amounted to nation building—the US presence needed to be extended. One unit replaced another, inertia played its role, and twenty years passed. Ultimately, if Iraq and Afghanistan were the types of wars that could be won with enough weapons and enough people to fire them, rotating new units into each theater would have worked. But they weren’t those types of wars, and it didn’t.

In any case, the point is not to argue that a rotational warfighting model is inherently bad—even if, in the cases of Iraq and Afghanistan, it clearly didn’t work. Rather, the point is this: we cannot afford not to learn from the past twenty years of wars. Too often, the passage of time chips away at an insightful reflection on the US military’s experience, or a sensible and useful distillation of complex subjects, or a narrative that pithily described an operational environment or strategic challenge, until it becomes detached from the deeper understanding on which it is founded and obscures any apparent need to dig deeper. It could be “twenty one-year wars.” Or it could be “fight tonight,” or “modernization,” or “great power competition.” What the US military needs most is dedicated servants who don’t stop asking what it means, challenging assumptions, recording their views, and continuing the discussion.

John Amble is the editorial director of the Modern War Institute at West Point and a veteran of the wars in Iraq and Afghanistan.

The views expressed are those of the author and do not reflect the official position of the United States Military Academy, Department of the Army, or Department of Defense.