A few days ago Shaikh Yamani gave a rare interview to the BBC. In it the former Saudi oil minister predicted that oil prices might peak at more than $100 a barrel, but in the long term they could just as easily drop back to the sort of low levels seen in the 1990s.

This is about as far from received opinion as you can get.

Most economists have been talking for a while of "a new paradigm" - a world in which oil prices settle into a stable price band of $50-60 a barrel.

But then again, these same economists were confidently predicting a $20-25 price bracket back in 2003, before oil prices started to climb. And climb. And continue to climb.

From the Gulf to the East Coast of the US, senior bank economists and oil market wonks scrambled to revise their forecasts.

With oil prices now edging into the 90s, you can hear the collective scratching of heads again. Should they hoist the price band higher? To $60-70, perhaps?

Whatever figure they pluck out of the air, you can be sure there will be a whole list of reasons they give to support their forecast.

On the one hand, there are the factors that keep oil prices high. These include declining fuel stockpiles in the US, the onset of winter or the American "driving season", political tensions, the growth of the Chinese economy and so on.

On the other hand, there are the factors that help bring prices down: an increase in Opec production, for example, or a declining US economy.

And just to complicate matters, there are the numerous unforeseen factors that keep oil markets on their toes: hurricanes in the Gulf of Mexico, or workers' strikes in Nigeria.

The list gets longer all the time.

Back in Shaikh Yamani's heyday - during the first oil boom and the 1973 Arab oil embargo - oil prices tended to reflect a simple formula of supply and demand.

Globalisation has complicated matters. These days, markets are generally more complex and have much fuzzier boundaries.

The use of futures contracts makes the picture fuzzier still. Because no trader really knows what events might influence the cost of oil three months down the line, any and every little piece of news today can have an effect.

Like the proverbial butterfly that sets off a tornado with one flap of its wings, a Turkish air raid in Iraqi Kurdistan can add 12 cents to the cost of a barrel.

At the end of the day, predicting oil prices is a mug's game.

Yet no economist would ever admit this. For a start, they would be out of a job. Because their real purpose is not to spin accurate forecasts, it is to conjure an atmosphere of confidence and calm.

Companies or governments need to be reassured that their latest multi-billion-dollar refinery or real estate project is going to work. They need to know the oil boom will last. They need to hear oil price predictions given in a strong, confident voice.

So it takes someone with the long experience and independent-mindedness of Shaikh Yamani to say that prices could just as easily go up as they could go down - to admit, effectively, that he has no idea what direction they might go in. But you can be sure he will say it in a strong and confident manner and he will charge you a hefty fee for his services.