Domain names with national characters

1. Consequences of national characters in domain names
   • Technical support in the domain name database
   • Other technical consequences
   • Legal consequences
   • Other consequences
2. Restrictions on characters under the .no domain
   • The official languages
   • Choice of characters
   • Comparison with relevant registers
3. Transition process
   • Equal treatment or preferential rights?
   • Choice of transitional schemes
4. The road ahead
5. References

The domain name system (DNS) was originally created to handle a very limited number of characters: the letters a-z, the figures 0-9 and hyphens. The background for this was partly that they were the only characters which could be found onall existing computers, and partly because they provided an unambiguous way to decide how names should be written. For example, there is no difference between upper and lower case. Software which handles domain names (e-mail programs, browsers, name servers, etc.) takes this into account, so that other characters in domain names are not accepted.

As the Internet has taken on a more prominent role in society, domain names have become more closely linked to both names and brands. At the same time, use of the Internet has extended from a primarily English-speaking environment, with no need for letters other than a-z, to communities with their own languages and writing. This has created a problem where a large number of names and concepts cannot be used as domain names in the way that they are written. In countries with written languages based on alphabets which are significantly different to the Latin ones, where large parts of the population do not normally use Latin writing, this is especially problematic. It is difficult for ordinary users to remember domain names which appear in Latin letters in announcements, advertisements, and so on.

Even in Norway, where the alphabet is based on Latin letters, there are problems associated with the lack of national characters. The solution so far has been to transliterate domain names, that is, to replace the national letters with the closest corresponding letters from a to z. The usual standard for transliteration of the most frequently used Norwegian letters is that æ becomes a, ø becomes o and å becomes a, so that the domain name blåbærsyltetøy.no ("blueberry-jam.no")is registered as blabarsyltetoy.no. Transliteration, however, makes the names more difficult to remember, since it weakens the link between the word and the domain name. In addition, the meaning of some words changes when they are transliterated. For example, the Norwegian word klær ("clothes") becomes klar ("clear" or "ready").

Because of such problems, international efforts to enable the use of national characters in domain names have been in progress for a long time.

1. Consequences of national characters in domain names

Why were national characters in domain names not introduced long ago, seeing the demand for them is so strong? The main reason is that the domain name system is so critical to the Internet. Nobody wants to make changes which could either upset the stability of the system or cause it to be split up into a number of subsystems which cannot communicate with each other. In other words, there are stringent technical requirements for solutions intended to handle national characters. The introduction of national characters will also have a number of legal and financial consequences. The sections below describe possible models for solutions, and their consequences.

Technical support for the domain name database

The technical aspects of introducing æ, ø, å and other national characters are wide-ranging. The domain name system (DNS) must be the same all over the world if the Internet is to function internationally. The technical solution for national characters should not require reconfiguration of all the millions of name servers in the world. In addition, it should function for all types of name servers and additional applications using domain names, for example, e-mail programs and browsers.

Currently, there are two primary technological approaches to the problem. One involves sending the domain names over the Net with the characters used by the local application. This means that blåbærsyltetøy.no, for example, would be sent over the Net with intact coding from the local computer. Since the name servers do not handle this type of coding at present, this approach would make it necessary to reconfigure all name servers. This is called a server-side approach.

An upgrade of all the name servers in the world, including countries which do not need national characters in their domain names, would be a difficult and expensive task. In addition, it would involve changing the domain name system itself, creating the risk that something could go wrong with a system on which the Internet depends. If the name servers are not upgraded, the consequence will be that it is not possible to access all Web pages and that e-mail will often disappear.

The other approach is to translate the local coding into ASCII compatible encoding (ACE) on the user's computer. In the ACE version of the name, only the characters a-z, the figures 0-9 and hyphens are used. The way in which this translation is performed (the algorithm) will determine exactly how the name will appear. This name can then be sent over the Net without problems and used in the domain name system without updating the name servers. Instead, users must install programs which can translate to and from ACE on their local computers. This is called a client-side approach.

