It is a chilling cyber-scenario as set out by Singapore government minister Ho Peng Ke. Instead of a backpack of explosives, he has warned - justifying new laws against "computer misuse" - a terrorist could create just as much devastation by sending "a carefully engineered packet of data" into the computer systems which control essential services, such as the power stations.

The supposed menace of "cyber-terrorism" has become the subject of alarming predictions of this kind since September 11, most of them originating in the new sense of vulnerability felt in the US. The subject conjures an image of Dr No come back to life with supreme keyboard skills. In the words last year of US homeland security director Tom Ridge: "Terrorists can sit at one computer connected to one network and can create worldwide havoc."

Not surprisingly, the first place where this anxiety has been translated into heavy-handed law is Singapore. The government there now has the discretion to take "pre-emptive action against computer terrorism", on the say-so of the security agencies, without ever being called to account. On Monday, some MPs feebly protested their fears that the new Computer Misuse Act might be... misused. Then they passed it on the vague assurance of Mr Ho that they should not "worry about abuse".

As more utilities become dependent upon computer programmes, the opportunities for cyber-mischief have multiplied, but the threat is now being wildly exaggerated. One hypothetical scenario envisages a cyber-terrorist targeting children through the production computer of a cereal manufacturer, which is instructed to increase the iron content to a dangerous level. In reality it would taste too bitter to eat.

It is prudent to ensure that vital systems are, in the jargon, "air-gapped" against external intervention. But one of the best safeguards is still the human eye: no critical service such as air traffic control should be allowed to function without human monitoring. Unchecked error, not terror, is the most likely threat.