The Internet’s Many Branches

The evolution of the top-level domain, a solution to a technical problem that has evolved into something with lots of cultural and marketing value.

By Ernie Smith

Today in Tedium: Last week, nearly lost amid all the talk of reopening states prematurely due to a pandemic, digital activists and nonprofits alike won a major victory for the internet: ICANN blocked the $1 billion sale of the .org top-level domain to a private-equity firm. A lot of stuff happened to get to that point, including outdoor protests while it was still safe to do those, but perhaps the lynchpin that might have killed the deal was the input of California Attorney General Xavier Becerra, who sent a message just before a pivotal decision. Becerra’s input, taking on more impact given the fact that his state was in the midst of a pandemic at that time, might have changed the minds of some key stakeholders at the Internet Corporation for Assigned Names and Numbers. Top-level domains have always been complicated and complex like this, in part because the original goals of these domains have frequently been subverted. Today’s Tedium dives into the evolution of the top-level domain. — Ernie @ Tedium

Be sure to check out our sponsor, Rype—they have an idea for something you can do during your forced downtime.

Rype

Now’s the perfect time to learn a new language. Try Rype.

Staying home to work these days? There’s never been a better time to take advantage of your time at home to learn a new language online. Take live 1-on-1 lessons on your phone, tablet, or computer with professional teachers from around the world. Get 7 days free today.

1985

The year that the Domain Name System was first introduced. The first top-level domain at the time was .arpa, which was intended basically to manage the internet’s infrastructure and serve as a bridge between ARPANET, the internet’s predecessor, and the newly global internet. (The top-level domain, which was intended to be temporary, ultimately stuck around and is used as the basis of modern-day reverse DNS searches.) Five other top-level domains came to life shortly thereafter: .gov, .edu, .com, .mil, and .org.

0508 hosts

A passage from a HOSTS.TXT file from 1983, before the Domain Name Service or TCP/IP protocol were implemented. (via ttkzw’s site)

Why the top-level domain system works the way it does

Do you have a list where you keep everything?

Perhaps you just write stuff down in it, just to remind yourself to cross it off. You do it by hand because it’s no big thing.

It’s a great way to manage, say, a grocery list, or a list of chores. But one can see the clear problems in managing a list of something really big. Like, say, the entire internet.

But this was once how the domain name system was managed—literally by hand, in a HOSTS.TXT file on a centralized server that was distributed to individual users, until the late 1980s. ARPANET was small back then, with just a few hundred servers.

But the internet was destined to grow bigger, and eventually this approach would prove too much for the internet to manage.

Fortunately, the organizers of the internet had ways to work on issues like this, and one of the most important was the RFC, or request for comment. This document format, dating to the earliest days of ARPANET, allows programmers and engineers to discuss technical, philosophical, and strategic issues in a consistent format.

And the RFC was the way that the internet’s earliest engineers formulated a way out of the HOSTS.TXT system.

Starting in 1983, engineers working on ARPANET at the University of Southern California’s Information Sciences Institute came up with a new organizational model for addresses that they called domains, which would eventually become one of the most important elements of the internet.

0508 paul

Paul Mockapetris, the inventor (with Jon Postel) of the Domain Name Service. (via GlobeNewswire)

Paul Mockapetris, who authored RFC 882, the initial document that laid out the need for domains, pointed directly to the challenges caused by relying on a centralized HOSTS.TXT file, as well as the increasingly complicated nature of sending an email on the internet.

“The size of this table, and especially the frequency of updates to the table are near the limit of manageability,” he wrote. “What is needed is a distributed database that performs the same function, and hence avoids the problems caused by a centralized database.”

0508 tree

Think of this giant tree as the internet, and all of the branches sticking out of it as different levels of domains. (Nareeta Martin/Unsplash)

The method he laid out took on the branches of a tree. Though we don’t really think of it this way now, the domain system was developed using a “tree” system, with each node representing a different, increasingly tiny branch. At the base of the tree were a handful of branches—the base top-level domains—with those branches breaking out into thousands and eventually millions of individual nodes, called secondary domains. Those secondary domains, managed by companies or individual users, can have subdomains, and those network nodes could be used for whatever—FTP servers, telnet, Usenet servers, or what have you. Many of the protocols we came to later use, including the World Wide Web, had not been invented yet.

