#icannfail number 27 - The IDN Debacle

#icannfail number 27 - The IDN Debacle

In the late nineties when application programmers began to understand how the fairly byzantine domain name system really worked, and began manipulating domain names in apps, besides the obvious policy mire, the issue of multilingual domain names emerged.

It's all very well to *say* "the Internet is for everybody", but if they don't speak, or read or write English, what good is to somebody?

So, work began in earnest to implement domain names in almost a dozen languages and not just the easy cases like "Oh with two dots over it" but the hard ones like the Asian languages: Korean, Chinese, Japanese and the even harder right-to-left scripts like Arabic, Farsi and Hebrew.

Around 1983 two of us wrote all the software for the first bilingual English/Arabic computers and manipulating Arabic characters was a real eye opener from a programming standpoint: 29 characters, not 26 - and six forms of each character: beginning form, ending form, stand alone, etc, because Arabic words are always joined together like our handwriting is. Arabic words are one a series of characters, it's one long "pen stroke" in the shape of a single word. Not a series of individual unconnected letters.

Plus there's the interesting case of "lam-aleph". Lam and Aleph are two characters, but when they appear next to each other they turn into a single Lam-Aleph character. This is a real hair puller for a programming counting characters. Then we have the descending and ascending diacritics like hamsa. It's complicated.

But, after 10 years and countless hundreds if not thousands of man hours it all works. I can cut and paste Korean, or Greek, or Russian or Chinese domain names. http://مثال.آزمایشی/صفحه‌ی_اصلی

Now people can read the net in their own language.

Except for two problems. Frist, ICANN wants to be paid, and paid handsomely every time they put two lines of text (ironically in computer they don't own or run) into one particular file they improperly have aegis over, to make these all work for everybody. Currently they exist in alternate root server systems all over the Asian subcontinent, although they will claim they're "private root", a semantic quibble - thye work for anybody that configures theur nameserver to use them - anybody anywhere in the world. "The Internet is for everybody" remember?

Second, they can't use any domain less than three words long.

As anybody who has ever heard the traveling ICANN roadshow knows, terseness is not their forte. Ironic because the point of the DNS they have aegis over is domain names were intended, by definition, to be mnemonic, they is an abbreviation of words, not words themselves. For efficiency, and to make it easier for people to remember.

This is why we have "" and not "

Get it?

Right, so the problem is, Chinese (and other) symbols are not letter, they're words. And the Government "Advisory" (really censorship) Committee has decreed that no new top level domains will be less than three characters, to protect their country's country code domains: de, uk, ro, etc.

This is a dubious principle at best, if they insists on "purity" of the "two letter country code" space then country code domains shouldn't act like generic TLDs. You can't have it both ways.

Some country code people suggest "oh, we're not com, we have a rigorous procedure, and an organization that's meant for entities in our own country and our laws are different than the US where the legal seat of all this is so we're not really "generic" or part of that .com mess.

And then there are ones like .gd ( - many "ccTLDs" that will let anybody get any name, heck, .tk is even free. That's as generic as you can get. That's more generic than dot com. At least com sounds like it might have something to do with commerce, not quite as generic as tk, the ultimate arbitrary string. I don't think even the people that maintain the ISO-3166 list it comes from could even tell you the name of the tiny island its for without looking it up.

To get some sort of dispensation from the rules we apply to "generic" top level domains is inconsistent with being one of them as well! Saying "but we're a two letter domain" isn't going to cut it, nobody is fooled any more: you're a *generic* two letter domain. Thanks for paying.

So, this two letter rule, whatever the merit of it, really only applies to the Roman alphabet. "commerce" got shortened to "com".

But in Chinese, a hypothetical character for "commerce" could not be used because it's only one character and ICANN has an equally foolish ban on single letter top level domains for reasons absolutely nobody can actually explain (hint: that's the way we've always done it) so that's right out.

So maybe "for commerce" in Chinese - two characters might not be so terrible if just plain old "commerce" isn't available. But now you run up against the GAC's two letter rule.

So Asian domains have to have at least, three characters as a top level domain.

Quick, this is a test: abbreviate the word "commerce". Use have to use three words though. WTF?

It's been a while since I've seen such a glaring inconsistency of logic like this, but it is probably to be expected from the byzantine sprawling road show that is the ICANN organization.

ICANN demands that each of the translated TLDs - say you'd like to see "com/net/org" in your own language - is subject to a $50K fee. It would me $157 million to translate the current tlds into all eleven languages under this ill-thought out policy. To recap: tens of thousands of man hours went into making the seemingly impossible work and one high rent California company wants a Kings ransom to pull the switch to turn it on. For what?

For WHAT? ICANN did nothing except stall the process. And in the height of hubris ten years of other peoples work to male an incredibly difficult linguistic feat of computer engineering work that moves forward the idea of universal inclusion and communications between people and ICANN wants to cash a large bunch of checks and do nothing more than that? And nothing happens till they're paid? That's extortion and let's not even get into "taxation without representation".

And what is it they do?

Well, lets look at an example of the massive disconnect between those who do the actual work and those who do nothing more than cash checks and run around the world saying "this is how it's gong to be, decisions have been made, you can go home now".

To paraphrase Lincoln: "Dear ICANN, if you don't want to use our Internet we'd like to borrow it for a while".

ICANN was supposed to be an institutional framework for IANA, instead it's just embarrassing; I agree with Feld and Rutkowski: shut it down and keep the NTIA away from the net. Three strikes, they're out.