Understand the structure of the domain name system before deciding what, if any, domain name(s) you want to register for your own site(s). Learn all your options such as subdomains of existing domains and alternative top-level domains -- one of these may suit you better than just registering another .com domain.
When the Internet was first created (as the ARPAnet, in 1969), its naming system was a flat namespace, where every location in the network was listed in a file (hosts.txt) that had to be updated every time anybody else entered the net. That was easy when there were only three computers on the net (as in 1969), or even a few hundred (as in the '70s), but as the network hosts grew into the thousands by the '80s, the "founding fathers" of the Internet decided to create a more logical system to name them. This was the domain name system. It's too bad that people these days seem to be trying to turn it into a flat namespace once again, coming full-circle.
The Introduction of Domain Names
Domain names were first proposed in the early 1980s as a structured naming system for the ARPAnet. The earliest official document describing their purpose and structure was RFC 882, written by Paul Mockpetris, generally regarded as the "father" of the domain name system, in 1983. This document has been obsoleted by later RFC documents (Requests For Comments, which despite their informal name are the place where many significant standards documents for the Internet reside), such as RFC 1591. The earliest document doesn't even have ".com", or any other of the current top level domains, in it; those came later. The basic definition of how the naming system worked came before any specific name assignments in it.
When the first top-level domains were implemented, around 1985, there was hardly a land rush to register them as there is for new TLDs now. Although registering a domain didn't cost anything, it required some technical expertise to get through the registration procedure, which required understanding how the system worked and setting up name servers which would respond to the new domain. No commercial Internet presence providers existed to do it for you! In fact, only institutions associated with government-sponsored research were allowed onto the net in the first place. That's why the original set of top level domains encompassed educational, governmental, military, commercial, organizational, and network-infrastructure entities, but provided no namespace for personal or hobby sites -- these were not expected to be a part of the network at all.
Of the slow trickle of early domain registrations, .edu names dominated, as universities were the big players on the net. In general, each domain name represented an entire network of dozens or hundreds of computers connected to the ARPAnet, or later the Internet, which were given hierarchically structured hostnames within the domain, like c.cs.cmu.edu for machine "C" in the computer science department of Carnegie Mellon University. Prior to the introduction of domain names, these machines had had hostnames in the hosts.txt file which had to be distributed to all systems on the net -- some attempts to introduce logical structure to these hosts had been made, using names like cmu-cs-c, but all such structure was informal and didn't give administrators at any level the ability to create names within their space without going through the global host table. Domain names, on the other hand, had a distributed system of servers where a request for c.cs.cmu.edu would proceed from the root server, to the .edu server, to the cmu.edu server, and finally to the cs.cmu.edu server which would have an entry for machine c. New names could be added at each of these levels by the administrator of its server, without having to consult others elsewhere in the hierarchy.
Soon after the generic TLDs were created, the two-letter country codes also were added, at the request of some of the handful of overseas entities with Internet connections who wanted a namespace that could be controlled on a country-by-country level without being subservient to the U.S. government. Although a .us domain was created also, for the use of the United States, it wasn't much used because Americans were already used to dominating the generic global TLDs.
We Haven't Had That Spirit Here since 1989
While a few .com addresses existed from the start, for companies involved in research projects along with the universities (apparently, the first .com domain registered was symbolics.com on March 15, 1985; it was then a couple of months before the next one was registered, think.com on May 24, 1985), true commercialization of the Internet didn't begin until 1989, when the U.S. government liberalized the rules to allow commercial activity unrelated to research. The first experience much of the general public had with domain names was when the commercial online services, such as CompuServe, began to interconnect their e-mail systems so that users could not only send mail to other subscribers of the same system, but also to other commercial online systems, and to the academic systems of the Internet. This required people to learn an address format new to them (though it had been in use for years by the academics) consisting of a username followed by an "@" (at) sign, then a hostname in domain form. Users of commercial online services found that their addresses ended in .com, as part of hostnames like compuserve.com. Users of the hobbyist network of dialup bulletin board systems, FidoNet, got addresses ending in fidonet.org (really long ones, like email@example.com, reflecting the hierarchical structure of FidoNet with zones, nets, and nodes), but, although FidoNet was immensely popular in the computer hobbyist community for a while, the commercial services got a wider general-public audience, so the dominance of .com in the public mindspace began to form.
