On September 26, 1995 a group calling themselves People Eating Tasty Animals registered the domain name peta.org and started a web site (Web). The purpose of the site is as "a resource for those who enjoy eating meat, wearing fur and leather, hunting, and the fruits of scientific research (and more!)" (Doughney). While there is nothing wrong with People Eating Tasty Animals running a web site, the controversy arose when PETA (People for the Ethical Treatment of Animals) found out about the site at peta.org. After PETA complained to InterNIC, the body responsible for domain name allocations, the domain name peta.org was put on hold, where it will remain until there is a court ruling on who gets control of the peta.org domain (Agmon, "Someone Else's Trademark"). The issue is a complex one because it deals with not just trademark issues but also with First Amendment freedom of speech: the right of People Eating Tasty Animals to parody PETA using the peta.org domain. However, since People Eating Tasty Animals cannot afford to take PETA to court and PETA does not seem inclined to take the issue to court as long as the domain name remains on hold, this controversy will never get resolved under the current system.

On May 18, 1994 Roadrunner Computer Systems, an Internet service provider in New Mexico, registered the domain name roadrunner.com and started using it to promote its' business (Web). The problems of Roadrunner Computer Systems started in December 1995 when Warner Brothers, Inc. complained to Network Solutions, Inc. (NSI) about the use of roadrunner as a domain name since roadrunner is a Warner Brothers trademark. Because the current policy of NSI, the company operating domain name registration under the auspices of InterNIC, takes only trademark status into consideration when controversies arise, it informed Roadrunner Computer Systems that the name was going to be placed on hold (Domain). Roadrunner Computer Systems has since obtained a "roadrunner" trademark from Tunisia but it was not accepted by NSI because it was not submitted within the first 30 days and the legal battle is now being fought between Roadrunner Computer Systems and NSI (Roadrunner). Away from the Internet, it is possible for multiple trademarks on the same word to exist if they are not competing. For example, both Domino's Pizza and Domino Sugar can have trademarks on domino but there is still only one domino.com domain name (Agmon, "The Relationship ..."). Under the current NSI policy, any trademark holder could conceivably take a domain name away from someone who does not have a trademark, even if the current owner has a more legitimate right to that domain name (Domain).

From these two examples and the large number of other disputes that have arisen, it is clear that the current domain naming system (DNS) does not deal effectively with trademark issues (Agmon, "Domain Name Disputes"). The current policy leads to either court battles or arbitrary decisions that do not take the specifics of the dispute into consideration. Another problem is that NSI currently has a monopoly in the major top level domains, or TLDs, including the .com TLD. A competitive system among multiple registrars would force the registrars to be more responsive to the needs of the domain name owners. The predominance of the .com TLD also creates potential problems. As of the end of March 1997, over 88% of the domains registered in the major TLDs were in the .com TLD (Landers). More TLDs need to be created to alleviate this overcrowding in the .com TLD and keep the databases small enough to allow fast access. Besides these problems with the current system, there are some other reasons for making changes. In March 1998, the agreement between the National Science Foundation (NSF) and NSI will expire, and the NSF has announced that it will not renew the agreement (Bordogna). The agreement is the basis for NSI's current monopoly on the major TLDs and when it is gone, the justification for their monopoly will be too. Also, registration of domain names is a lucrative business. It currently costs $100US to register a domain name for two years and Network Solutions has accepted over one million registrations (Landers). Increased competition would drive prices down and provide the end users with the most cost-effective options.

In order to make an informed decision about the future of the DNS, we must understand the reasons behind the DNS as well as the state of the current system. Once that is understood, we have to look at the proposals for the future of the DNS. While many ideas are more extreme than the current system, the major proposals by AlterNIC, NSI, and the Internet Ad Hoc Committee (IAHC) all deserve careful consideration. The adoption of the IAHC solution now seems probable but there is still strong opposition so other proposals should not be abandoned. Both the proposal by AlterNIC and the proposal by NSI have aspects which should be incorporated into the final solution based on the plan set out by the IAHC.

