Who makes the political weather? Who are the shamans who interpret the mysteries of the stars to those mortals who merely mark a cross on the ballot paper or (in the United States) pull the lever on the machine?
For Republicans in the state of Michigan who decide today whether to prolong John McCain's political resurrection, give Mitt Romney some consolation for his money, or take Rudy Giuliani for a final whirl, there is no time just now to ponder such apparently elevated questions. Nor is there any urgent need. Last week in New Hampshire the pollsters and the reporters could not be faulted. They called the polling precisely as it was.
For the Clinton-Obama show, however, which has the luxury of a few days' respite before its next outing in Nevada, what went wrong with the forecasts in New Hampshire should still be a burning question. Why did the advance polling and reporting conflict so sharply with the reality on the night?
The explanations have veered from the accidental – the Rumsfeldian "stuff happens" – to the last resort that it was all the fault of quixotic voters who changed their mind. The blogosphere throbs with allegations of fraud.
I have my own favourites among the explanations doing the rounds. Barack Obama exerted more JFK appeal as the underdog (as he was in Iowa) than as a presidential nominee in waiting. He was the modish choice, so more people told the pollsters they would vote for him than did. Perhaps Hillary's breaking voice played a part, in suggesting a new vulnerability. Or maybe her machine was simply underestimated in its ability to get out her core vote.
If just the advance polls had been wrong, such explanations might suffice. But the exit polls were equally wrong. And the only plausible explanations I have heard are these. First, that racism played a greater role once Obama had become front-runner and that some voters refused to admit their Clinton vote even after the fact. And second, that the sampling system used was geared to a past pattern of voting and failed to cope with a sharp increase in the Hillary-generation turn-out.
Unacknowledged racism is hard to deal with – although French pollsters factor in a percentage for shy National Front voters (which has the added advantage of often making the far-right result seem disappointing on the day). The sampling of voters by demographic group, however, is something that can, and may, need to be adjusted.
In the US, as in many countries with universal suffrage, women outnumber men as registered voters, and older women vote in larger numbers than any other group. Perhaps in an election where women voters believe that a woman has a chance of victory, the voting gender gap will widen.
New Hampshire apocrypha suggests an unsuspected element of female solidarity. Tales are told of women who went to vote, or changed their vote, at the last moment to counter the slights and bias they perceived to be directed at Mrs Clinton – the same slights, they felt, they had experienced themselves.
And the bias they say they felt, where did it come from? Not from the pollsters, but from the media. Watch US network television, scan the by-lines in the mainstream US newspapers, and count the women. They are "allowed" to report, especially if they look good, and one woman is accommodated on most talkshows. This is, after all, the 21st century.
But political punditry remains a man's game – an establishment man's game, what is more. And when the front-runners are identified, it is the (usually male) "stars", who follow the bandwagon. This year's Democratic primaries are male-female contests, in which the coverage is disproportionately filtered through the lens of male experience and expectations. In this, you might observe, the scene is not a whole lot different from here. Who are the political editors and observers who set the tone for mainstream politics? Half a century of feminism has left the two elite circuits for political journalism, Washington and London, with men easily outnumbering women. As Washington correspondent during the 2000 election, I was one of only a few "girls" on the campaign bus. In Britain, the number of women covering mainstream politics seems actually to have declined compared with 15 years ago.
Perhaps this does not matter. Perhaps a good pundit is a good pundit, as observant and perspicacious as the next. My sense, though, is that it does matter; and that as long as the imbalance obtains, the pack mentality and the presumption of a shared male experience will persist. If this is so, then the good news from New Hampshire is that the women used the ballot box to fight back.