An example of ACE translation might look like this:

The advantage of this solution is that it is not necessary to change the domain name system itself, making it possible to avoid the main problem associated with the server-side approach. In addition, the services function: e-mail arrives and Web pages are available even if not all users have installed the translation software. Users without such software on their computer will see the ACE version of names with national characters, and certain addresses will not look especially attractive or legible, but the addresses will work. The disadvantage is that the length of names with national characters may be a problem, and that users must install translation programs. Gradually these will probably become a standard component of browsers and other software.

Choice of technical approach

The Internet Engineering Task Force (IETF) is an open group of researchers, engineers, operators and suppliers responsible for the development and standardization of most of the protocols used in connection with the Internet. A working group under the IETF has taken on the task of finding a standard for national characters in domain names. After several years' work, there is now agreement on a proposal for a standard based on the ACE principle. The reason for the choice is that this solution will cause the fewest transitional problems and the least risk of something going seriously wrong with the domain name system. The proposal has been approved by the IESG (Internet Engineering Steering Group), and on 11 February 2003 the standard was completed when the Internet Assigned Numbers Authority (IANA) chose the identifier that the ACE version of the translated domain names is to include to show that the name has been translated.

The decision on the standard and the identifier is vital to the realization of national characters in domain names. Now the task of implementing it for all the registries all over the world has begun. For more information about the standard itself, see:

• RFC3454: Preperation of Internationalized Strings ("stringprep")
• RFC3490: Internationalizing Domain Names in Applications (IDNA)
• RFC3491: Nameprep: A Stringprep Profile for Internationalized Domain Names (IDN)
• RFC3492: Punycode: A Bootstring encoding of Unicode for Internationalized Domain Names in Applications
• The IDN group's Web site

Other technical consequences

Domain names appear in many places other than in the domain name system. They are published in Norid's Whois database together with information about who has registered them, they are specified on invoices so that subscribers can clearly see what they are paying for, they are recorded in customer databases in various connections, and they appear in letters from registrars to their customers. The introduction of national characters in the domain names will have consequences for all these systems.

As an example, somebody who receives e-mail from a domain name with national characters might want to look up information about who the domain subscriber is, as they can do for other domain names. So it must be possible to type national characters for lookups in the Whois database. The more types of characters and alphabets allowed in domain names, the more types of characters the database must handle. With an ACE solution there will be two versions of each name with national characters, one with the national characters, blåbærsyltetøy.no, and one with the translation to ASCII, xn--blbrsyltety-y8ao3x.no. It should be possible to look up both versions in the Whois database and get the same answer.

The same issues apply to billing of the annual fee for the domain name and to customer databases which are to include the domain name. In both cases, both versions of the domain name must be treated as equivalent, and it must be possible both to enter national characters in the database as well as printing them on the invoice.

It follows from this that the greater the number of national characters that must be accepted, the more flexible and complex the programming supporting the registration process must be, in order to handle the various alphabets. If many different alphabets are to be accepted, such great demands will be placed on the flexibility of the supporting software that there is probably no single billing program or database that could fulfil the requirements. It could then become relevant to introduce a registration model where the central registry, in our case Norid, works only with the translated version with a-z, and where the task of presenting the domain name in the national alphabet is assigned to the registrars. On the one hand, a model of this kind requires that the registrar has the necessary technical competence. On the other hand, it creates the potential for using a large number of different alphabets. The model will enable registrars to specialize in specific alphabets and make the most of this specialization as a competitive advantage. For example, one registrar could work with Hebrew, while another could specialize in Chinese characters.

Regardless of where the responsibility for presenting the names is placed, a greater number of alphabets will result in higher costs in the registry system, which in turn will be reflected in the prices of domain names. The costs associated with introducing the characters typical of the Norwegian languages are not especially high, since very few characters are involved and much of the supporting software in Norway uses these characters already. However, software in general use offers little support for other alphabets such as Hebrew, Chinese and Cyrillic, and introducing these will entail costs on a completely different scale.