The rationale behind the DNS is fairly simple. On the Internet, computers are uniquely identified by a 32-bit number separated by periods called an IP, or Internet Protocol address. IPs are analogous to a person's house being uniquely identified by their postal address (Murphy). For example, my computer is located at 171.64.177.78 and Stanford University's web server is located at 36.190.0.136. While these numbers are easy for a computer to remember, they are extremely difficult for people to memorize. The DNS was created to provide a hierarchical and recognizable method for identifying places on the Internet (Murphy). To show the contrast, it is much easier to remember rover.stanford.edu for my computer or www.stanford.edu for the university's web server.

The DNS is based on a few basic principles. Every domain name is unique, hierarchical, distributed, and interdependent (Murphy). First, each full domain name corresponds to a unique IP address. For example, www.stanford.edu is the only domain name that matches to 36.190.0.136. An extra advantage of using the DNS as an abstract level of unique locators above the IP addresses is that the hardware underlying a given domain can be easily changed with only the entries in the primary and secondary name servers needing to be updated instead of every single reference to the location.

Second, all domain names are hierarchical (Murphy). Segments of the domain name are separated by periods. The highest level segments are on the right and they get more specific as you go left. On the far right is the top level domain or TLD. Currently the major top level domains are .com for commercial entities, .org for other organizations, .net for Internet related sites, .gov for government agencies, .edu for educational institutions, and country codes. The country codes are under the control of the respective governments so the focus of the controversy surrounds control of the generic TLDs. Below the TLD comes the second level domain. The second level domain or SLD is what can be registered by an individual or a corporation. Examples of SLDs are cnn.com and mcdonalds.com for CNN and McDonalds respectively. Everything before the SLD are subdomains (Murphy). These subdomains can be whatever the SLD name owner wants. The most common subdomain is www for the primary web server. In the above examples, www.cnn.com and www.mcdonalds.com are the primary web servers of CNN and McDonalds.

Finally, domain name information is distributed and interdependent (Murphy). Each domain name server is responsible for a small subset of the total domain name information. Having a centralized database with domain name information would present large scale logistical problems. Keeping the IP address of every domain name up to date would be extremely difficult and the rapid growth of the Internet would cause the database to become too unwieldy to be feasible. With distributed name servers, updating information only needs to be done on the primary and secondary name servers and exponential growth can be dealt with by increasing the number of name servers. Due to this distributed model, the DNS is completely interdependent (Murphy). Since each name server only maintains an extremely small portion of the total domain name information, the domain name servers rely on the information stored on other domain name servers so that anyone connected to the Internet can access a resource on any part of the Internet.

With the domain name information spread among numerous domain name servers, a computer must have a way to find the correct domain name server. To do this, the computer first checks if the local domain name server has the information.

If your local domain name server does not have the desired information, it checks a root server to find out where information for the TLD is stored.

Your local name server then queries the TLD server to find the local server for the SLD.

That server will give your computer the IP address of the desired location.

Your computer can then request the information from the computer which stores it.

By breaking the domain name query into these multiple parts, it is possible to handle exponential growth by just increasing the number of servers at each level. This distribution also allows the time it takes to process a query to remain almost constant no matter how fast the number of domain names increases. By being hierarchical, unique, distributed and interdependent, the DNS is capable of handling the rapid expansion of the Internet.

The current system of domain name allocation is administered by InterNIC, a collaborative project between the National Science Foundation and NSI. Under their agreement, NSI is responsible for domain name registration ("About the InterNIC"). NSI currently has a monopoly on registering the five major TLDs, .com, .org, .net, .edu, and .gov, and under their system, registration is on a first-come first-serve basis. This first-come first-serve model has led to some problems. In one example, the Princeton Review, whose biggest competitor is Kaplan Education Center, registered kaplan.com (Agmon, "Domain Name Grabbing"). As noted above, most controversies arise over trademark issues and the current policy of Network Solutions is to put disputed names on hold until the courts make a legal ruling on the case.

