That dilemma is whether to put Labor first and an independent second, or an independent first and Labor second.

It depends on what sort of seat you are in. If it is a Labor-held seat, it will not matter. But in a Liberal-held seat where there is a reasonably strong progressive independent, the difference can be critical.

It sounds counterintuitive, even mad, but in the latter circumstance the most diehard Labor supporter should put the independent first and the Labor candidate second to get an overall better result for Labor.

What? How can putting Labor second rather than first improve Labor's overall result?

Here's the theory.

Where there is a strong independent, the fate of the sitting Liberal (who usually polls first but with less than 50 per cent of the first preferences) will depend on whether Labor or the independent comes second.

If Labor comes second, rather than third, the Liberal will win. This is because polling and electoral history suggest the independent's preferences will split at a ratio around 65-35, with 35 per cent going to the Liberal. That is usually enough to push the Liberal over the 50 per cent line to victory.

If, on the other hand, a bunch of strategic Labor supporters put the independent as number one and Labor as number two. This could result in the independent coming second and Labor third. The Labor candidate would be eliminated, and the preferences distributed. About 90 per cent of Labor's preferences would go to the independent and less than 10 per cent to the Liberal, polling and electoral history suggest.

The 90 per cent of Labor preferences would push the independent over 50 per cent and give the independent the seat. The less than 10 per cent of Labor preferences going to the Liberal would not be enough for victory.

Here's an example where just one Labor diehard putting the independent first denies the Liberal Party a seat which the Liberal Party would win if the Labor diehard put Labor first.

Say the results come after our diehard has put Labor first on her ballot in an electorate of a tad over 100,000 voters. After the exclusion of a few shrapnel candidates, the result of the remaining 100,000 votes is 45,000 votes to the Liberal (45 per cent), 27,499 to the independent (27.5 per cent minus one vote), and 27,500 votes to Labor 27.5 (per cent).

In this instance, the independent is excluded because she has one fewer votes than Labor and her preferences are distributed. They go 65 per cent to Labor and 35 per cent to the Liberal candidate. So 35 per cent of 27,499 is 9625. Those 9625 get added to the 45,000 first-preference Liberal votes making 54,625, or 54.62 per cent, and the Liberal is elected.

But just say the diehard Labor voter has read this column and puts the independent first and Labor second. What happens then?

The independent has 27,500 first-preference votes (one extra which came from our converted diehard) and Labor has 27,499 (down by the diehard's single vote). The Liberal has 45,000.

In this case, the Labor candidate is excluded. The Liberal gets only 10 per cent of the 27,499 preferences - that is, 2750. When added to the Liberal's 45,000 first preferences, it totals 47,750 or 47.75 per cent. The independent gets 90 per cent of the 27,499 Labor preferences, which is 24,749 votes. When added to the independent's 27,500 first preferences, it comes to 52,249 votes or 52.25 per cent. The independent is elected.

This example, which proves the point, uses just one vote and gets a fairly large difference in final outcome - so there is plenty of room to move the parameters to more realistic scenarios and get the same overall conclusion. Labor voters should vote strategically in Liberal-held seat where there is a strong progressive independent if they want a worse overall result for the Liberals.

Labor does not actively urge its voters to put independents first and Labor second in these key seats, but it is definitely "running dead" in them - resourcing little and providing low-profile candidates, hoping that the Labor candidate comes behind the independent and gets excluded before them, so the Liberal is denied the seat.

READ MORE:

All of this definitely does not apply in the Senate. Voters can safely put the major party of their preference first, because in the Senate preferences flow both ways: up from excluded parties and down from any votes over the quota needed to get elected.

Importantly, in the Senate voters should vote for all preferences, whether you vote above or below the line, so your vote does not exhaust before the count moves to the last seat. If you restrict yourself to just six above the line or 12 below, your vote is very likely to have exhausted before the count between the last two candidates is decided - and those two candidates could be as disparate as a Greens candidate and One Nation candidate, or between one of the major parties and a minor party, and your vote will not be there.

In the ACT, a lot has been made of the possibility that the Liberals' Zed Seselja or Labor's Katy Gallagher might lose to an independent.

But for the past 50 years, ever since the territories got two senators each, the major parties have won one each in each territory. It was designed like that.

A candidate needs just 33.3 per cent of the vote after preferences to win one of the two seats. My guess is that Gallagher will win easily, by either getting a quota on first preferences, or in the unlikely event she fails that, via a few Green preferences to help get her over the line.

Seselja, on the other hand, is almost certain to fall below quota on first preferences, as he did by a bit in 2019. His trouble is that this time, of the nine minor parties from whom he might get preferences, only two are on the right: the anti-vax "Informed Medical Options" party and the UAP. So unless he gets more than about 28 per cent of the first-preference vote, he will be in big trouble.

If the 50-year record of one-each-for-the-majors is broken and David Pocock wins, it will be part of the radical change coming over Australian politics in which independents get to hold significant power.