Legal consequences

The legal issues relating to domain names which contain national characters will on the whole be the same as for the domain names in use today. The choice of name will be the applicant's responsibility. She or he must ensure that the name will not lead to unlawful interference with the rights of a third party and that it is not discriminatory, defamatory, pornographic, or illegal in any other way.

However, the introduction of domain names with national characters will have a number of legal consequences. The introduction itself opens a larger namespace, which in itself increases the number of possible conflicts over rights. In addition, some alphabets include characters that look identical to those in other alphabets, for example, characters in the Cyrillic alphabet and in the Latin alphabet. This creates the risk of problems with both rights and security.

For example:

1. ВЕНЕМОТН [Cyrillic letters]
2. BEHEMOTH [Latin letters, Helvetica]
3. BEHEMOTH [Latin letters, Times, italics, red]

(This example may look different in different browsers, depending on which fonts are installed.)

1 and 2 are written in Cyrillic and Latin letters respectively. Coincidentally, they look almost exactly the same, but have completely different meanings. The difference between 2 and 3 is in the colour and font, which does not involve any difference in meaning. For the coding of characters, only differences in meaning are taken into account. So a computer will consider 2 and 3 to be identical, and completely different from 1. From the user's point of view, however, all three words are identical, although the appearance is slightly different. Users are accustomed to seeing words printed in many forms and colours without this changing the meaning of the word.

So what is the problem? If a user who receives an e-mail message with the domain name in Cyrillic letters (1) cuts it out and pastes it into another program, the address will be Cyrillic. However, if he receives the domain name on a business card or letterhead, he has no way of knowing whether the actual domain name is written in Cyrillic characters, Latin characters, or a combination. Domain names are printed in many fonts and colours, and if the user decides to type Latin letters when the text is actually in Cyrillic letters, his e-mail could end up in completely the wrong place.

In addition to possible security problems, there are certain legal issues associated with this. Is the trademark BEHEMOTH specifically associated with the Latin alphabet, or does it also cover other alphabets where the characters happen to look the same, but are not pronounced the same and do not have the same meaning as in the Latin alphabet?

Here, too, the complexity and the risk of problems increase in proportion to the number of alphabets to be supported. If the range of characters that people are allowed to register is restricted, it will be possible for the registry to check that the alphabets permitted do not contain characters which look the same. It is possible to solve the problem illustrated by the example above by supporting only one of the alphabets, such as the Latin one. In this case, sensitive e-mail sent to the Cyrillic version of the domain name in error would be returned, because registration of this name is not allowed. When there is only one visual version of the characters in a Norwegian domain name, the person who holds the rights to the trademark or company name BEHEMOTH does not need to consider whether the rights extend to all characters which give the same visual impression as the word in the trademark.

Resolving disputes out of court

Regardless of this change, Norid plans to establish an alternative conflict-resolution body to evaluate disputes over rights associated with domain names. The introduction of national characters could have a significant impact on the costs involved in resolving such disputes. The conflict-resolution body must have the competence to be able to handle all characters and alphabets which may be registered, and to assess whether they represent infringement of a third party's registered or unregistered rights. This is no problem as far as Norwegian characters are concerned, but immediately becomes more complex if, for example, Chinese characters are to be assessed in relation to a Korean trademark, where the same characters may have completely different meanings in the two written languages. Such evaluation will require special competence in proportion to the number of alphabets and the complexity of the alphabets for which registration is permitted. In turn, the greater the number of alphabets permitted, the more expensive the dispute-resolution scheme will become, and these costs must be covered by the ordinary domain subscribers.

Other consequences

Although it is technically possible for the domain name blåbærsyltetøy.no to function all over the world, this does not necessarily mean that it is a good name in the international context. International customers will have major problems in finding the company's Web pages if they only have the domain name on a business card and must try to enter characters they do not have on their keyboards. People who choose domain names with national characters should therefore think carefully about whom they are trying to reach with the domain name, and whom they probably will not reach, just because of the national characters.