Given the problems in the current system managed by NSI, it is not surprising that alternative proposals exist. However, some of these alternative proposals are even more extreme than the current system. One example is a randomly generated system of TLDs proposed by engineer Paul Vixie in 1995. This proposal would create multiple registries, each with its own randomly generated TLD using base-36 numbers(1). Either the number could be a random 4 digit number such as S6EG or a letter, digit, digit, digit order, for example A123. In either case, any combination that represented a word in any language would be omitted (INTA). In this scenario, no TLD is preferable over any other, but one could no longer remember web sites such as www.sportsillustrated.com or www.cnn.com and access would have to be through some other medium such as a business card, Internet search-engines, or a directory service. Directory services to locate web pages already exist, Yahoo being the primary example, so while there is very little to criticize in Vixie's proposal, there are too many commercial and advertising interests invested in the current system of recognizable domain names for this proposal to have any chance of success.

A proposal that combines the random number assignments of Vixie's plan with the ability to register a vanity domain name has been introduced by David Maher and Robert Frank. Under their plan, normal new assignments of TLDs and second level domains or SLDs would be randomly assigned a string of letters and numbers. However, there would be an option available to trademark holders to register a vanity domain name for an extra fee, similar to license plates (INTA). For example, I might register daryl.com as a vanity domain name instead of getting a randomly assigned domain.

For these vanity names, a special search would be conducted to ensure that there are no conflicts with existing vanity domain names and a special procedure would be followed if a conflict did arise (INTA). This proposal is not popular both because people dislike the idea of randomly assigned domain names because they are difficult to remember and because the special procedure to follow in the case of a vanity name conflict is not defined.

Another proposal is to group trademarked domains by their International Trademark Classification System number. The ITCS number is a two digit number used to identify the class of the trademark (i.e., group 25 for clothing, group 36 for financial, or some other group) (INTA). This plan would allow for distinction between some of the trademarked groups; however, it does not take geographical location into consideration so an Acme Clothes Co. from France and an Acme Clothes Co. from Germany could both be entitled to the same domain name under this plan. While this is not a problem for their regional trademarks, the domain name is international so applying the same classification to both regional and international trademarks would be difficult. With such a large area for disputes, this plan would create more problems than it would fix.

One solution to the problem of geographic overlapping is to incorporate geographical identifiers as part of the domain names. For example, instead of microsoft.com, we would instead have microsoft.redmond.wa.us for Microsoft whose headquarters is in Redmond, Washington. One large concern of any non-local business is that this system would tie them down to a single physical location (INTA) which would be very disagreeable to corporations such as McDonalds wanting to reach potential customers all over the world, not just near its headquarters. Another problem is that it would be necessary to go all the way down to the level of street address to entirely avoid duplication in this system (INTA). While quite feasible and possibly a useful solution, many corporations and individuals dislike this proposal, perhaps because it seems to remove the international, abstract feel of the Internet. Another problem is that national governments have jurisdiction over domain names under their country code so the Internet would be subject to censorship and other political issues.

While many of these proposals never got past the initial stages, one alternative proposal by AlterNIC has already been implemented. AlterNIC provides an alternative to InterNIC's current system by maintaining its' own set of root servers. As shown in the diagrams above, root servers maintain a database of the locations of the TLD servers. By running its' own set of root servers, AlterNIC has the ability to add new TLDs and currently allows anyone to register a new TLDs or just a SLD under one of the new TLDs in a manner similar to NSI's registration of SLDs. While this proposal is strong in the technical, economic, and social aspects, the legal aspects create some problems.

