On to part 4 in my series of comments on the New Scientist magazine series “Eight things you didn’t know about the internet”: “Is there only one internet?”, by Ben Crystall.

As with all the questions so far in this series, this one brings up a series of subordinate questions, the answers to which are not simple. The most basic of these is how one defines “the Internet”, and how one decides its boundaries. Mr Crystall starts with this:

Probably — for now. The internet is a disparate mix of interconnected computers, many of them on large networks run by universities, businesses and so on. What unites this network of networks are the communication languages known as the Transmission Control Protocol and the Internet Protocol, collectively TCP/IP.
He goes on to suggest that human-language issues, dealing with Russian and Chinese character sets, could fracture the Internet.

It’s hard to say what will actually happen in that regard. The IETF has been working diligently on internationalized domain names and email address internationalization, to try to stave off such fragmentation. Indeed, the mechanisms are in place, and there already are domain names registered in Russian, Chinese, Korean, Japanese, Hebrew, and other languages that use character sets that used to be incompatible with the Internet’s Domain Name System. The internationalization protocols allow such languages to be represented in ways that are compatible, but that can be displayed in the native formats... for the most part, everyone is happy.

On the other hand, even if Turkey can have domain names in Turkish, what happens when they decide to block access to YouTube? And China... even though China has more Internet users than the U.S. does, what happens when they, too, block YouTube, along with other web sites, including the New York Times? Is it still all one Internet then?

Apart from the censorship question, we clearly create parallel networks all the time. Corporate firewalls, for example, break off pieces that are otherwise physically connected. When I worked at IBM, I had access to services on the “big-I Internet”, as we called it, as well as those on the “little-i intranet”. We used the same protocols, and we were connected to the same Internet servers and services. Yet now, from outside the firewall, I can no longer get to what used to be my internal web or email servers.

Doesn’t that mean that there’s a separate Internet in IBM, and in most other large companies?

It’s likely, too, that even as a customer of a typical Internet service provider, if you try to send email directly to an email server on the Internet, you’ll be blocked. You’re expected to go through your ISP’s email infrastructure, an attempt to clamp down on zombie computer “botnets”.

The reality is that what we call “the Internet” comprises a collection of distinct networks that use common network protocols, and that more or less agree to cooperate. And, in fact, there is a parallel network called Internet2, set up for academic and research use (as was the original Internet, back in the day).

So... the answer is “Yes and no.” There’s certainly one, single “Internet” that we all think of as the place we go to for Google and Amazon, Facebook and YouTube, the New York Times, the BBC, and the Sydney Morning Herald. But, ultimately, it’s made up of countless little networks, along with the standardized “glue” that holds it all together.