Just when we thought it was safe to think that Hillary Clinton would comfortably win the US presidential election and that, for better or worse, the US and hence the world would continue with business as usual, reality has come back to bite us on the assumption. Clinton is still likely to win but, reflecting an underlying fragility in her potential voter base, states that were leaning in favour of Clinton have weakened a little in their support for her.

Clinton is currently campaigning in Florida, given that, in a tight electoral race, this is the state that will likely tip the balance in favour of the winning candidate. Having Michelle Obama at her side has proven to be a great electoral asset. Indeed, many potential Democratic voters would rather vote for Michelle Obama than Hillary Clinton.

Setting aside the increasingly dynastic character of US politics implied by this preference, Hillary’s husband and a couple of recent Bush presidencies, M. Obama is seen as untainted by the grubbiness of political participation. She also does not carry the baggage that Hillary has been obliged to drag around with her since Bill was Arkansas governor.

Setting aside alternative preferences, it is Hillary, not Michelle, who has to win this race. At this point, less than two weeks out, Clinton is polling ahead of Trump in Florida by around 3.5%. This is better than trailing by 3.5% but, when right-libertarian candidate Gary Johnson’s vote of 2% is factored in, that margin is looking less solid.

Similarly, the other must-win state, Pennsylvania, is still favouring a Clinton win, but it has also softened, if slightly less so than Florida. Some observers in the US have described this as Trump picking up momentum, but it seems less that he has gained momentum than the underlying cynicism that has informed this campaign and voters’ perceptions of it is playing out as inconsistent voter intentions.

[Rundle: how Trump remade the GOP and unmade America]

While chances of winning still favour Clinton, Trump’s chances have grown in the last couple of days from around 10% now, according to some assessments, up to 17%. This is still a weak position, but it does show that Trump is not entirely out of the race.

One factor that has not been included in most poll assessments is that, setting aside normal poll variability, is that while polling among registered Democratic and Republican voters is likely to be accurate, non-registered voters are much more volatile in their voting intentions. Some non-party registered voters have been changing their minds on a regular, almost daily, basis and could, quite possibly, change their mind on way to the polling station.

Trump’s campaign is now playing to this voter disenchantment, which is the foundation of such appeal that it does have. Rather than have a positive message for voters, knowing it has around 30% or so locked in, the Trump campaign is seeking to even the playing field by discouraging those inclined to support Clinton from turning out on election day.

This then becomes a campaign of negative attrition. If enough former Bernie Sanders voters can be persuaded that Clinton does not represent the values they want to see in the White House, and to therefore stay at home, non-registered voters turn out in decent numbers and decide they, too, are tired of “politics as usual” and the “Trump rump” remains locked in, despite intentions expressed to pollsters, the election might be closer than most pundits think.

For those who do not favour the “let’s shake it up and see what happens” school of political outcomes can take heart from two factors. The first is that Trump and Clinton were running more or less neck and neck in the polls in mid to late September but, since the presidential debates, this is no longer the case; Clinton now has a substantial lead, and while that might diminish, it probably won’t drop sufficiently to allow a Trump victory.

The second factor, which plays best to the most cynical (or realist) observers, is that the book-makers are offering odds of around 1.2:1 for a Clinton victory, but 5:1 for a Trump victory. What they are saying is that, in a two-horse race, Trump could “do a Bradbury” if Clinton crashes in the next 12 days. But, assuming she can stay upright and there are no other major changes, to borrow a line from Damon Runyon: “The race is not always to the swift, nor the battle to the strong, but that’s the way to bet.”