By going ahead with the implementation of their proposal before trying to create a consensus, AlterNIC bypassed most worries about the technical feasibility of their proposal. In the process of demonstrating that their proposal could work, however, AlterNIC also showed the inherent problems in having multiple solutions and the need to create a consensus. Currently, only a small fraction of Internet users can view sites in the new top level domains because most domain name servers use InterNIC's root servers which do not list the TLDs registered with AlterNIC. The only sites able to find sites in the new TLDs are those where the domain name server references one of AlterNIC's new root servers instead of a standard InterNIC root server (AlterNIC.NET, "How To..."). Until a broad consensus is reached, most domain name server operators will not switch to AlterNIC's root servers simply because they have not been tested for as long or under as much usage as InterNIC's and are therefore not considered as reliable.

Economically, AlterNIC's plan is more attractive for end users than the current NSI policy. Instead of charging $100 for the first two years and $50 every year after (InterNIC), AlterNIC only charges a $50 one-time setup fee and $24 annually to register a SLD(2) (AlterNIC.NET, "AlterNIC DNS Fees"). While NSI seems to have more services to offer, most users only need the domain name registration and it would be move cost-effective for consumers if the registration service did not have to subsidize auxiliary and usually unnecessary services.

The AlterNIC initiative seems to fit very well with the open, democratic system of cooperation that allowed the early Internet to flourish. Unlike the current system where NSI has a monopoly on registration, AlterNIC even allows other companies to serve as registrars for other TLDs. While open cooperative systems were common back when the Internet was being developed as a research and educational tool, commercial interests now play a large part in the development of the Internet so the viability of such an open standard is questionable considering the stability requirements of the business users.

The largest problems with AlterNICs implementation are legal and dispute related concerns. As long as it is only accessible to a small portion of the Internet, large companies who want to reach as many customers as possible will not use the AlterNIC for domain name registration and therefore there have not been many disputes. However, if it was accepted as a mainstream approach to domain naming, many more disputes would arise and AlterNIC's complete policy on dispute resolution states: "any disputes between parties over the rights to use a particular name are to be settled between the contending parties using normal legal methods" (AlterNIC.NET, "AlterNIC Domain..."). One major problem is that a lawsuit or any kind of court proceeding is extremely expensive so small companies and individuals will be at the mercy of large corporations. Another concern is whether or not the courts are well enough informed about the Internet to deal with disputes effectively and fairly. Besides these problems, the policy of sending all disputes to the courts would mean the courts would develop the dispute resolution precedents and guidelines based on individual cases instead of developing dispute resolution guidelines through consensus building of the Internet community and approval by the accepted guiding organizations of the Internet. Leaving all guidelines to the courts would be a major departure from the current Internet practice and would itself represent a precedent that could lead to governmental regulation of the Internet; a fundamental change in the way the Internet is controlled. With these major problems, it is clear that AlterNIC's proposal does not effectively address dispute resolution or other legal issues.

As the organization that controls most of the current DNS and therefore has the most experience running and organizing a DNS, NSI brings a voice of experience to the debate over the future of the DNS. From its' experience, NSI has developed five guidelines it believes are critical for the future of the DNS. First, regulation must be very limited. The DNS should use free market competition to promote the highest quality of service and the most cost-effective methods of registering domain names. Second, TLDs should not be considered a "public trust". There are enough SLDs in each TLD so that every person on earth (approximately 4 billion) could have 25 million distinct domain names. However, it is most likely that TLDs will be restricted to three to five characters long for simplicity and ease to remember. While this limits the number of TLDs by a great deal, there are still enough so that they aren't in any danger of being used up. Third, there should be no limit on the number of TLD registrars allowed. Market forces, not arbitrary limits, should determine how many registrars are needed. Fourth, the administration of the root servers and IP addresses are critical and should be controlled cooperatively since they are fairly limited. Fifth and finally, the administration of the DNS and IP addresses should be given a legal basis instead of the current system where the right to control these functions is based only on historical precedent (Network Solutions). Based on these principles, NSI has come up with a proposal for the future of the DNS.