2. Restrictions on characters under the .no domain

The client-side approach has been chosen as the standard, and there are thus no technical consequences associated with the domain name system itself involved in supporting an unlimited number of alphabets. In all cases, all domain names will be encoded as strings which consist only of the characters a-z, 0-9 and hyphens, so that the domain name system does not need to know which alphabet the coding is intended to represent. A domain name with Chinese characters will be treated in the same way as one with Cyrillic characters in lookups to find the IP address to which they belong.

Outside the domain name system, however, there will be major financial, legal and security-related consequences related to the choice of the number of alphabets to be supported. As described in the sections above, the costs associated with supporting systems and conflict-resolution bodies will increase with the number of alphabets to be supported. These costs must in principal be covered by everyone who registers a domain name, whether it contains national characters or not, and it is therefore reasonable to place some limitations on the alphabets which it would be natural for the Internet community in Norway to pay for. In addition, there are significant practical problems associated with having many alphabets with characters that look the same.

Norid proposes that the number of alphabets permitted under the .no domain should be limited. Since only Norwegian organizations can register domain names under the .no domain, it would be natural to restrict these to the official Norwegian written languages: bokmål, nynorsk, nordsamisk (North Sámi), sørsamisk (South Sámi) and lulesamisk (Lule Sámi). A delimitation like this will mean that the introduction of national characters in domain names will have minor financial consequences for the registration of other domain names; neither will security holes arise because of characters which look the same without actually being identical.

To ensure that these delimitations are in line with the wishes of the Internet community in Norway, the issue has been discussed in Norid's advisory council for policy issues, Norpol. Norpol includes representatives of users, industry and the authorities. Its objective is to provide Norid with an opportunity to gather signals on the subject of domains from the Norwegian Internet community.

Norpol recommends that registration should be permitted only for characters which are letters and which belong to the official Norwegian written languages. The council recommends that neither characters from other languages nor non-language symbols should be supported. Norpol also recommends that if the introduction of Sámi characters involves high costs, the benefits of the change should be weighed up against the costs, and the work should be coordinated with Sweden and Finland if possible.

The official languages

The official written languages in Norway are bokmål, nynorsk, nordsamisk (North Sámi), sørsamisk (South Sámi) and lulesamisk (Lule Sámi).

The report "Nordic Cultural Requirements on Information Technology" ¹⁾ includes the following characters in the Norwegian and the Sámi alphabet:

Characters marked "1" are specified in the report as necessary for correct orthography; characters marked "2" are used in some names, but are not part of the normal orthography.

To ensure that the list of characters is as up to date as possible, Norid has asked the Norwegian Language Council and Sametinget (the Sámi parliament) to provide an overview of the characters which currently form part of the official written languages in Norway. This is especially important for the Sámi written languages, since what is marked as Sámi in the report covers several Sámi written languages, including some which do not have the status of official written languages in Norway. In line with Norpol's advice, Norid proposes that only characters which are letters and which belong to languages with official status in Norway be supported.

The Norwegian Language Council includes the following letters in bokmål and nynorsk:

• the letters a-z
• the letters specific to Norwegian: æøå
• the characters à, ç, è, é, ê, ñ, ò, ó, ô, ü, ä

The Norwegian Language Council points out that the apostrophe (') and ampersand (&) are also characters used in Norwegian writing. The apostrophe is used very often, for example, in contractions of words, while the ampersand is seldom used. The introduction of characters which are not letters will lead to a number of technical problems, and Norid will therefore follow Norpol's recommendation to allow onlyletters from the official written languages in Norway. Norid has however noted the Norwegian Language Council's suggestion, and will reconsider it if developments result in a solution to the technical problems associated with implementation of such characters.

The Norwegian Language Council also comments that there are a number of borrowed common nouns and names that are in normal use in Norway, especially personal names and geographical names, which contain characters that do not form part of the normal Norwegian alphabet. These are primarily characters in normal use in the languages of the Western European cultures. The council provides a non-exhaustive list of examples of these characters. Norid regards a category of this nature as problematic, since questions can be raised about which such names are to be regarded as so special that the characters in the name should be permitted. Again, we support Norpol's recommendation to allow only letters included in the official written languages in Norway, emphasizing those that either form part of the normal orthography in the written languages or are included in Norwegian personal names. Norid has however noted the comment from the Norwegian Language Council, and will reconsider it if it becomes feasible to provide for a wider range of characters than those in the official written languages at a later stage.

