The war in Afghanistan revealed significant advances in the US armed forces' ability to gather information and use it quickly to strike at an enemy. The so-called "sensor-to-shooter" gap had been narrowed considerably.

"I think there was immense progress," says Nick Cook, aerospace consultant to Jane's Defence Weekly. Before Afghanistan, the Pentagon's target was to be able to fire a weapon within 10 minutes of spotting a target.

"Now people are thinking of shorter periods than that."

According to President George W. Bush: "Our commanders are gaining a real-time picture of the entire battlefield, and are able to get targeting information from sensor to shooter almost instantly."

Gathering information about an enemy and using it to hit him before he can hit you is perhaps the oldest goal of military strategy. Before radios and aircraft, it could take months.

But the 10-minute target emerged after the Gulf war because defence chiefs were unhappy that they could not hit Iraqi missile launchers in the time it took to pull one out of a hiding place, fire a Scud, and hide it again.

In theory, advanced sensors, digital communications and precision weapons make it possible to reduce the gap to seconds. In practice, it is not so easy.

The Kosovo campaign in 1999 showed US weaponry was good at hitting fixed targets such as buildings and bridges, but less good at hitting moving targets: very few Serb tanks were destroyed.

The mix of technology, information and human decisions that should make it happen is known by one of the defence world's more complex acronyms: C4ISTAR - command, control, computers, communications, intelligence, surveillance, target acquisition and reconnaissance.

The challenge for modern commanders is to synthesise the mass of data available from many sources into a picture of the battlespace that they can understand - and especially one that gives them confidence that targets are correctly identified.

A second element is to circulate information across a network including anyone who might need it, including commanders in and outside the theatre of war, and aircraft in the air and ships at sea. Then, command and control must be structured so as to give quick authorisation to hit a target and give an order to fire a weapon.

Using digital and internet technology to do all this is known as "network-centric warfare". It is a central thrust of Pentagon policy and a favourite theme of Vice-admiral Arthur K. Cebrowski, appointed by Donald Rumsfeld, the defence secretary, to spearhead the planned Bush-era transformation of US military capabilities.

Though not a weapon in itself, defence experts see network-centric warfare as a "force multiplier" enabling much more effective and quick use of the formidable US arsenal. It also makes US forces less vulnerable to attack, because damage to one headquarters or ship will not disable the whole network.

Commanders in Florida ran the campaign in Afghanistan using data from an extensive bank of sensors ranging from satellites to human eyes.

They received images from satellites, U-2 spy planes, Joint Stars battlefield surveillance aircraft, Awacs early warning aircraft, a British Canberra photo-reconnaissance aircraft, and Predator and Global Hawk spy drones.

To these pictures were added signals and electronic intelligence gathered by US Rivet Joint and British Nimrod R1 aircraft, and eyewitness reports of US and other special forces soldiers operating alongside anti-Taliban forces.

Most of the data feeds arrive at an operational headquarters separately. Bob Hoffa of Northrop Grumman says: "It's like looking at a variety of TV screens and watching four different football matches at the same time."

Experts say there is a long way to go in overcoming this first barrier: massaging the data into a coherent picture.

Second, the quality of images needs further enhancement so commanders have no doubt in their minds when selecting targets. The next generation of sensors will allow much more accurate identification.

Third, the mass of real-time data has to be rapidly communicated. Bandwidth is a limiting factor, and will become more so as the number of unmanned spy planes multiply. More high-speed datalinks between aircraft are also needed.

Fourth, the role of human beings in the information chain needs study. Pilots of advanced fighters have on-board computers that sift information and make decisions for them. They know a missile is heading towards them, but they do not know how they know. Network-centric systems will do the same for all the information coming to commanders.

Thomas K. Adams, a former officer writing in the US Army War College Quarterly, said: "The explosion of available information inevitably results in information overload and flawed decision making . . . soon it becomes obvious that the slowest element in the process is the human decision maker."

Yet each strike must go through an authorisation process, sometimes reaching up to the highest level. Targets have been lost while necessary approvals were sought - but some officers would argue this is better than giving a mistaken order that might kill innocent people.

For commanders, this could become more rather than less awkward. In the world of network-centric warfare, the president may be watching the same pictures that they are.