I see that Justice Minister Simon Power considered Dame Sian Elias' prisoner amnesty idea for all of five minutes before dismissing: "The Government makes the law on behalf of New Zealanders who elect them, judges take that law and apply it. And that's the end of the matter."

Elias is only the Chief Justice, what would she know? Far better to listen to the Sensible Sentencing Trust's Garth McVicar, whose knee jerked predictably in the same direction.

In fact, there was nothing in our most senior judge's carefully worded lecture that other criminal justice experts hadn't already said. For those familiar with the territory, it was neither outrageous nor unusual.

In the US, for example, as many as half the states are considering early release schemes as a response to prison budget deficits and overcrowding. As one sociologist has said, the research is "completely clear" that the move won't increase recidivism or crime rates. Only the politics get in the way; "People just don't understand the basic math here."

The US is the world champion of incarceration, locking people up at the rate of 756 per 100,000. Its population accounts for 5 per cent of the world's total but 25 per cent of the world's prisoners - which means "either we're the most evil people on earth or we're doing something wrong", says Democratic senator Jim Webb, who's pushing for a radical overhaul of America's prison system.

According to Webb, the American criminal justice system is broken; he thinks Americans need only know how to count to understand their prison problem.

This should give us pause considering we've been hell-bent on emulating them. Longer sentences, three strikes and you're out, overcrowded prisons - whatever problems the Americans have, we're headed in the same direction.

Already we imprison at a higher rate than Australia, the UK, and Canada. The prison muster has grown from 7500 a year ago to 8400. A crisis looms. If we carry on as we are, we'll reach an imprisonment rate of 200 per 100,000 (or just under 11,000 prisoners) in 8 years.

To be clear, Elias accepts that "retribution is a proper response for serious crime". Even the most rabid prison reformer agrees that the truly bad and dangerous should be locked up. There's no debate on this; everyone wants less crime, safer streets. But is our tougher, more vengeful approach working?

In the US, according to the Sentencing Project, much of its 600 per cent rise in the jail population since 1972 can be attributed to changes in drug laws and the passing of numerous "tough on crime" laws intended to put more people in prison for longer. One analysis of imprisonment rates between 1980 and 2001 concluded the growth was related to changes in sentencing policy.

Which would be fine if more prisoners equalled less crime, and harsher sentences really deterred criminals. But the evidence doesn't bear this out. A major international analysis of 50 research studies involving more than 300,000 prisoners has found that prison doesn't deter, and longer sentences increase the likelihood of reoffending.

Indeed, prisons are largely irrelevant to reducing crime and making communities safer.

Supporters of harsher sentences will often cite falls in US crime rates to bolster their case, but the reality is more complicated. While it's true that crime rates fell in states with above average increases in imprisonment rates, states with below average increases in imprisonment actually experienced a greater decline in crime rates.

For example, "tough on crime" Texas had a 144 per cent rise in imprisonment between 1991 and 1998, and its crime rate fell 35 per cent. But New York's crime rate declined even more (by 43 per cent), despite an increase of imprisonment of only 24 per cent. New York continues to experience historic lows in crime and a declining prison population; and may now have to close four prisons.

One UK study estimates that if we increased the prison population by 25 per cent, we could achieve a 1 per cent reduction in the crime rate.

But at what price? The average cost of keeping an offender in prison is nearly $100,000 a year, while the average cost of a community-based sentence is just over $10 a day. It bears repeating that the more we spend locking people up, the less we can spend on things that make a real difference. One estimate puts the cost of a child at school as a quarter of that needed to lock up an offender.

Elias' common sense approach includes a focus on community education, early intervention, improved mental health and drug and alcohol treatment, more community-based sentences, and improvements in the probation service.

It's not rocket science, or even radical. But we can't get there without community engagement and understanding. Politicians like Power are driven by electorate expectations and anxiety. Elias says we need to accept there are no silver bullet fixes, "that risk cannot be eliminated and that the costs we are absorbing to try to do so are disproportionately expensive" - and that real progress requires social change.

Which means we need to know more than how to count to understand our prison problem.

Elias' lecture was given in honour of noted criminal defence lawyer Shirley Smith, who argued that we needed to find out "why blameless babes become criminals".

Smith wrote, "To reduce crime it is necessary to identify what makes criminals and deal with the causes ... This is the only long-term, effective way to help victims, to reduce their numbers. Punishment does not work."