NSI's vision for the DNS has four basic parts. First, to provide a legal authority for the administration of IP addresses and the DNS, NSI wants a two-year transition period where the administration is transferred from the volunteer IANA (Internet Assigned Numbers Authority) to a special US government committee in order to institutionalize the process and allow an international authority to be created. After the two-year transition period, the newly created international authority would take over control of the functions currently controlled by the IANA. Since no international authority currently exists to take control of the system, one needs to be created and the transition period would allow a stable shift of control. NSI defends using the US government as an authority during the transition period based on its historical funding and development of the Internet (Network Solutions).

The second part of the proposal involves the selection of an independent contractor to manage the root servers and TLD databases. This contractor would be chosen through a competitive application process, presumably by the IANA or the transitional authority. Although not stated in the proposal, I assume that this contractor would be under the direct oversight of the controlling international authority to ensure a sound and stable system. Another duty of this contractor would be to set minimum financial and technical standards for parties interested in becoming TLD registrars (Network Solutions).

The core of NSI's proposal is in the third section. This would allow an unlimited number of TLD registrars provided they met the minimum standards as set by the contractor running the root servers. Each registrar would have exclusive control over up to three TLDs, allocated on a first-come first serve basis. The registrars would be responsible for maintaining the locating databases for their TLD as well as paying fees to support the contractor running the root servers. This system leaves all dispute resolution procedures up to the individual registrar as well as allowing registrars to establish name recognition for their TLDs and would create the competitive environment sought in the NSI guidelines (Network Solutions). This part of NSI's proposal clearly favors NSI. Besides being the main established registrar, NSI currently controls .com, .net, and .org, the three most popular TLDs. NSI would certainly try to maintain control of these TLDs under the new system and would therefore have a huge head start over other registrars. The final aspect of NSI's proposal calls for the creation of three more regional IP Registries to expand the current system of IP allocation worldwide (Network Solutions). This part of the proposal is already being implemented and so its' inclusion is merely a show of support for that policy.

On technical grounds, NSI's proposal is very sound. By allowing exclusive control of the TLDs, compatibility issues between databases maintained by different registrars will not be a problem. Since the technical aspects of this proposal just involve the current system expanded to allow multiple registrars and more TLDs, the technical feasibility of this proposal has already been proven.

NSI's proposal takes a free market economic approach. While this seems to provide the highest quality of service in the long run, there will be many instances of registrars that are unable to stay competitive or for other reasons cannot provide a stable system. This will lead to many disruptions for small portions of the Internet community as some registrars go out of business and take their domain name servers offline but it should not affect the Internet as a whole since the strongest registrars will be the ones providing the most stable service and attracting the most users.

The social consequences of this plan are difficult to foretell. By creating proprietary registrars, the system will no longer rest on the ideas of consensus and information sharing that seem to be the core values of the Internet. However, these registrars will all be dependent on compatibility with the root servers so there is very little potential for problems to arise between registrars that would affect large portions of the Internet community. Under the likely scenario that NSI maintained control of its three most popular TLDs, it would start with a base of over one million registrants, a huge advantage over other registrars who would be forced to build their systems from the ground up. An important aspect of changing the DNS is to remove the current monopolistic situation but it is unclear if this plan will give other registrars a fair chance to compete. In any case, with the NSF withdrawing funding and oversight from the DNS, private businesses will play a major part in financing and developing the updated DNS. From the proposals given, a combination of private business within a larger framework that ensures compatibility seems to be the best solution. The difficult question is how to separate the for-profit businesses from the non-profit controlling structure.

