Key Points:

There has so far been a rule under MMP - support parties' vote shares fall at the next election.

For instance:

* New Zealand First in coalition with National, 13.3 per cent of the party vote in 1996, 4.3 per cent in 1999; supporting Labour, 5.7 per cent in 2005, 4.1 per cent in 2008.

* Alliance in coalition with Labour, 7.7 per cent in 1999, 1.3 per cent in 2002.

* Progressives in coalition with Labour, 1.7 per cent in 2002, 1.2 per cent in 2005, 0.9 per cent in 2008.

* United Future supporting Labour, 6.7 per cent in 2002, 2.7 per cent in 2005 and 0.9 per cent in 2008.

The Greens did better while in a support arrangement - 5.3 per cent in 2005 to 6.7 per cent in 2008. But their support was only by abstention and they were often in opposition.

The message is, don't get too close. Tails on governing dogs get smaller, not bigger.

Act nevertheless reckons it can double its 3.6 per cent in this year's election to 7 per cent in 2011.

To do that it will need to: 1) slipstream a popular Government which converts still more defectors from the other side, 2) hope National's centrism spills despairing votes its way, or 3) convince about 80,000 voters that it has demonstrably influenced National for the better in the economy and/or in personal safety.

The first is unlikely. Very few governments - Labour's first and fourth did manage it - win a higher vote after one term in power. Labour's fifth did lift from 38.7 per cent in 1999 to 41.3 per cent in 2002 but relocating Alliance voters account for that.

Act's best hope is the last option, to morph into a social conservative party, for which our liberal policy settings have left space and on to which it could tack radical economics in a minor key. Rodney Hide has taken the party partway there. But he is not there yet. That leaves the second option.

But a rise in Act support in 2011 might trigger a bleed from National across the divide - centrist voters fearing radical economics - or reflect a National bleed across the divide occurring anyway, as Labour bled to the Greens this time.

Helen Clark calculated that risk of triggering a bleed across the divide in 2002 and 2005, and chose partners centrewards of Labour over parties more radical than Labour.

So if National were to have to choose pre-2011 between Act and the Maori Party, it will take the Maori Party.

In any case, National spies what it calls a "sea-change" opportunity not only to win general-roll Maori middle-class votes but to lock in Maori Party electorate voters who vote Labour with their party votes - thus, in essence, purloining Labour votes.

If you count the Maori Party's party vote as being on the Labour side, that side's total party vote among parties which won seats was 44 per cent to the National side's 49.5 per cent. A modest swing in 2011 could reverse that, especially if New Zealand First's 4.1 per cent in 2008 broke mainly Labour's way.

Some see in this a bias in the MMP system for Labour. In fact, in 2002 Labour admirers of Scandinavia excitedly imagined they might be on the way to emulating that region's near-perpetual social democratic governments.

They were wrong. Most United Future and New Zealand First voters leaned to National and they went back in 2005.

Clark was nearer the mark in asserting a relatively fine balance between the two sides. In other words, as the centre is continually redefined by governments, oppositions and circumstances, the weight shifts from one side to the other, as under first-past-the-post. The MMP system doesn't have a bias - voters do.

Yet National wants shot of MMP and will run a referendum. That is partly to appease its backwoods where MMP is considered a devilish trick. But National also knows that a more electorate-based system - such as supplementary member (SM) - would heavily advantage the two big, old parties which win almost all electorates.

A two-vote SM system on the basis of this year's voting figures, with the party vote proportional only of the 50 list seats, not the whole 122, would have had National governing alone (65 seats). A one-vote system assigning proportionality on electorate vote shares would advantage the two big, old parties even more, because they win more electorate than party votes. So would any reduction of list seats' proportion of the total.

An SM system would also advantage National over Labour because Labour piles up votes in dense, lower-income safe electorates. In 1978 and 1981 Labour won more votes but National governed. First past the post did exhibit a systemic bias.

One small party would do well under SM. Its use this time would have added a list seat to the Maori Party's five electorate seats, thus enhancing its potential leverage in a tight race between the big two.

But for Act and the Greens and other small parties with more than just a single electorate seat, MMP deals a far better hand. But only, history says, if they don't play pin-the-tail-on-the-dog.