Which gave the result something of an open feel. In RFC 920, which laid out requirements for what a top-level domain could look like, the approach, as explained by authors Jon Postel and Joyce K. Reynolds, felt fairly amorphous:

Domains are administrative entities. The purpose and expected use of domains is to divide the name management required of a central administration and assign it to sub-administrations. There are no geographical, topological, or technological constraints on a domain. The hosts in a domain need not have common hardware or software, nor even common protocols. Most of the requirements and limitations on domains are designed to ensure responsible administration.

(Postel and Reynolds, side note, were two of the internet’s most important early figures, helping to organize and manage a system that would eventually outgrow them by leaps and bounds. As I wrote in 2015, Postel died weeks after the creation of ICANN, the organization that was launched effectively to replace him. Before that, he was largely a one-man band, with the occasional help of Reynolds.)

In practice, domains would turn out to be much more than administrative entities: signifiers of identity, important status symbols, branding essentials, the starting point from which we used the internet. Famous battles have been fought over domain names.

But in the 1980s, a much larger goal was more practical: To decentralize the process of connecting to a host. It had a technical reason to exist, not just for reasons of vanity.

“The big deal,” Mockapetris told Wired in 2012, “is that it allowed everyone to manage their own networks and change the parameters of their networks and have those changes be visible everywhere.”

During these early days, there were just a handful of requirements for administering a top-level domain, including the existence of a “responsible person” to manage the domain; the use of a server to manage and control the domain; a minimum number of hosts (the document recommends at least 50, but says it’s a “soft” total); and an official registration with a domain registrar.

But despite all this, the top-level domain would only be handed out sparsely for the first decade and a half of its existence, with a focus on country codes in particular. With few exceptions beyond physical location, the original five user-accessible top-level domains—.gov, .edu, .com, .mil, and .org—would become the primary domains that people would see in the English-speaking world.

Slowly but surely, though, the internet grew from a digital communication system to a basic element of our daily lives, and by the time the top-level domain system had seriously been reevaluated to allow for different types of names, some of the prior top-level domains—particularly .com—had become immensely crowded places.

0508 symbolics

The digital museum that now sits at Symbolics.com, the first dot-com domain ever registered.

Five notable examples of early domain names

  1. The first non-.arpa domain. Early domains generally were either existing technology companies, universities, or had some tie to ARPANET. The very first domain registered, nordu.net, represents education and research networks through the Nordic countries.
  2. The first dot-com. On March 15, 1985, the .com top-level domain came online with the registration of symbolics.com. The namesake firm, which produced early computer workstations dedicated to the programming language Lisp, was one of a small handful of technology companies that was given access to a single-entry domain name in 1985, with public registration only announced the next year. The domain, which recently celebrated its 35th anniversary, now serves as a museum of early internet history. Symbolics, which had ties to the defense industry, shut down in the mid-1990s.
  3. The first dot-org. Registered by the MITRE Corporation in 1985, mitre.org represents a federally funded nonprofit that supports research and development centers around the country.
  4. The first educational domains. On April 24, 1985, a group of five domains were handed out to some of the United States’ best-known universities: the University of California, Berkeley; Carnegie Mellon University; Purdue University; Rice University; and the University of California, Los Angeles. Despite the concept of the top-level domain originating from the University of Southern California, that school didn’t get a top-level domain until August of that year.
  5. Early domains registered by individuals. During the earliest years of the Domain Name Service, domains were generally registered by companies or institutions, but at least two domains went to people: Broken.net, registered to a programmer named Jason Matthews in 1986 (and currently inactive); and toad.com, registered to digital activist John Gilmore (a cofounder of the Electronic Frontier Foundation) in 1987.

“Trademarks are registered in a system that permits many companies to share a name legitimately without interfering with each other, such as Sun Photo, Sun Oil and Sun Microsystems. Domain names only permit one user of a name; there is only one sun.com, which Sun Microsystems registered first. Neither lawyers nor governments can make ten pounds of names fit into a one-pound bag.”

