The answer, or at least part of it, is found in Malcolm Gladwell's brilliant piece on underdog strategy. In order for an underdog to win against a superior opponent (a la David v Goliath), they almost always have to play the game differently. The Cuban Revolution is perhaps the textbook example of this; unsurprising, since Che wrote the book. There is a lot that enabled the Davidian triumph in Cuba - a climate that allowed for tenacity and endurance, a public polarized, the aggressive repression of the General Strike making revolution the only viable alternative, and the relative newness of oppression all helped mobilize people against a "goliath". So the traits the Davids in conflicts value (tenacity, moral, and mobility) were all possible, while the goliathian nature of Batista's Cuban Army was rather minimal. That the revolutionaries chose to fight a guerrilla war matters little - victory was impossible without it, and was all but guaranteed fighting that way.
Bolivia, on the other hand, lent itself less well to this style of warfare. Che's best soldiers were Cuban and inexperienced with the region. Che set up far from the politicized masses, and instead operated amongst xenophobic peasants (this limited mobility needed to offset a larger military). US support was offered early and often to Bolivia, making Bolivia's army far better at beign a Goliath than Batista's army was. And it was very, very hard for the kind of tenacity that had enabled success in Cuba to prevail in Bolivia without popular support and mobility. Not that Che didn't try, but whereas in Cuba momentum built and the army was able to both sustain and expand its efforts, in Bolivia the revolutionaries became more and more hindered, forced on the defensive, and unable to ever get momentum going their way. When the final confrontation came, it was between US-trained Rangers, who are as close as goliathian powers get to matched tenacity and mobility against an irregular force. Combined with, in this case, superior weaponry, numbers, and control of the field, Che didn't really have a chance.
The case can be made that because Che was unable to get a guerrilla movement going in Bolivia, he was fighting more conventionally, but with a grossly mismatched force. While that claim holds merit, I think it overlooks the additional hardships forced upon his warband by trying to fight a guerrila war. It's exhausting, and it changes engagements from those forced upon you in defensive positions or actively sought out with superior strength, to the possibility of being engaged at any time. Armies fight conventionally, even when they are grossly mismatched, because it is easier to do. Forts, stable supply lines, brief high-intensity engagements all are less draining on an individual soldier, and on an army, than being active all the time. Fighting battles like that means you have to fight fewer of them. Washington and Vo Nguyen Giap, both noted practioners of Guerilla warfare from their respective eras, were both eager to transition their army from guerilla tactics to conventional - early on, that transition often led itself to a series of defeats (outmatched when playing Goliath's game, after all). But why did they consistently choose it?
Gladwell, using sports as one of his main examples, lets social pressure be what undermines the usage of the David-esque full court press. Well, that, and the sheer difficulty in sustaining the kind of pressure that kind of gameplay requires. Social pressure is a hard factor to argue for war - war, being rather existential, is not the arena where general's care about what the opposing commanders must think of them. So the second explanation must be the one that allows us to understand Washington and Giap. While both could win a guerilla war indefinitely, they had a hard time sustaining it. Fighting conventionally meant putting ones soldiers through a few big battles, rather than existing in a state of constant warfare. Fighting big meant battles with clear-cut results, and the temptation of immediate and significant victories is potent. More meaningful than trophy victories, however, is the nature of what soldiers have to fight. If you are fighting all the time, you need lots of soldiers willing to be immediately ready to press every advantage. Skill is less relevant, as tenacity and endurance are the meaningful quantities. But that's exhausting when it's a 48 minute basketball game, and moreso when it's an actual war. Skill, on the other hand, can be improved at a less breakneck pace, and enable one to win the few engagements one fights in. When Goliath armies beat David armies, it's because they've gone ahead and adapted to Davidian war. But Goliath armies will always, always prefer the Goliath-on-Goliath combat.
The relevance of this (besides the sheer awesomeness that is the Malcolm Gladwell article and the Steven Soderbergh film) is that the US military is, and has been for at least two decades now, the impossible-to-defeat Goliath. In Gulf War I, Iraq had the world's 4th largest military. The war was grossly one-sided, and the United States lost 113 soldiers. When everyone plays by the winners rules, the winner will be incredibly hard to stop.
The wars the US is fighting today do not favor Goliaths. The tenacity of forces in Afghanistan has been demonstrated - the Taliban are back, when by all accounts they were ruled dead by 2002. In order to win, as the US did by aiding Bolivia, they have to make overwhelming force applicable to this kind of combat. There are generals currently doing just that - perhaps most famous is David Petraeus, while the most interesting developments are coming from "The Accidental Guerilla" author David Kilcullen. They are the ones formulating the tactics that will win this.
The challenge, as has often been leveled agaisnt the US, is a willingness to fight the kind of sustained warfare needed. For the Taliban, they don't have a choice - the war is existential. For many factions in Iraq, their survival depends on mastering guerilla conflicts. For the US, we have to be willingly to commit to a war with no set-piece battles. And this calls into question the whole role of war with humanitarian ambitions, or of international police action. Iraq is abhorred by the left because it was unnessecary. It is very much a victim of attention fatigue, and perhaps a worthy victim. Afghanistan, however, is a conflict that the US seems more committed to, without an immediate existential threat. The Taliban are as much foreignners in the country as the Cubans were in Bolivia, and the public longs for stability. The possibility of an eventual US victory is real. The challenge, of course, is sustaining the effort needed. The Taliban have to do it. We don't.
Editor's Note: Since writing this post, a few other examples of the attitude needed towards Afghanistan have come to mind. Over at doubleX, Vanessa M. Gezari has a comprehensive and concise post up about the change in generals for the Afghanistan conflict. From later in the Salon review of Kilcullen's book comes these amazing quotes, which more or less sum up all that is needed, and all that is difficult about the US operating in Afghanistan.
Security is the single overriding need of people living in weak states, and Kilcullen points out that rule of law doesn't fall too far behind that. (It's especially difficult for Westerners who have no cultural memory of living under chaotic conditions with no viable civil authority to grasp just how scary this can be. We tend to fixate on the menace posed by authoritarian tyrants like Saddam Hussein, forgetting that a person killed in a civil war is just as dead as someone killed by a dictator.) One service the Taliban has excelled in providing to Afghans in isolated regions, for example, is dispute resolution. When you're quarreling with your neighbor about who owns those goats, if there isn't some authority, however merciless, to appeal to, things can degenerate into a Hobbesian state pretty quickly. The Taliban first gained power in the 1990s by offering just this sort of adjudication, and they haven't forgotten how well it worked. Unless the West helps the Afghans set up better civil institutions, they'll use it again.
Kilcullen makes it clear that efforts to set up viable governance in places like Iraq and Afghanistan must involve established local power structures (like Iraq's tribes) and customs (like the elaborate Iraq blood-debt resolution ritual known as the "suhl"). The neoconservative pipe dream of making over Iraq and Afghanistan as Western-style democracies has to be set aside. Most difficult of all, homegrown police, politicians, judges and other officials who aren't either corrupt or pursuing a purely sectarian agenda have to be put in place. Kilcullen thinks this can be done, but it will be really difficult and really expensive. And it will take a while. And the result is unlikely to be the sort of government Americans admire.