Tuesday, September 27, 2011

Accented characters in domain names are a terrible idea

Normally I ignore the various e-mails I get from CIRA (the Canadian Internet Registration Authority, which oversees .ca domains). This includes requests to vote for CIRA boards of directors, and other stuff that I don't remember because I ignored it.

However the current consultation on internationalized domain names for .ca domains is worrisome.

At the consultation home page you can follow the links to the (relatively straightforward and plain-language) proposal, and provide feedback on how they should implement it (not whether it's a good idea).

There is also a discussion forum that discusses four questions. The first three are variants on "isn't this such a great idea?" and the fourth is "any other comments?"

There are some recurring themes--almost entirely emphatically opposed to this idea as proposed--and I submitted the following comment to the thread (note: my questions are literal, not rhetorical. The intent is for the consultants and CIRA to consider those issues; I don't need to know the answers myself).
I've read through the entire thread and I have a number of comments:

I am one of those Ottawa-dwelling folks who spends a lot of time thinking about how to properly deliver services to my members, clients, and correspondents in the language of their choice. I hope the consultants thank CIRA for the big paycheque and tell them how bad an idea this is.

The logical decision would go like this:

1. It is fair to allow people to register accented domain names.
2. In order to prevent abuse, "Names that differ only in accents should not be allowed to be registered to different organizations." (as worded by ahooper in comment 60)
3. This would require considerable resources to implement.
4. To pay for this implementation, either domains will have to get a lot more expensive, or CIRA and/or DNS operators will have to take a big hit and allow all variants to be registered to the same user for the current price
5. If the goal of the policy is fairness (point 1), it would be unfair if the policy reduced the accessibility of domain names to individuals and organizations with limited finances.
6. Similarly, it would be unfair to require DNS operators to absorb the full cost of this implementation
7. Therefore, the benefits of the proposal are far outweighed by the costs of implementing it.

Other new comments not already raised:

I have experienced problems with automatically-generated usernames based on the user's name (e.g. FirstnameLastinitial). For StéphanieR, there are some OS/Browser combinations where she will not be able to enter her username; the system will interpret it as StÈphanieR and reject the user. I've experienced similar problems with database exports that have similar errors if they aren't opened in Windows. What will CIRA do to ensure that input from all browser/OS combinations will be correctly interpreted by the DNS?

The company I work for recently received its registration certificate for a couple of its French trademarks. The name on the government-issued certificate was spelled IN CAPITAL LETTERS with obvious accents omitted (i.e. MON NOM REGISTREE). How would we be able to enforce our rights to the domain monnomregistrée.ca if even Industry Canada is ambivalent to accents? (My mind is further blown to learn that CIRA operates under the auspices of Industry Canada)

People need to recognize that not every domain registrant is a corporation with millions of dollars of annual revenue. It is not so simple for some organizations to register the variants. I am involved with some community groups whose annual income is in the tens of dollars, and the domain registration is one of the few expenses. While they are all Anglophone groups, there are likely Francophone groups in the same situation who cannot afford both mélange.ca and melange.ca (not to mention mèlange.ca and mêlange.ca). Fredjubs' suggestion in comment 71 (that business information sugh as a GST number be required for domain registration) does not work for individuals nor for non-incorporated organizations (like many community groups). Further, there are (or at least there should be) privacy implicaitons if more and more information (passport number?!?) is required just to register a domain name.

Are accented characters case-insensitive, as unaccented characters are in DNS?

Elaborations of and rebuttals to other comments:

Elaborating on other comments: I'm offended by those who think that the only people who speak French are in Québec. There are many Franco-Ontarians (including many towns, like Sudbury, where many do not speak English), Manitoba was the first province to be officially bilingual (though it isn't any longer), and New Brunswick is recognized as being officially bilingual in the Constitution. So any solution based on "only for .qc.ca domains" is entirely insufficient. I don't consider myself to be one of these people, I'm merely offended by the total ignorance of those who make this suggestion.

Reiterating cbehnke's point in comment 53: When I give a website over the phone, and I say président.ca (in French), it will be very cumbersome for me to specify to the message recipient whether or not the domain has an accent, especially if the person is not french speaking.

Entering accented French characters is not a "solved issue":

As mentioned elsewhere, IDN is fine in countries like Germany or Spain where the entire population is expected to know how to speak and type accented characters in the dominant language of the respective country. However, since many, if not most, in the Internet-using population that corresponds to .ca domains do not normally speak and type accented characters, I believe this will become a net obstacle to universal access to the internet, not an enhancement.
- On macs (and apparently on Linux) it is easy to do. However, you must know how to do it. Those who generally correspond only in English (including with bilingual francophones who have domains with accented characters) will not know how to use it.
- On Windows, you have a few cumbersome options: if you have French installed, AND you have activated the desktop/menu bar icon for fast keyboard switching, AND you alreay know what keys to use to enter accented characters, then it's easy. If you don't, then you can use Character Map, which I'm sure a vast majority of users wouldn't even know where to look for it, if they knew it existed, and even then it is not simple or convenient.

User Laplante in comment 52 linked to a Microsost KB article on international keyboards. First, a user shouldn't need an FAQ to type a simple domain name. Second, how will a user know to visit that article when they discover they don't know how to type an accented domain name? You could put a link on the homepage of the domain name, but the user won't be able to go there! Catch-22!

I echo amadha's analogy in comment #45 about bilingual names vs. English and French names. Instead of typing "ottawapubliclibrary.ca", I have to type "biblioottawalibrary.ca", which is much more cumbersome, and harder to remember, even as a bilingual person. The point of the analogy is that it's intended to be more fair but ends up being more confusing.

pdobeda@acm.org in comment 86 suggests that two names with similar names will have already addressed the issue with their corporate registrations. This is a false assumption: for example Apple Computer and Apple Records both operate under the name "Apple" (not to mention numbered companies). And while accents aren't usually used with them, acronyms are often shared among many organizations.
If you have additional comments, or just want to give your "no" vote, I suggest you head on over to the IDN Consultation website and speak up before we all get inundated with spam messages from ínfÒ@vÌ4grá.ca.

- RG>

1 comment:

David said...

Accents and keyboards are one of those examples of how American English Canada really is.

All over Europe, country-specific keyboards are the norm, even in English-speaking Great Britain. This may have been due to most countries having their own currency with its own symbol until the launch of the Euro (whereas Canada, of course, just used the dollar sign used in the US). Countries with multiple official languages have keyboard layouts to accommodate the characters used in their official languages.

Canada, too, "on paper" has keyboard layouts to accommodate English and French. Indeed, Canada has two such layouts, French Canadian and Canadian Multilingual:

But good luck finding one for sale anywhere*, and if you do, chances are it'll cost you a lot extra. Some keyboards in the federal government are 'Canadian', but all too often the IT guys have simply left the OS (i.e. Windows) set to US:US so naturally the keyboard "doesn't work".

As with things like date formats and paper sizes, Canadians are essentially American in all but name.

*The exception is with netbooks, but here all hell seems to break loose. I recall seeing some pretty negative comments from some purchasers of netbooks that had international or Canadian multilingual keyboard layouts (I imagine the Asian creators of these netbooks couldn't imagine Canadians being such pricks when it comes to their own official keyboard layouts). In fact, the negativity is still ongoing:


Perhaps Lucien Bouchard was right all those years ago: Canada is not a real country.