— John Gilmore, a cofounder of the Electronic Frontier Foundation and the owner of the very early domain toad.com, in a 1996 letter to the editor in The Economist. Gilmore’s point reflects the complex nature of trademark in the domain name era, and while there have been some legal remedies in both national and international law over the years, the fact is that it requires active monitoring to prevent related domains from falling into the wrong hands.

0508 bird

In this metaphor of a picture, you’re the bird and you’re surrounded by internet. (Matt Seymour/Unsplash)

How the top-level domain situation became complicated

Initially, the ownership of the top-level domain system rested with the federal government through ARPANET, but as the internet grew, management of the top-level domains went to different institutions.

This as in part driven by both demand and shifting laws. While the internet was originally intended for purely government and academic pursuits, despite all the technology companies that staked a claim on the network early on, in practice, nothing really stopped companies from connecting to and promoting services on the internet, and by 1992, the Scientific and Advanced-Technology Act effectively ended the de facto ban on commercial use.

Over the years a series of companies gained and retained ownership of various top-level domains. For many years, the primary registrar of the three primary top-level domains—.com, .net, and .org—laid in the hands of Network Solutions, which was acquired by Verisign during the early 2000s essentially to gain access to the company’s domain management functions. (Fortunately, registering these domains is no longer in the hands of a monopoly.)

Over time, some domains switched hands; in the interest of putting the top-level domains with owners who made sense for their respective missions, the .edu domain was handed to the nonprofit Educause, while the recently contested .org domain went to the Internet Society-affiliated Public Interest Registry.

But after the gold rush of the 1990s, the internet’s basic domains started to feel fairly full. They still do. It’s hard to find a .com domain name of five characters or fewer that hasn’t already been registered. As a result, new types of top-level domains started to trickle out, with seven decided on by ICANN in 2000: .aero, .bit, .coop, .info, .museum, .name, and .pro.

Later efforts to democratize the process of registering for top-level domains, launched by ICANN starting in 2008 and expanded upon by a 2011 board vote, proved controversial to observers, because of the “land-grab” nature of the process and the apparent favoritism the approach offered to companies. If you had $185,000, you could get a top-level domain, which meant that prospectors with enough money could take control of entire top-level domains—and many, such as the domain registrar Donuts Inc., did.

“The new ruling means that from this year, we may well start to see sites ending in .pepsi, .apple or .microsoft,” a 2012 Wired article warned. “A number of companies have already laid a stake in the ground regarding a particular term, with Canon making it very clear that it wants .canon, while the Mayor of London has announced that he is interested in the .London suffix.”

(Each of the generic top-level domains mentioned in the quote were eventually registered, by the way, with the exception of .pepsi—the soda giant didn’t see the value in it. Not that anyone is actually using them at this time.)

The attempted sale of the .org domain to a private equity firm stood out because of what it represented: a top-level domain name associated with a specific kind of integrity suddenly put into the clasping maw of private equity. Few top-level domains were built with integrity in mind, but .org had turned out to be one of them.

So when the Internet Society tried to sell the Public Interest Registry, it created a backlash because of what it represented. Fortunately for fans of integrity, the internet was strong enough to fight back.

$5M

The amount that Verisign pays the tiny nation of Tuvalu each year to administer the .tv domain. The domain, which has become popular with streaming sites, is a huge boon for the country, whose largest industry, the licensing of territorial fishing rights, brings in $19 million per year. “As sites utilizing .tv grow in prominence, Tuvalu’s domain on the web may eventually supersede that of its seas,” The Washington Post wrote last year.

0508 libya

Libya was in the midst of a civil war just as interest in its two-character top-level domain was peaking. (Public Domain Pictures)

Should you register a domain from the Libyan government? Probab.ly not.

At some point the pressure created by the heavy use of the original top-level domains for commercial use found a release valve of sorts when some creative entrepreneurial types took advantage of the fact that country codes offered slightly fancier ways to stand out than .com did.

