Eric's Beijing Diary

The recent UN summit on the Internet has given the world body a new picture of some of the peculiar features of that inscrutable monster.

I am not sure why the Europeans are so interested in managing the Internet domain structure, unless it is just the fact that they don't trust the Americans. But the push to "internationalize" the management of the Internet displays a misunderstanding of just how the Internet works. We often say that "nobody owns the Internet." This is true in terms of content and who can or cannot be involved. But from a technical standpoint, the Internet structure is inherently hierarchical. This is because the Internet has millions of hosts, and every one of them needs to have a unique address. We call this address an "IP address," because it is a function of the Internet Protocol. So the first problem is to make sure that every address is unique. But there is another problem. For users, Internet addresses are not identified by a number. They are identified by "friendly names." Hotmail's current IP address is 64.4.32.7 So if you open Internet Explorer and type in that number, you will get Hotmail's website. But nobody does that. There are a couple of reasons why we do not use numbers to identify websites. First, they are hard to remember. We do have internet search engines, of course, but most people have a group of websites they need to go to frequently, that they need to be able to remember easily. The second reason, is that sometimes IP addresses change (perfectly exemplified by the problems I have been having with my student web site). For a small company, this could be something as simple as changing an Internet Service Provider. If websites were only identified by numbers, then every time something like this happened, clients of this website would somehow have to be notified of the change. But because we use friendly names, it is a simple matter of changing the DNS mapping. The mapping between the friendly name and the IP address is made once on the DNS server, and the end users don't know the difference. Much, much simpler, but someone has to be in charge of this. The agency is called Internet Corporation for Assigned Names and Numbers, a non-profit organization charged with assigning IP addresses to Internet hosts.

When you type in the URL for a website, your browser issues a request for a mapping. That is because in order to reach the website you are looking for, your computer needs to know the IP address. The request for this mapping goes first to your local DNS (Domain Name System) server, and is then passed up the line to the root server. The root server takes care of mapping for the root domain, and refers to the top level domains (.com, .net, .org, etc.). There are 13 or 14 root servers positioned in various parts of the world. (I can't remember all the locations, because I am not teaching Microsoft systems here in China, so I don't have any DNS servers set up.) Each domain is then responsible for the mapping below that level of the hierarchy. For example, microsoft.com would manage the mapping for support.microsoft.com. So the organization takes care of assigning domain names, and mapping IP addressed to those names. The technical management of this whole process cannot just be picked up and moved at will.

In addition to all this, there must be a structure to facilitate the development of internet technology. This would be the Internet Engineering Task Force. The IETF is a very informal organization. The IETF does not go out and recruit engineers to develop the Internet. They just show up. Engineers who care to be involved in some aspect of Internet development choose to participate in one of the many working groups managed ( I use the term loosely ) by IETF. They meet together, drink coffee, shoot the breeze and come up with ideas. As casual as these confabs are, they do generate the ideas which become the technology that drives the Internet. Eventually, of course, this technology needs to be codified, and this is done in the form of RFC's (Request for Comments), such as RFC 1548, the original document establishing the Point-to-Point protocol which you probably used to connect to the Internet to read this.

Occasionally, the writers thereof are in a bit of a nonserious mood (did I say "occasionally?") and they come up with stuff that is a bit far-fetched, such as the "Coffee Pot Protocol," or the "Carrier Pigeon Protocol."

The bottom line, really, is that the Internet is just very, very California. And it always has been. The Internet was developed by the US. Military in 1969 (back when Al Gore was a little boy). But the US Military doesn't do stuff like that on their own. They elicited the help of universities like U.C. Berkeley, just as they worked with the University of Chicago in development of the Atom Bomb during World War II. There was quite a bit of technodevelopment coming out of California during this time. The Palo Alto Research Lab, funded by Xerox, developed such innovations as windows (not capitalized, since it was not a brand name then) and the mouse. Unfortunately for Xerox, the executives in the board room were too thick-headed to appreciate the value of such innovations as a little box with a ball in it that made a dot move around on a screen, so they discarded it. Bad for them, and good for Apple, and eventually Microsoft, who stole it from Apple. But I am digressing...the point is that the tremendous volume of creativity and innovation that resulted in such phenomenal developments as the Internet is not likely to be surrendered to a few dunderheads in the EU who don't like to have the Americans dominating anything. And their wrong anyway. The Internet is not an American innovation. It is a West Coast, Californian innovation. I think most people who know anything about the history of the Internet (including Secretary General Kofi Annan) would be uncomfortable with the idea of moving Internet management to Europe. Personally, I would be uncomfortable with moving it to New York.