Along Comes the Web
Then, in the 1990s, the World Wide Web exploded onto the scene. Beginning in 1990 as a system at European physics laboratory CERN to allow researchers in high-energy physics to share information, the Web started slowly (and for a few years trailed the menu-based Gopher system as the leading system for structured online information), but started a massive growth in popularity once the graphical browser Mosaic came out in late 1992. First it was used by people who already had access to Internet-connected computers, mostly college students, but then the mass media started noticing it, and this fueled a demand for commercial Internet providers, which started springing up in early 1995. Also, the online services like Prodigy, CompuServe, and AOL added Web browsers to their software, so suddenly a mass audience from outside the traditional "geek" circles was using the Web.
The Incredible Shrinking URL
When the masses first flooded onto the Web, most of the sites they browsed were still on academic servers, since few commercial web hosting providers existed. URLs tended to be long, technical things, like:
(I put a linebreak in the middle to keep this from stretching beyond some systems' window width, but this is really supposed to be one line.)
Immediately, people involved in developing for the Web started working to shorten and simplify this. Most of these simplifications made a good deal of sense, removing unnecessary complexities from the structure of URLs. However, in hindsight, maybe it would have been better if they hadn't done this, since this produced a drive to keep simplifying things even to the point of eroding some of the useful structures -- once people got used to URLs sometimes being short and simple and un-techie-ish, the developers of Web sites wanted all of their URLs to be that way, and didn't mind abusing the domain name system to do this.
To see how this progression occurred, look at the above URL and note the ":8080" -- this is a port number. Web developers quickly realized that if they put all Web servers in the default port, 80, they could omit that portion of the URL, shortening it to:
Then, by learning to use default index files so that they could give a directory name alone without a filename after it (something not all Web developers "get" even now -- see my Web Tips article), they can shorten it to:
Next, they could make sure that their main home page was in the directory the server pointed at, rather than more deeply nested in a directory structure that was designed when the Web was only one of many uses for the computer the server was on; this cuts a few subdirectory names out of the URL:
If "J. Smith" was just an ordinary user of that computer, this might be the furthest he could go at simplifying his URL, but if he were the system administrator, he might move his homepage to the server root and cut out the need for a "tilde" and username:
Now we're getting somewhere. At this point, the sysadmin may realize that the URL would get even simpler if he could create an aliased name in his department's DNS (domain name service) -- instead of just using the hostname of the machine the server happens to be running on, he can create the hostname "www" and make it equivalent to it. This lets the URL become:
Enough administrators did this that it rapidly became the "standard" for Web addresses, and soon hardly any of them were on hostnames that didn't begin with www. (In the long run, though, this proved to be a transitional stage; the trend these days is to drop the www and use the "naked" domain name with no hostname preceding it. This would have been unthinkable just a few years ago, while the Web was just one of many different services being used on Internet-connected machines, and didn't yet have the enormous prominence it now does -- administrators wouldn't have wanted to tie up the root domain of their site by aliasing it to their Web server in preference to the servers for all other services such as FTP or Gopher.)
At the same time, browsers were being made more intelligent at completing URLs that were missing some parts, since they could fill in the most common values themselves. In particular, the http:// part could be inferred. Also, people could get away with omitting trailing slashes (though in some cases this produces a wasteful extra server redirect). Thus, though it wasn't technically speaking a proper URL, people got used to addressing the above site as:
But by this time, .edu sites like this were becoming a minority compared to commercial sites, which were sprouting up at a rapid pace. Unlike the university sites, which were usually grouped by department in a hierarchical structure, the commercial sites were usually designed to promote the company and its products in a unified manner, and thus had addresses of the form www.companyname.com, not needing subdomains for departments and divisions, as these didn't usually have separate Web sites. Thus, it came about that the public expected Web addresses to be of the form www.something.com (or just something.com, as the www part became increasingly considered optional).
The Marketing Types Take Over
As the Web became a part of the mainstream, the character of the people involved in creating content for it changed, from a technical orientation to a marketing orientation. This became reflected in their choice of site addresses. Since the general public had come to expect Web addresses to be short names "framed" by www. and .com, the "marketing types" were determined to give them that, whether or not it was appropriate in a given case.