Sametinget includes the following letters in North Sámi, South Sámi and Lule Sámi apart from a-z, æ, ø, å:

• North Sámi: á, č, đ, ŋ, š, ŧ, ž
• South Sámi: ingen
• Lule Sámi: lulesamisk: á and ń

To ensure an up-to-date list of characters which are used in names, including characters outside the normal orthography, Norid has asked Norway's national population register (Folkeregisteret) for advice. The national population register allows the following letters to be registered in personal names:

• the letters a-z
• the letters specific to Norwegian: æøå
• the characters ä, é, è, ô, ö og ü

Conclusions regarding the selection of characters

On the basis of information from the Norwegian Language Council, Sametinget and Norway's national population register, Norid has created an up-to-date overview of letters which form part of the official Norwegian languages. The letters in question - except a-z - are shown in the table below together with the Unicode identifier²⁾ for each letter and an indication of the written language in which the letters occur. Characters used in personal names are shown in a separate column, instead of being divided between the two groups of written languages.

Characters marked "1" are necessary for correct orthography; characters marked "2" are used in some names, but are not necessarily part of the normal orthography.

In line with Norpol's recommendation, Norid proposes that these characters, together with the letters a-z, the digits 0-9 and the hyphen character be permitted in domain names under the .no domain.

Comparison with relevant registers

We have also gathered information about which letters the Brønnøysund Register Centre allows for the registration of organization names in the Central Coordinating Register for Legal Entities. Currently, organization names may only consist of the letters a-z and the Norwegian letters æ,ø and å. Other characters or letters from other languages are not permitted.

Since Norid plans to enable registration of all letters which are permitted both in this register and in Norway's national population register, the limitations should not be perceived as restrictive in relation to what is customary in other respects for the registration of names in Norway.

3. Transitional process

When the range of characters to be supported has been selected, a proposal must be made for a transition process from the current regulations to regulations permitting national characters. This transition can be undertaken in many ways:

First come, first served

The date and time for the transition to the new regulations is to be stipulated. At the specified time, it will become possible to submit applications under the new regulations. The applications will be processed in the order in which they arrive at the registry.

Drawing of lots

During the transition, it will be possible to accept applications under the new regulations for a limited period. These will be stored without any registration being undertaken. At the end of the period, no further applications will be accepted. At the same time, duplicates in which the same organization applies several times for the same name will be removed. When only one copy remains, the system will "draw lots" to determine the sequence for processing applications for the same name. When the draw has been completed and all the applications have been processed, the system is opened for the ordinary receipt of applications in accordance with the new regulations.

Preferential rights

Many subscribers who currently have transliterated domain names want a domain name that is written "correctly" when it becomes possible to register national characters. If these subscribers are to have preferential rights, this can be achieved by allowing only subscribers with such transliterated domain names to apply, and then only for the corresponding domain with national characters. In this case, this means that the person who subscribes to blabarsyltetoy.no will be able to apply for blåbærsyltetøy.no. At the end of such an application period, there would then be a general transition to the new regulations, whether on a "first come, first served" basis or by drawing lots.

Equal treatment or preferential rights?

The first question which must be decided is whether all applicants are to treated in the same way, or whether some applicants are to be given preference over the others. One of Norid's basic principles for the administration of the .no domain is neutrality, and equitable treatment is a central part of this principle. This does not exclude the possibility that we can have procedures which distinguish between various groups of applicants, but with equitable treatment within the group. There must however be good reasons or particular requirements for special treatment of a group of applicants.

The current regulations state that"Registration of a domain name does not grant greater or different rights to the name than the applicant had previously. Registration does not entail any restriction on rights that others may have to use the name." The fact that a subscriber has already registered a similar name thus gives no preferential rights to a new domain name.

