The basic premise of the peak oil theory seems unremarkable. Oil is a finite resource and therefore almost everyone agrees global oil production must reach a peak at some point, before going into decline.

The controversy arises over two questions: When will it happen? And what are the consequences?

Professor Richard Heinberg claims to know the answers. He has produced four books and a DVD on the subject of oil depletion, and has given more than 300 public lectures all over the world. He was recently invited to New Zealand as a peak oil expert by (among others) the Green Party and Engineers for Social Responsibility.

In his books, he answers the when question on the basis of a modelling approach devised by oil geologist Dr Marion King Hubbert. In 1956, Hubbert used his approach to predict the 1970 peak of US oil production with remarkable accuracy. Using similar methods, contemporary peak oil theorists are now predicting a global peak in the first decade of this century.

In fact, Professor Heinberg believes that this event may have already happened. Production of conventional oil seems to have peaked in May 2005, he says, and all liquids including natural gas liquids, tar sands and even biofuels seem to have peaked in July 2006.

But apparent peaks in production have happened before. As Professor Heinberg himself points out, global oil production seemed to have conclusively peaked in 1979 before making a dramatic resurgence in 1984. Similarly erroneous claims for peak oil were made in 2000.

And, at a global level, Hubbert-type methods have repeatedly made false predictions forecasting peak oil to occur as early as 1989 (and in most years thereafter).

Part of the problem is that such predictions rely on an assessment of the amount of oil that can ultimately be taken from the entire globe. Although Professor Heinberg maintains that this is two trillion barrels by most estimates, some respected geologists have put the figure at nearly double this value. The US Geological Survey's mean estimate is over three trillion barrels.

Furthermore, Hubbert-type methods involve fitting equations to historical oil production data. In the comparatively unregulated US market this is fairly straightforward, but political and economic influences make it more problematic for the global situation.

These factors have led to criticism of the modelling methods of peak oil theorists. Cambridge Energy Research Associates, a US-based energy consultancy, is damning in its assessment, saying that peak oil theory is garbage. Highly-respected energy economist Michael Lynch has described peak oil theorists as practising pseudo-science and claims that: "The quantitative models used by peak oil theorists would earn a university student in elementary statistics a failing grade".