DNS articles Brad Templeton Home Brad Ideas (My Blog) ClariNet
|
|
Goals for a domain name system
In order to figure out how best to design or fix a domain name system, you need to first figure out what it's for. And that means enumerating not just what you think a domain is for, but what other people want, for in the end the ultimate goal is to give the users what they want. Some of the goals can be reconciled, some can't. Some lead to good results and some are more selfish and lead to bad ones. To do this we first should look at how people -- both owners and users -- make use of domains today, and what they want from them, and how badly they want it. I've identified several goals, and ordered them only roughly. These goals are mostly human goals, because naming systems are really for use by humans, who say them, hear them, read them and type them. The goal of getting levels of indirection for computers is also present but much easier to solve and thus at a lesser level. Indeed the technical goals are really only there to serve human goals. Summary goalI have found that the following problem statement sums up a remarkable number of the goals into one issue. "I want to be able to meet a person, and tell her my E-mail address, and have it be easy to remember and type accurately so that when she enters it, she easily and without complication sends mail to me, and never to anybody else, for as long as I want to keep the address. And she wants the same thing." ReliabilityThe most common thing done with a domain is looking it up. Often this is automatic, and users are not even aware of it. It happens many billions of times a day. People want name look ups to be fast and reliable. ReadabilityNext most frequently, people read domain names, and get some sort of information from them. You read a domain when you read the "From" header of an E-mail you get, or when you move your mouse over a link in a web page, or you see a domain (in many forms) in an ordinary document, both on and off the web. People desire their domains to mean something, and people who see a domain and use it get expectations about what the domain might point to from how they read it. ExpressionA sub-goal of readability has led people to express themselves through domain names. They pick domain names that say something, sometimes even controversial things, as a message to those who read them. The above goals reflect the vast bulk of uses people put names to, but there are others. MemorabilityPeople want a name that others can remember, when it has to be given in non-written form. People want to be able to tell somebody an E-mail address and expect the other to remember it. They want to see a URL on a billboard or in a TV ad and remember it when they get to their computer. This is a step up from phone numbers, most of which are not memorable. TypabilityAssociated with memorability is the ability to type in a domain and get it right. This applies to any domains you hear in person or on the radio. When you hear a domain you would like to be able to type it in and have it point to the right place. At this point, little effort has been put into this goal in any system. The typability goal might suggest that two homonyms could be considered the same (such as "phone" and "fone") and that similar symbols like "0" and "O" should be considered the same. (Upper and lower case letters are already considered the same in the existing DNS.) PermanenceWhen people get a domain, they would like it to last forever, or at least until they say otherwise. Of course, all dispute resolution systems which can take a domain away are at odds with this goal. UniquenessPeople want a domain to point precisely at one thing. If I give you my business card with an E-mail address on it, You and I both want that address to mail to me, and me alone. You might consider this a subset of reliability. Cheap and for everybodyPeople would like the domain name system to be cost-effective, with minimal cost to both the holders and users of domains, financially and technically. In particular, people want the system to scale to a huge size, to create a means to provide pointers to not just big companies and institutions but all the people in the world. Flexibilty and InnovationWhile often not a goal people will overtly cite, all systems benefit by allowing freedom of individual choice so that they can foster experimentation and innovation in the system. Goals and customer demands vary and change with time. It must be possible to innovate and try new systems. As few rules as possible should be cast in stone, other than those than protect fundamental rights. FairnessMost people want the domain system to be "fair" in one fashion or another, which broadly means not untowardly benefitting one party at the expense of others. Some feel that first-come-first-served is a fair allocation scheme in a namespace, while there are other arguments that suggest that it may be possible at some levels to serve everybody, and not need to give preference to the first (or anybody else.) Another "fair" system is to put names to auction, to the highest bidder. Another could be a lottery. There are others that might exist, for the various definitions of fairness that people have. Low-hasslePeople want a system free of disputes, or in which disputes can be handled easily. Of course this is hard to reconcile with some views of fairness. Those who think a domain should "fairly" belong to them want tools to make that happen. Those who are the targets of such action want the process to be more friendly. Of course, the real world's dispute system (the courts) still usually trumps whatever system domain names might have. FindabilityPeople want to be able to find the right domain for an entity when they don't know the domain, or have heard it but forgotten it. Some wish this goal to be solved in the name system, others feel it is a goal best served by tools like search engines. GuessabilityClosely associated with findability is the desire of some to be able to simply guess a domain name from other information about a person or company, without the use of tools like a search engine. Many people, who have existing well known names in the world outside the net, would like users to be able to guess their domain name. For example, Ford Motor Cars is keen on having the domain "ford.com" because many would guess that this would be their domain. This goal of guessability is one of the greatest sources of conflict. No CybersquattingWhile this goal derives from several of the other goals, it is worth a mention because it is one of the few goals some users have actually gotten incoporated into law, though with much debate about just what cybersquatting is. At the core of this goal is a desire that others can't block you from getting a name you feel is the right one for you. However, the term tends to refer to people registering a domain deliberately to block you from getting a domain they predicted you would want, ideally so they can get money to hand it over. Because of the debate over the definitions, anti-cybersquatting roles tend to interfere with other goals like permanence, and low hassle. Prestige & ValueMany believe certain domain names contain inherent prestige and value. For example, the ordinary English language words and phrases, appended with ".com" attained great value. Some of them, like "drugstore.com" and "business.com" sold for millions of dollars. While trademarks have value built by their owner rather than inherent value, the current DNS allows ownership of ordinary words and phrases (which trademark law does not.) These ordinary words have an intrinsic value and prestige that says, "This company is important." Drugstore.com, one realizes, must be an important pharmacy site, or at least one that got there early. These names also tend to be memorable and guessable, to boot. Some domains, like trademarks, have also acquired value beyond their intrinsic one. People have made considerable investments in them which they wish to keep. ControlOne of the "darker" goals is the desire not simply to own a given domain name for its values, but to stop others from owning related domain names. Some of this is the correct application of trademark law, but many parties have the desire for much broader control. For example, Ford Motor Company, which I listed above, is not only keen on owning ford.com, but took action to take away "fordsucks.com", which was not really a domain Ford wanted for itself, nor a trademark violation. It also took action to take away "jaguarcenter.com" (Ford now makes Jaguar cars) even though it is a site about the endangered cat. Of course, trademark issues play a major role in both the desire for control, and the mechanisms of control. The trademark system allowed parties to gain ownership in certain geographical reigns of various symbols, phrases and words in the context of industries in which they sold goods and services. Many have wished to transfer those rights into the domain name system. No ControlA counter to that goal is that people don't want outside forces running the system against their wishes. They don't want any one powerful party (that isn't them) to have too much say or control, with the ability to take away their name unfairly or to enforce rules they don't desire. In particular, people don't want their use of the system to be subject to the laws of a country other than theirs, and they don't want one country's laws to dominate. In particular, since the USA created the system, people outside the USA are concerned over U.S. dominance of the system. Save the LegacyThe most frightening goal which most people would rather not exist is the desire to remain compatible with, and not destroy, the legacy systems. Some have large investments in the legacy and would suffer major trouble if it were simply replaced, even with something better on all other accounts. The biggest piece of the legacy is of course, ".com" |