China is a permanent problem for our national defence, one that will only get worse. This month it's taken a step toward establishing itself militarily in the South Pacific.

We can't assume our US alliance is a permanent solution. No one knows whether Washington will want to protect us many decades from now.

Australia must itself become stronger. The issue comes down to this: defence costs money, a chunk of the national economy, and the strongest way to expand the economy is to expand the population.

There's no particular population level that will make us absolutely safe, but every additional taxpayer reduces risk.

The 26 million people in Australia today will be 43 million by 2066, give or take 6 million, according to a five-year-old forecast by the Australian Bureau of Statistics.

That implies average growth of 1.1 per cent a year, compared with 1.6 per cent since the beginning of the 20th century.

We should aim for much more, perhaps 2 per cent a year and a population 50 years from now of 70 million. The way to achieve this is obvious and simple: opening the immigration tap.

What's not at all simple is distributing that population and especially preventing Melbourne, Sydney and Brisbane from becoming unacceptably large.

I wonder if there is a country in the world that talks as much about decentralisation as Australia while doing so little of it?

Our biggest achievement in this area has been to build up Canberra to more than 400,000 people by supplying government jobs. Then there's Albury-Wodonga, which has reached about 100,000.

These are trivial numbers compared with the tens of millions we should add to the population for security - or even compared with the gentle growth the ABS is forecasting.

We need a really powerful decentralisation policy, one that will move hundreds of thousands of people per year.

The policy should be to rapidly expand many current towns and smaller cities to populations of 1 million or more each. We would do this successively, starting with, say, two locations. As their growth got under way, we'd shift attention to another one or two, and so on.

Most of these future big cities would be on the coast, because Australians obviously like to live near water, but some should be inland. Some could, in fact, be in outstandingly beautiful locations.

The map below shows 44 possible places for future major cities, each generally 100 to 200 kilometres from the next, leaving plenty of nature between them. Each (except maybe Darwin) has an acceptable climate - though we could argue about what is acceptable.

About one-third of the locations are in fact popular tourism destinations. Except for Darwin, their average summer maximum temperatures are no higher than 32 degrees, about the same as Singapore's year-round average.

Next, we need a force to drive their growth, and it must be financial. The way to get people into these smaller places is to offer deep tax discounts.

Suppose income-tax rates for people in the chosen locations were reduced by 24 percentage points. In that case, there'd be no tax on $90,000 a year, which is the average Australian wage without overtime.

To get jobs into the cities, the benefit would have to be shared with employers. They'd pay lower lower wages but, after tax, employees would still receive more.

Such incentives should be enough to get backing from locals when expansion begins, or at least those who are working.

The Treasury should be on side, too. There would be a cost to the budget, but it would be offset by the promise of more future taxpayers and avoidance of a crippling defence burden.

As each new city grew and became more appealing for new arrivals, the tax concession would be phased out, according to a schedule published at the outset. It would reach zero when the population had hit the target of about 1 million.

The starting concession for each place would depend on its starting size.

We would still want to encourage population growth in other places apart from Melbourne, Sydney and Brisbane, so we'd still need our current decentralisation policies, notably those that encourage immigrants to spread out.

MORE AGE OF THE DRAGON:

Especially where new major cities start as modest towns, they could be designed for minimal environmental impact. For a start, they could have strict perimeters, prohibiting sprawl and mostly use desalinated water.

Their area should be minimised by keeping housing at an average medium density. Most people would live in townhouses, but there would also be high-rises in the nicest locations, with some bungalows on the edges.

Excellent roads and public transport routes would of course be in the master plans, which would also provide for populations growing beyond the targets, perhaps to 2 million.

It should be possible to fit 2 million people into a coastal city that reaches only about 15 kilometres inland. The exact dimensions would depend on local terrain, including mountains and the floodplains of unbanked rivers, and the need to preserve areas for biodiversity.

Eventually, a fast-rail line could link the coastal cities from Cape York Peninsula to the Spencer Gulf.

Don't dismiss this because it seems too imaginative.

Maybe there's a better way to ensure distribution of a rapidly growing population. If there is, I can't think of it.

But I'm sure that for national security we must get more people as quickly as we can.

Australia, you have a choice: populate or perish.