Let me say at the beginning that I do not think that the existing mess in Afghanistan at present is the fault of the Obama administration. The president inherited it from George Bush, and simply did not have time between taking power in January and the Afghan elections of this month to carry out a radical change of course. If, however, the administration fails to change course after the (predictable) debacle that these elections have become, then the responsibility for subsequent disasters will indeed rest with President Obama and his team.

The Afghan election has lessons that go far beyond Afghanistan. It illustrates the folly of relying on democracy and elections to provide solutions to complex issues of state-building, absent a whole set of other preconditions. One of these is for Washington to have a clear idea of what election results it wants, what election results are possible and what if anything it can do to influence those results.

Instead, both the Bush and Obama administrations drifted along with the Afghan electoral process, the results of which were always going to be a choice between the very bad and the absolutely disastrous; and were then going to have to explain to the American public and the publics of key U.S. allies (notably Britain) why bringing about this awful choice was worth the lives of dozens of U.S. and allies troops. It now seems likely that more British soldiers have died in Helmand province over the past four years than Afghan citizens voted there in the first round of these elections. How do you explain that to those soldiers' parents, wives and children?

The reasons why people did not go to the polls in Helmand, despite the recent military operations to drive back the Taliban, are perfectly obvious--and should have been perfectly obvious long before the elections. They are summed up in one simple question: Would you, dear reader, risk your life and those of your family to vote for Hamid Karzai?

On the one hand, the threat from the Taliban to voters and their families was not diminished by the latest military operations because it is a permanent threat. In the perception of the people of the area--and almost certainly in reality--sooner or later U.S. and British troops will return to their bases; and then the Taliban will return to the villages and impose ferocious punishments on those who voted. On the other hand, why run this risk for President Karzai, a man for whom Western officials in private--and even sometimes in public--have nothing but contempt, and whose administration, after eight years in power, has done precisely nothing for most ordinary Afghans?

But Karzai was always going to be the good alternative in these elections (given that, unfortunately, Ashraf Ghani never stood the slightest chance of winning). The only real alternative was and is Abdullah Abdullah. His past membership in the Northern Alliance (responsible for numerous atrocities against ethnic Pashtuns in the 1990s and after the overthrow of the Taliban in 2001) makes him detested by a great many Pashtuns. His election would almost certainly lead to a new wave of Pashtun support for the Taliban, the collapse of whatever remains of government authority in many Pashtun areas, and in the long run would very likely point toward the partition of Afghanistan along ethnic lines. All this is quite apart from the fact that the election was always going to be subject to massive rigging (to be fair, by both sides), with a high probability that the results would lack real legitimacy in the eyes of many or even most Afghans.

If the United States persists in declaring that all of this notwithstanding, the elections have been legitimate and democratic, then Hamid Karzai will survive in power. But he will lead an administration that will be even more crippled, ineffective, and beholden to warlords and drug barons than it was before the elections. And as a result of Karzai's re-election, any really new U.S. strategy for Afghanistan might have to be postponed for another five years, until the next Afghan presidential elections--which will come after the next U.S. presidential elections. If only for its own chances of re-election in 2012, this is something which the Obama administration would be very foolish to go along with.

Instead, in my view, the Obama administration should adopt the following strategy: If the legitimacy of Karzai's victory is seriously challenged by Abdullah to the extent that the June Iranian elections were, then Washington should agree that the results lack credibility and cannot be recognized, and use this as an excuse to move from a presidential to a prime ministerial system. A loyah jirga (grand national assembly) can be called to legalise this constitutional change. A caretaker prime minister can then be appointed who would attempt to bring more conservative Pashtuns into his cabinet. If Washington decides that this is a step too far and Karzai must remain as president, his loss of authority and legitimacy should nonetheless be made the basis for forcing him to appoint a prime minister to assume most responsibility for the actual running of the government.

This should be accompanied by a move to legalise the presence of political parties in the Afghan parliament, prior to the parliamentary elections next year. The Taliban should be actively encouraged to form a political wing and to take part in these elections--along the lines of the bizarre, but in the end very helpful, system in Northern Ireland, where even at the height of the British campaign against the IRA, its political wing, Sinn Fein, remained a legal party and stood for election.

Negotiations should be opened with the Taliban high command on a peace settlement in Afghanistan, the offer on the U.S. side being the promise of an American military withdrawal within a long but fixed timetable, conditional on progressive Taliban ceasefires across increasing areas of the country. Where the Taliban does not agree to a ceasefire, military operations should continue. Pakistan should be used as an intermediary between the United States and the Taliban leadership.

In other words, the United States should pursue a strategy of talking and fighting at the same time. The goal would not be an early peace settlement or an early American and Western military withdrawal, since the first is impossible and the second would be disastrous. Rather, this strategy would recognise that it is to a great extent the Western military presence that is driving support for the Taliban in many of the Pashtun areas.

This strategy would seek both to diminish this support and engage the Taliban leadership by the promise of withdrawal, while leaving Afghanistan's long-term political future open. Our approach to the future shape of Afghanistan could then be guided by two developments or the lack of them: The emergence of the Afghan National Army as a force which is militarily effective and--equally importantly--genuinely multi-ethnic, in the sense of fully representing the Pashtun sections of Afghanistan's population; and reconciliation between different political and ethnic forces in the country, including the Taliban.

If these two conditions can be met in the course of a long (say five to ten years) negotiating process, we can continue to aim at some form of truly united Afghanistan. If they cannot be met, we will have to accept de facto partition along ethnic lines, even if we do not call it by this name.

Unlike the present U.S. strategy, this would involve talking to Mullah Omar and the Taliban leadership, initially through Pakistani intermediaries, later directly. The present strategy, of trying to get bits of the Taliban to surrender to the Karzai regime, is by contrast doomed to failure. After all, if there are two obvious "bits" to aim at, allied with the Taliban but not part of it, they are Gulbuddin Hekmatyar's Hezbe-Islami party and the tribal network of Jalaluddin Haqqani and his family in Pakistani Waziristan and Afghan Khost.

But Hekmatyar and Haqqani were offered huge amounts of American money to join the Western side in the autumn and winter of 2001-2002, when U.S. power and prestige were at their height--and they refused. Why should they agree to come over now, when the United States and the Kabul regime look so much weaker? The present strategy, of trying to break off local Taliban commanders, reminds me of something Wellington said before the Battle of Waterloo, when he was asked if he expected significant defections from the French Army. He replied something to the effect, "No-one worth a damn. Maybe a couple of Marshals."

I hope that I am wrong, but as far as I can see, sooner or later we are going to have to recognise the twin facts that the Taliban are the most important political force in the Pashtun areas, and that it is to a very great extent our presence which makes them so. This recognition most certainly should not imply a rapid and humiliating Western scuttle from Afghanistan. Equally, it should imply the abandonment of empty talk of remaining in Afghanistan for a generation, eliminating the Taliban, and building an Afghan democracy.