The technical types actually managed to hold a great degree of power for several years into the marketing incursion, since things like registering a domain name and creating a Web site were somewhat arcane and required technical skill -- thus, the marketing people had to come to technical people for assistance in getting these things accomplished, and the "techies" could sometimes talk them into doing something different from the www.SomeSillyName.com they originally wanted. I was in that position myself as I worked for an ISP in the days when the Internet was beginning to explode into the popular consciousness, and managed to convince several clients to use .org instead of .com for nonprofit organization sites, and to use subdomains for branches, chapters, and subsidiaries instead of registering separate domains. At the time, a non-technical person entering the Internet still tended to think of him/herself as a "stranger in a strange land," and look for guidance from those who knew the language and culture better. Thus, through about 1996, domain registrations were still largely being done along the correct structural lines (most campaign sites for the 1996 elections were at .org addresses, the proper ending for noncommercial organizations, while by 2000 most were improperly using .com).
But this changed as the less-technical people poured into the Internet in great waves, overwhelming everything that came before. Eventually, corporate and marketing types felt like they were the ones in charge, and were no longer just visitors in the realm of the "tech-types" -- their own ignorance of technical details was no longer something that needed to be remedied by consulting with those who knew better, but rather it was something that ought to be imposed on everybody else. The domain registration process became automated via Web form and took less technical skill to complete. Technical people could be bypassed altogether, and were sometimes taken out of the loop completely in the process of deciding what domain names a company or organization would register. Even the technical professionals of the Internet became less knowledgeable, as the huge demand for Internet-related professions caused a large number of people less intelligent, knowledgeable, or well-educated about the structure of the Internet to begin offering their services as professional developers.
Speculation and Cybersquatting
As early as 1994, there were some people anticipating the big demand for domain names for marketing purposes. A journalist managed to register mcdonalds.com to demonstrate how somebody could "cybersquat" on a corporate trademark if the company doesn't grab the name first. MTV veejay Adam Curry failed to find any corporate interest in getting that music channel onto the Internet, so he registered mtv.com himself and set up an "unofficial" MTV site there -- then he quit MTV, putting him in conflict with his former bosses who suddenly realized they wanted to have that address for themselves instead of letting him run off with it. Eventually, a big rush occurred to register domain names that might become valuable. Up until September, 1995, registration was free, but Network Solutions, the then monopoly registrar, convinced the U.S. government (for whom they'd been operating domain registration for an annual fixed payment) that they should start charging for each registration now that this had become a commercial matter. That it now cost money to register a domain didn't stop the speculators; in fact, speculative registrations actually increased, as perhaps some people were showing some restraint in registering names while they were free, and hence a "public resource", but now that they cost money, they could be thought of as private property to be hoarded unlimitedly.
Also, now that Network Solutions made money on each domain that was registered, their incentives shifted. Before this, they encouraged registrants to understand the naming system and register names that made sense within it. They sometimes rejected registrations if they were in the wrong top level domain for the type of organization, or if the same organization tried to register multiple domains that would be better served by the use of subdomains. But now, with more money to be made by unbridled registrations, they stopped trying to control it. .com was the first to go completely uncontrolled, with no attempt to verify that the applicants were actually commercial entities. They tried, for a while, to keep .net and .org under a degree of control, but this fell away after a while, leaving only .edu, .gov, .mil, and .int with restrictions that were actually enforced.
For a few years, the Network Solutions site at least still had text that explained the distinction between the different TLDs and encouraged registration in the appropriate one, but ultimately the "marketing types" took over completely there, and the site was redesigned to remove all such references (except buried deeply in FAQ or glossary pages), replacing them with marketing blurbs urging registrants to grab all three of the .com, .org, and .net domains to "protect" their brand names. Why not... this tripled the registration fees Network Solutions received!
Thus, over the next few years, there was a big rush to register every domain anybody could think of. People got domains to speculate on their eventual value (a handful of high-profile, high-price sales like of business.com fueled such speculation, though the vast majority of names never managed to sell for more than their registration cost), to "cybersquat" on trademarked names in the hope of getting extortion money out of the trademark owner, or to "typosquat" on names similar to those of existing sites in the hope of drawing accidental traffic. Corporations started registering every variation and permutation of their product names to get them before the cybersquatters did. Advertising agencies thought it was "neat" to have a different marketing-slogan-gimmick domain for every campaign that they could flash on the TV screen during their commercials, getting the world to beat a path to "WhizBangSuperSpringSale.com". Enough companies even changed their corporate names to something ending in .com so that "dot-coms" became an established part of the language to describe Internet-oriented companies, at first while hyping them, then later, after the big market bust of 2000, to deride the empty hype of the earlier period.