As with AlterNIC's proposal, NSI's does not deal with the difficult question of dispute resolution. Leaving it up to the individual registrars means there will be as many variations of procedures as there are registrars. While this method might eventually develop good solutions through trial and error, there will be many problems arising with the poorer dispute resolution policies and intellectual property protection will be very difficult since there will be different dispute resolution procedures in every applicable TLD. Also, a large number of problems have arisen with NSI's current dispute policy and it seems unlikely that it would change under this system so a large portion of the problems would remain. The lack of an overriding dispute resolution procedure to eliminate many of the problems and complexities of NSI's current system is a major flaw in the proposal.

The proposal that appears most likely to get implemented was created by the IAHC (Internet Ad Hoc Committee). The IAHC was created by the IANA to review and update the DNS. In their full proposal released on February 4, 1997 and in updates since that time, the IAHC is by far the most detailed proposal presented. With the support of the IANA and ISOC (the Internet Society), the main groups responsible for creating Internet standards, it is understandable that this plan is generating the most support. However, the powers of the IANA and ISOC are only derived from their implementation of a perceived consensus of the Internet community so responsiveness to public feedback is critical to the success of the IAHC plan.

The IAHC plan is based on a memorandum of understanding (MoU) signed by representatives of 57 organizations at a conference in Geneva (Wylie, "Net..."). The International Telecommunication Union (ITU) will act as a depository for the MoU (IAHC Proposal). With both MCI and UUnet, two of the largest providers of the Internet infrastructure, backing the MoU as well as the traditional authorities for creating Internet standards, the plan carries a lot of clout (Wylie, "Net..."). The advisory body for the new system will be a newly created Policy Oversight Committee (POC). The members of the POC will be chosen by designated Internet and international organizations. Under the IAHC proposal, seven new TLDs will be created:

While it is technically possible to have as many TLDs as SLDs, that number will remain limited under this proposal until a stable DNS system is in place that can handle a large increase in the number of TLDs. The new TLDs will be shared among the registrars with the existing TLDs added to the pool of shared TLDs when NSI's monopoly expires in March 1998. To oversee the registrars a coordinating group called the Council of Registrars (CORE) will be created. CORE will be a non-profit association in Switzerland, subject to Swiss law. Since the registrars will be competing with each other for business, a neutral third party is needed to actually maintain the database for each TLD and allow all registrars equal access. These neutral parties, or stewards, will be selected and overseen by the CORE. Both the CORE and the stewards will be funded by fees imposed on the registrars (IAHC Proposal). Initially the IAHC plan called for limiting the number of registrars to 28 spread worldwide to insure the system would remain manageable during the transition period. However, after the signing of the MoU, the IAHC was disbanded and replaced with an interim Policy Oversight Committee (iPOC) (IAHC Press Release, "Structure..."). This new iPOC responded to widespread public concern and decided to let anyone that meets minimum technical and financial qualifications become a registrar for the new TLDs (IAHC Press Release, "Lottery...").

The dispute resolution procedure proposed in the initial IAHC report outlined a unified system of dispute resolution procedures that has been clarified in a recent release by the iPOC. To simplify the procedure, a group of Administrative Domain Name Challenge Panels (ACPs) would be created to settle disputes over domain names. While having no legal power over any parties involved, the ACPs would have authority over the SLDs and would have to power to transfer rights to a SLD from one party to another and in exceptional cases of trademarks known world-wide, ban other parties from using a specific SLD in any TLD. Disputes between the parties could still be taken to court and any decisions reached by the courts would be followed by the registrars (iPOC). The ACPs, although they will probably need some adjustments once they are put into practice, would greatly improve the procedures for resolving domain name disputes. In most cases, they would provide a simplified manner to find a fair solution.

On technical grounds, the revised IAHC proposal is sound. While more difficult to implement with the shared databases, by using third-party stewards to actually control the databases, technical conflicts should not arise. Economics provide some of the sharpest criticism of the IAHC proposal. Although the proposal allows competitive registrars, it remains to be seen how effective the cost cutting will be since the actual costs of maintaining the database are incurred by a third party steward and shared between all the registrars. The steward, although overseen by the CORE, is not subject to the same competitive pressure to provide the most efficient service possible.

