Despite a persuasively convincing imitation, the Howard government isn’t yet dead in the water, the dismal polls notwithstanding. It could still scrape back, but minus Howard.

There are simply too many variables to be absolutely cocky about a large Rudd victory.

If we use the polls as a guide, we have the following order of probability:

1. A very large Rudd victory

2. A large Rudd victory

3. A narrow Rudd victory

4. A tied result

5. A narrow Howard victory

6. A large Howard victory

7. A very large Howard victory

Let’s apply the Rumsfeld logic to the situation (and try very hard not to end up with Iraq as the answer). There are known knowns; there are things we know we know. We also know there are known unknowns; that is to say we know there are some things we do not know. But there are also unknown unknowns – the ones we don’t know we don’t know.

What we do know is that Howard is on the nose everywhere; WorkChoices is a stinker. What we also know is that he lies and that interest rates continue to rise. But what we know we don’t know is how this will translate into seats won, lost and held on Saturday night – the only thing that matters.

We know that we know that many people make up their minds in the last week, even on the day – perhaps a quarter of the electorate. We know that we don’t know how many of these have responded to the polls and whether that response was a provisional or a final answer. We also know that we don’t know how even the swing will be.

Rudd needs 16 seats to win government, but let’s assume the Liberals pick up the two ALP held marginals in the west, Cowan and Swan, which looks at least possible. That makes 18 seats for Rudd to seize.

Clearly there will be a swing of sorts and a not insubstantial one. We can with some confidence therefore write off in likely falling order the following 16 government seats: Kingston (SA), Bonner (Qld), Wakefield (SA), Makin (SA), Braddon (Tas), Wentworth (NSW), Bass (Tas), Moreton (Qld), Solomon (NT), Lindsay (NSW), Eden-Monaro (NSW), Bennelong (NSW), Dobell (NSW), Deakin (Vic), McMillan (Vic) and Corangamite (Vic).

That first tranche of possible marginal losses falls with a swing of 5.3 per cent or less, but it excludes two government-held marginals in WA, Stirling and Hasluck, which would fall to Labor with swings respectively of 2 per cent and 1.8 per cent.

At this point, the ALP would need to pick up two more seats – and if the government manages to hold one of the above, then three more seats.

Watching the count at home on television, the benchmark seats then become the crunchers: Boothby (SA) with a swing of 5.4 per cent, Page (NSW) 5.5, Blair (Qld) with 5.7 per cent, and La Trobe (Vic) 5.8.

If Labor takes these, then government is Kevin Rudd’s – but only just.

Should there be a continuing swing against the coalition, the following seats requiring a swing of under seven per cent become the ones to watch:

Herbert (Qld), Paterson (NSW), Kalgoorlie (WA), McEwen (Vic), Longman (Qld), Cowper (NSW) Sturt (SA) and Robertson (NSW).

Anything above this in magnitude will see Labor in with a comfortable majority and could even threaten Peter Costello in Higgins, requiring a swing of 8.8 per cent to fall.

But these are simply the unknown unknowns lurking in the twilight zone of the voters’ minds. The task for Labor historically and practically is not easy; it verges on the unprecedented.

A revised order of probability using the Rumsfeldian method might read:

1. A large Rudd victory

2. A narrow Rudd victory

3. A narrow Coalition victory

4. A tied result

5. A very large Rudd victory

6. A large Howard victory

7. A very large Howard victory

It is not therefore entirely inconceivable that the Coalition could yet be back with, say, a two-seat majority.