Law and Policy
The rise of "cybersquatting" brought on lawsuits from trademark owners. Soon lawyers and courts were being dragged into this issue, and Network Solutions jumped in with a dispute policy that many claimed was unfair against non-trademark-owners. Anybody who was just trying to use domain names as a technical addressing system was now lost in all of this noise and confusion.
The big companies argued in court that somebody else with a domain name that was similar to their trademark was confusing the public into thinking their site was affiliated with the trademark owner. While there was much truth to this, to a large extent the companies brought the situation on themselves by their own use of endless marketing variations of their name for the domain names of their sites. If they had used domain names rigorously in their original structure, using subdomains of their main corporate domain, then the public might have learned to understand the concept that, for instance, only domains that were within citibank.com belonged to Citibank. If a con artist were to try to use a site at CitibankAccounts.com to dupe consumers into providing personal info to a site not really associated with that bank, he wouldn't have nearly as much success in a world where Citibank didn't really use a whole heap of variant names itself, so it's plausible that they might be using that one too. Thus, corporations that are now complaining about the need to police their name and lots of variations of it in every top level domain, and who are opposing the introduction of new TLDs because of the increased trouble and expense this causes, are really the ones at fault.
But, since the corporations were the ones with the big bucks, domain name policy has been increasingly controlled on their terms, so that the various domain dispute policies have tended to be biased in the direction of trademark owners, and the introduction of new TLDs was delayed and wound up having extensive "sunrise" provisions giving a trademark owner much greater rights in cyberspace than in the real world.
I Think ICANN!
There was much dissatisfaction over the monopoly profits enjoyed by Network Solutions in domain registration. This ultimately led the U.S. government, which still had formal control over Internet structures due to it originating as a government research project, to create a new organization, ICANN (The Internet Corporation for Assigned Names and Numbers), to oversee the domain name system and related issues.
This has had mixed results. It did succeed in at least partially ending the monopoly power of Network Solutions. The functions of "registry" (the keepers of the central domain database) and "registrar" (those who deal directly with the people, companies, and organizations who register domains, provide the user interface to register and modify those domains, maintain the "WHOIS" data giving more details on the registrants, and take payments for registrations and renewals) were separated. Network Solutions (or, actually, its then parent company Verisign) retained the registry, the behind-the-scenes outfit that controlled .com, .org, .net, and .edu (but .edu was spun off later to a non-profit consortium, and .org was reassigned later; .net could eventually follow... or not) and gets a few dollars in wholesale registration fee for each domain registered. This is much less than the money it made per domain when it was the monopoly registrar, but it's still a lot of money with the millions of domains registered. Meanwhile, the registrars are now in a highly competitive market, with Network Solutions (now no longer owned by Verisign, which kept the registry but spun off the registrar) still one of them, but many others offer lower prices and better service. Domain registrations for .com and .net still ultimately go into the registry owned by Verisign, no matter who you register them with.
ICANN also imposed a uniform dispute resolution policy, which I discuss in detail elsewhere.
Another change promoted by ICANN was to add additional top level domains. These were run by different registries, hence ending the monopoly of Verisign. (Actually, there have always been country code domains, run by different registries, but they have been less commonly used especially in the United States, where the .us domain was organized in an unwieldy manner -- it finally started to allow direct second-level registrations in 2002.) This new TLD process was very slow to get going, due to the great amount of political squabbling, but it has finally happened, with seven new domains active... though none of them have really caught on with the public yet. Supposedly, a few more are on their way soon.
ICANN has come under a lot of criticism, some of it warranted, but some of it appearing to be the sour grapes of people whose own schemes to make money off domain names have been thwarted by ICANN policy. One source of contention is the alternate roots, which some promote as a way to break the ICANN monopoly, while others criticize as bringers of chaos to the naming system, where show.biz can end up going to different Web sites depending on which ISP you're using -- URLs would no longer be uniform or universal.
