Millions of computers and users, nobody in charge. Why isn't it a disaster?
You are staring at the Aardvark Preservation Society's home page: The ".de" at the end of the address tells you that the page is stored on a computer that is probably in Germany. At the bottom of the page is a reference to the Aardvark Liberation Front's home page stored on a computer in Surinam (country code ".sr"). You click on the link and you find yourself confronted with the news of the Front's latest raid.

With some 10 million host computers, around 40 million users and no central authority, the Internet should be a disaster. Yet here you are looking first at a page stored in a computer in Germany, then at one in Surinam. How is that possible?

The answer is that the Net is a surprisingly organised creation that works rather like a giant Post Office. Information whizzes around the globe via electronic sorting offices with postcodes being checked against the biggest postcode directory in the world.

The Internet is what is called a packet switched network. This means messages are broken down into small chunks of data, ranging from 296 bytes up to a typical maximum of 1,064 bytes.

Each packet has an address where it must be delivered. Much as the Post Office will take a letter from point A to point B by various routes depending on circumstances, so these packets are dispatched round the Internet via various communication links. If one link is too busy, another is tried. It might be a more complicated route, but, like the post, the message eventually gets through.

When you click on an address, known as a URL (Uniform Resource Locator) on a Web page, your Web Browser breaks down the address and looks for the host name, the name of the computer out there on the Internet to which it must connect.

In a typical URL, such as, the host name is the part between the first two sets of "/"s, in this case

Once your Web Browser has extracted the host name, a message is sent to a large computer at your Internet service provider called the Domain Name Server (a domain covers a group of computers at an organisation). The DNS searches its database for the host name that you have typed in and looks up the computer's address. This is a numeric address which the Internet "understands", roughly like a house address. So it will define country, city, street and house number in Internet terms.

Once the host name has been converted into an address by a DNS, a message is sent out over the Internet saying that you are requesting a certain page of information stored on that host. So that the information can find its way back to you, you are assigned a temporary address which is sent with your request for information. This is like sending a stamped addressed envelope when you send off for a fact sheet from a TV programme.

The messages normally move around the Internet along leased lines owned by the various Internet access companies. In the UK, most of the major Internet service providers have their main telecommunication systems connected together at Telehouse, a bomb-proof telecommunications and computing back-up centre in London Docklands. Here, in a system called the Linx, all the telecommunications lines come together and messages flow between the various British networks. If you are requesting information from a foreign server, the request passes on to other countries' equivalents of Linx until the eventual host is contacted.

This connecting is done by special communications computers called routers, which work like telephone exchanges or, continuing the postal analogy, like sorting offices. From the type of address, the router knows roughly where the destination computer is and passes on your request down the line.

So, you might be requesting a page from a computer in the Far East. Much as a letter from London to a small town in northern Scotland might be partially sorted in a sorting office in London and then further sorted on a night mail train, so the electronic message might be passed on to Japan for further sorting or redirecting.

Each router studies the way that messages have flowed through the various connections it makes to other routers, and chooses what it thinks will be the best route to the destination.

When your request for information finally arrives at the host computer, it contains your return address. The page you are looking for is located, sent off in packets and eventually, a few seconds later, it is delivered back to your PC in packets.

Unless you move on to another page or send a "stop" message, the whole page is sent. Then the remote computer forgets you exist. Any connection you make by clicking on a URL, whether it is to the same computer you have just accessed or another one on the other side of the world, is handled as a completely new sequence of messages. Many people believe that they first connect to one computer, then from that computer by clicking on a URL, they connect on to the next, to the next and so on. That is not the case. It is as if you are sent your fact sheet and that fact sheet contains addresses for further information.