Even if we should choose to go against the regulations and grant preferential rights, cases where there is doubt will occur. Several versions of the same word may be registered. For example, the three subscribers who have registered bat.no, baat.no and boat.no (the Norwegian word "båt" means "boat") could all argue that they registered the domain name because they could not obtain båt.no. It will not be obvious which of these three should have a preferential right to båt.no.

It is also relevant to argue that if the registry is to take into account anyone's "rights" on the basis of a previous registration, one should also take into account that someone may hold rights based on a trademark, company name or other registered or unregistered rights. Such considerations lapsed when the current regulations were chosen, because the Internet community desired regulations which gave applicants great freedom of choice and which at the same time paved the way for rapid, unbureaucratic and reasonable procedures for processing cases.

Norid cannot see that good enough reasons exist to give preference to certain applicants based on previous registrations, especially since the current regulations expressly do not grant other rights through the registration of a domain name than those which the applicant held previously. We therefore recommend a transitional process in which all applicants are treated in the same way. Norpol supports this view.

Choice of transitional mechanism

Provided that all applicants are to be treated in the same way, Norid recommends that the transition should take place through drawing lots. Alternative transitional mechanisms and the consequences of these have been put forward to Norpol, which supports Norid's recommendation for the use of drawing lots.

Why draw lots?

Drawing lots is a mechanism with which the Norwegian Post and Telecommunications Authority, among other institutions, has positive experience from other allocation situations. This transitional mechanism was also used during the liberalization of the regulations for allocation of domain names in February 2001, and its implementation at that time proceeded virtually faultlessly.

It is a fair mechanism in which everybody has an equal chance to win. The mechanism also provides a variety of technical and practical advantages:

• The transitional period will give Norid time to check the applications for formal errors and provide feedback to the registrar if an application is rejected. The registrar can thus submit a corrected application before the draw itself starts.

• Removal of duplicate applications means that an applicant will not be able to improve his or her chances by submitting several applications for the same name. Neither will an applicant benefit from going to many registrars at the same time and burdening the registrars with duplicates.

• The date during the transitional period on which the application arrives will have no significance. It will thus not be critical for the applicant if the registrar or Norid experiences technical problems on a given date. There will also be no distinction between the registrars who check their e-mail service themselves and those who have chosen to rent the service from another Internet provider.

The disadvantage of the mechanism of drawing lots is that the implementation will take somewhat longer than for a "first come, first served" transition. This is primarily a problem for applicants who are not actually affected by the change but must still wait because of the transitional period. For this reason, we have previously chosen to implement minor changes without the use of drawing lots. The change we are now facing will however have far-reaching consequences for many applicants, and we have therefore concluded that the advantages of drawing lots clearly outweigh the disadvantage of a longer transitional period.

4. The road ahead

As this document shows, there are a number of complex issues related to domain names with national characters, and some background knowledge is required if one is to see the consequences of different choices. A general hearing is thus not necessarily appropriate for deciding which regulations the .no domain should have regarding national characters. Instead, Norid has chosen to prepare a proposal based on the signals Norpol has given regarding the desires of the Internet community in Norway. In addition, relevant authorities on the Norwegian written languages have been contacted directly for comments. As final quality assurance of the proposal, a comparison has also been undertaken between the number of letters we propose and the number permitted in other name registers in Norway.

On this basis, Norid's starting point is that the proposal is in line with the wishes of the Internet community in Norway. It is however possible for interested organizations and individuals to provide comments on the proposal. Comments should be sent to info@norid.no by 22 April 2003. Mark all comments "Domain names with national characters". After this date, Norid will prepare a final proposal and forward it to the Norwegian Post and Telecommunications Authority for final review.

Norid will start the implementation programme when the new regulations have been chosen. Detailed information about the new regulations and transitional procedures will be published at least three weeks before the introduction itself.

5. References

1. Nordic Cultural Requirements on Information Technology
   INSTA technical report
   STRÍ TS3, 1st edition, 1992
   ISBN 9979-9004-3-1

2. Unicode http://www.unicode.org/