As time goes by, opposition to ICANN continues to increase, and it is coming under scrutiny by many people and institutions, including the U.S. Congress, which is questioning whether the agreement between ICANN and the Commerce Department, which gave ICANN its power over Internet namespaces, should continue. It hasn't helped things that the original agreement called for half of ICANN's board to be elected by popular vote of Internet users, but in fact only a smaller number of board members were ever elected, and there are no plans by ICANN's entrenched insiders to allow any further elections. The ICANN bylaws are silent on what to do about the "At Large" board members after their initial term, because the concept of how to organize an international election of Internet users was too unsettled at the time to be enshrined permanently there; instead, ICANN was expected to come up with a workable method for future elections by the time they were needed. But they haven't; instead they have adopted a new structure which does away with At Large board members altogether and just lets the cabal of "stakeholders" run the whole show. ICANN management responds in a nasty, sneering manner whenever anybody points out their suppression of democracy, calling objectors "radical critics" and claiming that Internet elections are unworkable. The biggest critic is Karl Auerbach, who was elected as the North American representative on the ICANN board, and is best known for suing ICANN (successfully) for the right to examine their financial records without signing a restrictive agreement. The U.S. government might eventually pull the plug on ICANN; congresspeople have expressed some skepticism in hearings. The future is still very uncertain at this point.
In all the debates about domain names, many of which I cover more extensively in the other pages of my site, lots of inappropriate metaphors get bandied about. Messages sprout up often in various forums that say "Dot Com is King!" (So what is "Dot Org", queen, prince, or court jester?) Others say that .com is the winner, .org and .net are distantly in the money as "place" and "show", and all other TLDs, old and new, are also-rans. A recent message even said "Rock on, Dot-Com!", as if this TLD were a rock group, in addition to being a monarch and a racehorse.
Those messages were from proponents of the use of .com as if it were the only TLD in a flat namespace. Opponents often buy into the same metaphors themselves, when they promote some alternative TLD, like a country code being "repurposed" as a generic TLD (.tv, .md, .cc), a new TLD being added by ICANN (.info, .biz), or an alternative root system's TLDs (new.net) as a "dark horse" they expect to make a big burst of speed in the future and unseat .com as rightful ruler of the namespace (to make an awkward shift from horseracing to royalty in my metaphor use).
All of this is misguided. If one thinks of the domain name system in the manner in which it was intended, as a logical namespace for the naming of entities on the Internet, then it makes no sense to regard one TLD as "ruling" over the others, or of beating them in a race, or of "rocking on" for its fans and groupies. They're simply different parts of a naming system, intended for different purposes, just as in the telephone numbering system area code 212 is for phone numbers in New York City while area code 213 is for Los Angeles. Do you speak of one area code, or zip code, as "ruling" over the others?
In what way has .com "beaten" .org, or .edu, or .museum? By having more Web sites and email addresses? There are more commercial businesses than universities or museums, certainly, but does this mean that the more numerous entities "rule"? This isn't a situation where majority-rules applies. If you're looking for higher education, you'll go to a university to find it, not a gas station, even though the latter exist in greater numbers. If you're looking for a university Web site, you'll look in .edu, not .gov or .com. The different TLDs are different, not "better" or "worse" than one another.
Since I mention area and zip codes, I'll point out that those addressing systems have in fact sometimes been contorted and abused by the "marketing types" in order to contrive a postal address or phone number that suits their desired image. Businesses will sometimes get the street in front of them renamed, and get exceptions to the house numbering system, to get the address they want, like "One Disk Drive" for a computer company or "711 Winners Circle" for a casino. There are varying degrees of prestige associated with particular streets, cities, and even zip codes and area codes, which will sometimes be taken into account when people and businesses decide where to locate. Phone numbers will sometimes be contorted to spell things for a business campaign, and there have even been disputes involving toll free numbers that are remarkably similar to domain "cybersquatting" cases, where companies are accused of taking away one another's business by getting an intentionally similar phone number. Some companies have felt compelled to get the same number in each of the several toll-free area codes, 800, 888, and 877, much like registering the same domain in .com, .org, and .net, defeating the purpose of the creation of these additional area codes to increase the number of available toll-free numbers for the many companies that want them.
However, the difference in such cases is that they are acknowledged by practically everybody to be a minor sideshow of the addressing systems as a whole. Nobody claims with a straight face that the system of assigning telephone numbers or the system of assigning street addresses exists exclusively, or primarily, for the purpose of marketing, branding, and trademark protection. Rather, those addressing systems exist primarily to provide logically structured identifiers for all the telephones and postal delivery locations in the world. That they're occasionally used or abused for marketing purposes is incidental, and it doesn't cause the entire system to be restructured in a way that's less logical but better suited to the marketing people or the corporate lawyers. So why is the domain name system treated otherwise?
Standards, Policies & Historical Documents
This page was first created 11 Aug 2001, and was last modified 10 May 2015.