Socially, the IAHC's plan seems to continue the Internet's democratic traditions. However, serious concerns have been raised about the process the IAHC used to create their plan. Both the White House and the Association for Interactive Media (AIM) which has over 300 member companies have raised concerns that the plan does not represent large portions of the Internet community (Macavinta). Additionally, the speed with which the plan is being implemented raises concerns that the iPOC is not responding to the concerns of the numerous critics or trying to reach a consensus. Another specific concern is how the the Internet will be affected by having the governing CORE explicitly subject to Swiss law instead of under the control of an internationally based organization. Also, the US Government opposes the IAHC plan largely on concerns that placing the MoU under direct control of the ITU, a UN organization, will lead to intergovernmental control and possibly taxation of the Internet (Wylie, "U.S..."). One area where the iPOC has responded to the concerns of critics is in the method of dispute resolution. Although fairly experimental, the revised ACPs in the IAHC's plan provide the best method for dealing with domain name disputes. The ACPs do not remove the Internet from the power of the courts but provide a much simpler alternative to litigation to deal with legitimate concerns.

One of the fundamental principles of the Internet is to innovate, try new ideas, and eventually come to a consensus on a standard for a given technology so it can be spread across the Internet. With NSI under contract with the NSF and AlterNIC operating on its own, there has been significant innovation and effort put into coming up with solutions for the future of the DNS. With the support of the IANA and the ISOC, the IAHC's proposal will eventually become a standard. However, to be successful, it must follow the guidelines used by the IANA and the ISOC in their historical leadership of the Internet. To do this, the iPOC must respond to the major concerns of critics and develop a general consensus. While the level of consensus historically used may not be feasible with the variety of groups now represented on the Internet, the iPOC can still modify the proposal in order to come as close a possible to a consensus. First of all, the iPOC needs to slow down the timetable for implementing the new DNS so that major concerns with the proposal have a chance to be addressed. Secondly, useful portions of the proposals from both AlterNIC and NSI need to be incorporated into the IAHC's plan. From the example set by AlterNIC, we see that it is technically feasible to allow unlimited TLDs as well as SLDs and so the iPOC, the current successor to the IAHC, should make clear that the system will head in the direction of unlimited TLDs once the DNS is operating smoothly with a limited number of TLDs. From NSI's proposal we must consider the questions of economic efficiency. Shared databases managed in a non-competitive manner will definitely provide more stability for the end user that relies on the name servers to use the Internet. The tradeoff on this aspect of the proposal is between early stability and future cost-effectiveness because competitive proprietary databases would in the long run be more cost-effective based on simple free market principles. During this time of turmoil and change for the DNS, stability is more important than a minor cost difference because either new system will still be more effective than the current system. Also, once we have multiple registrars established in the field, there will be fewer problems with registrars going out of business during a switch to a free market, proprietary system. After the system with shared databases is running and stable, a transition to proprietary systems should be considered using the then existing oversight committee and current registrars to facilitate the change.

The way it stands now, the IAHC plan provides the best foundation for the future of the DNS. The worries of the US government that use of an international organization such as the ITU will be the basis for Internet taxation are probably unfounded. Not only would the US government be able to block such a move politically but users of the Internet would strongly resist any kind of direct tax. Fees to cover the cost of usage on the Internet were hard enough to implement and they were justified. While the plan for handling domain name disputes will need some adjustments, the revised proposal is by far the best out there and it should cut down on the number of legal problems. It's also possible that the ACPs will create a precedent that will eventually lead to a form of online jurisdiction, independent of any national courts. That would be the true evolution of the Internet as its own entity.

Please ask before using any part of this paper for any reason. Thanks!

Click here to return to the rest of my website.

Please send me comments, suggestions, ideas or even random thoughts! dpuryear@cs.stanford.edu.