These domain hacks were clever and fun, and effective branding for new companies. Perhaps the most effective over the years has been .ly, which adds a slight twist of whimsy to many English words, an attractive thing when your goal is effective branding.

The problem, of course, is that the country of Libya has traditionally been a political hot spot, in part due to its longtime leader Muammar Gaddafi, whose influence carries on into the tensions of an ongoing civil war dating back to 2014—a followup to another civil war that culminated in Gaddafi’s violent 2011 death. Of course, when people register a domain, they don’t think about any of this.

And that proved a problem for Bitly and other sites that used these top-level domains. Bitly, of course, had a technical reason for its use case—it was trying to minimize the number of characters used in a link at a time when 140 characters was a big deal. During the 2011 conflict, Libya at times attempted to shut off its internet, which led to concerns that the .ly sites would stop working—a problem when every tweet on Twitter at the time was using the link-shortening service.

Fortunately, as CNN Money wrote, there was never a real risk:

The root servers connect a user to the server that the website is hosted on. But those servers don’t always physically exist in the countries their TLDs are assigned to. Sites with the .ly TLD have five root servers, but only two are based in Libya: The other two are located in Oregon, and one is in the Netherlands.

For a site like bit.ly with a .ly TLD, all five of the root servers would need to go down at the same time for access to the site to be restricted. It’s nerdy, but integral to understanding why bit.ly kept chugging along when the Libyan government killed the Internet.

That wouldn’t have been true a few years prior, though. In 2004, as Priceonomics notes, the domain was put into ICANN’s hands after an incident in which all of the domains disappeared.

0508 bitly

Bitly, as it appeared in 2010. (via Ready Designs)

Still, the political and social implications likely played a role when Bitly started redirecting to a dot-com domain instead. (Other domains weren’t so lucky—Letter.ly’s registration had expired with no easy way to resolve the problem, while vb.ly, a URL shortener co-owned by the sex columnist Violet Blue, was taken offline for moral reasons.)

The Libya domain hack even affected presidential campaigns. In 2011, at the height of tensions in Libya, Mitt Romney’s campaign announced that it was dropping his Mitt.ly domain in favor of Mi.tt, a much friendlier domain from Trinidad and Tobago.

(It could always be worse—as Art.sy found out in 2013, when it dropped its top-level domain, administered by the Syrian government, in favor of a .net domain. In its case, a civil war did knock it offline.)

These two-letter country codes were doled out based on international standards and are effectively good top-level domain names for “domain hacks” by chance. Only a handful of region-intended domain names, such as Catalonia’s .cat, have maintained their original regional and cultural goals.

Of course, when these country codes were put into place, they were not intended for commercial use in this way. It was a happy accident that, in some cases, turned not so happy.

In the case of my own domain, Tedium.co, people often ask me why I didn’t go for the dot-com. The short answer is that it’s not for sale. The firm that owns it, Reflex Publishing, has a reputation for acquiring valuable domain names, sitting on them, and refusing to sell them.

They own some of the world’s most valuable domain names, tedium.com included. Do I need the extra m at the end of the name? Not particularly, especially given that their domain doesn’t really go anywhere notable at the moment.

They’re in it for the long game, and the long game ultimately has a lot of value for them. Game respects game.

Fortunately for me, the government of Colombia has been pretty laissez-faire about letting their country code being used for non-Colombian reasons, and actively promotes its country code for the very use case I use it for.

The top-level domain, over the 35 years of its existence, has evolved from a solution to a technical problem into big business, then into a useful political tool, then (more recently) into a rallying cry.

You probably pay no mind to the two or three letters at the end of most of the domains you put in, but those letters are the branches from which the rest of the tree that is the internet grows.

--

Find this one an interesting read? Share it with a pal! And definitely give Rype a good look if you want to learn a language or two.

Ernie Smith

Your time was just wasted by Ernie Smith

Ernie Smith is the editor of Tedium, and an active internet snarker. Between his many internet side projects, he finds time to hang out with his wife Cat, who's funnier than he is.

Find me on: Website Twitter