Hot on the heels of other ICANN Internationalized Domain Name (IDN) Top-Level Domain (TLD) launch errors1, we now have another example of ICANN's failure to comprehend the differences between IDN and ASCII names, this time to the detriment of potential IDN registrants and the new IDN generic TLD (gTLD) Registries. This gaff really makes you wonder whether the SSAC and Multilinguism departments at ICANN have ever met.
I'm sure you are by now aware of the late-to-the-party security concerns regarding "name collisions" resulting in "NGPC Resolution for Addressing the Consequences of Name Collisions”.
This resulted mainly from router manufacturers and corporations deciding to use "made up" TLDs for their internal and sometimes public-facing systems. Bad idea. Things like ".home" for routers. Godaddy even used host names ending in "corp.gd" for internal mail servers2.
Anyway, somewhere along the way it was decided that all the illegal hostnames appearing in the DNS root server NXDOMAIN stats must be caused by people setting up silly hostnames in routers and other configuration errors, thus every NXDOMAIN entry must be a security problem.
So now we get to the stage where the first 4 new gTLDs get launched, and ICANN publishes the "block list" that is going to fix all these security issues. Let's concentrate on the 2 Cyrillic new gTLDs — .сайт (= .site in English) and .онлайн (.online in English) — those are quite nice TLDs if you ask me. Let's take a look at the blocklists published: xn--80asehdb & xn--80aswg
Despite the fact that the Registry for both of those TLDs (CORE) never intends to offer ANY ASCII registrations in these 2 TLDs, the block lists have a huge list of ASCII SLDs they have to specifically block. If we ignore all those ASCII strings needlessly listed, we're left with the Cyrillic strings.
According to ICANN, all these Cyrillic strings are "collisions" and therefore a security concern. Let's take a look at a sample of them, and translate them into English for those of you that don't understand Cyrillic characters…
|бг||(xn--90ae)||bg = Bulgaria.online|
|булбанк||(xn--80aba1bco2b)||bulbank.online A large bank in Bulgaria|
|европаплюс||(xn--80adi0angbeo3k)||europaplus.online Moscow FM radio station 106.2 MHz|
|интерны||(xn--e1afpbok0e)||интерны.online (Russian TV series)|
|ігри||(xn--c1akx0g)||games.online (in Ukranian)|
|люксфм||(xn--j1abetp1d)||luxfm.online (FM radio station)|
|месси||(xn--e1agkva)||Messi.online (football player)|
|яндекс||(xn--d1acpjx3f)||Yandex - the most popular search engine in Russia|
Sorry for the long list, but it really demonstrates the magnitude of ICANN's error.
Do any of the above looks like made-up hostnames people are likely to have loaded into router configurations? Is there even any router configuration interfaces that let you type Cyrillic into a hostname field and automatically convert it into the punycode required to correctly use it in DNS?
Another question: how many of you, when sitting on a search engine HTML page on your browser, accidentally grabbed the mouse, clicked on the URL bar instead of the search bar and typed in a search term (lets say "online game" for instance) and then been a little embarrassed when the browser has gone off to the DNS and produced an error because there isn't a domain "online%20game". I know I have and I'm willing to bet you've done it too. No doubt they're all in the NXDOMAIN stats.
Now, if you take a look at that big list of ICANN-designated "collisions" above, doesn't it look like it could be a list of the top Russian search phrases that include the word "online"? Myself, I think that's exactly what it is. Maybe there is a browser somewhere out there that is used regularly by people in Russia (or another country that uses Cyrillic) that, when confronted by a search phrase typed into the URL bar, converts (DNS illegal) spaces into (DNS legal) .'s instead of URL encoded %20's and heads off to the DNS for a resolution attempt?
Here's a quick test that suggests this is indeed true — go to http://www.yandex.ru and paste "онлайн" into the search bar then press the space bar. Look at the list of suggestions that appear. All of the 2 word ones that appear in the drop-down list are in the .онлайн blocklist. Whoops!
The other potential reason for such "collisions" is that new Internet users in these countries, not knowing the history of the Internet and the old DNS ASCII-only restrictions quite reasonably expect that when they type a Cyrillic domain name into the URL bar, it's going to work. I can imagine they get quite confused when it doesn't and isn't that the whole reasoning behind IDN TLDs in the first place? ICANN finally launches these IDN gTLDs and just beforehand hobbles them by placing the most potentially popular domain names in a "collision" list.
Can you imagine the confusion that is going to ensue when CORE is finally allowed to open registrations to the general public and the most popular choices are artificially blocked? "I'm sorry the enormously popular phrase you have chosen to register has been deemed to be a security threat by ICANN and so cannot be registered".
The Internet users keen to try out the new IDN gTLDs in their web browsers are most likely going to try the domains blocked by this list (what domains would you try first, if you heard .online was alive?) — it might take them 20 tries before they finally strike an obvious hostname that isn't on the block list. Opera — a very popular browser in Russia — is going to report back "Network problem – Check that the address is spelled correctly, or try searching for the site." (well, in Russian of course), along with a google search box, sending the excited IDN new gTLD experimenter off to a search engine in .com/.ru, thus negating the whole damn reason for launching these IDN gTLDs in the first place. A lot of them might give up before then and decide these newfangled IDN gTLDs don't actually work.
Fadi Chehadé, the CEO of ICANN, is a supporter of the concept of IDN's (yay!) and is fluent in Arabic. Perhaps he should take a long hard look at the blocklist for .شبكة ("network" in Arabic — one of the other of the first four new gTLDs to launch)3, have a think about the arabic words in that blocklist4, maybe consult with some language and foreign SEO experts and then make some urgent adjustments to the blocklist methodology in time for these first 4 new gTLDs to launch without restrictions?
1 (a) Intellectual Property Constituency blocking of Verisign transliterated .com special launch requirements, based on verifiably erroneous and weak data, to the detriment of hundreds of thousands of existing IDN.com/org Registrants. See:
(b) ICANN require that the ONLY domain that is allowed to be resolved when the TLD is first launched is the ASCII string "nic" — IDN new gTLDs are not even allowed to offer the equivalent string in the script the entire TLD will operate under. Just another ICANN cultural error. This no-activation window was chosen to match the 120-day period that the CA/Browser Forum gives its certificate authority members to revoke clashing certificates. Have they even issued any certificates in xn-- format?
2 "Received: from unknown (HELO gdmailer05.dc1.corp.gd) (22.214.171.124) by m1plcorpmail001.prod.mesa1.secureserver.net"
4 Hint: It includes world.network islamic.network arabic.network AlJazeera.network and health.network
By Andrew Gardner, Internet User
|Cybersquatting||Policy & Regulation|
|DNS Security||Registry Services|
|IP Addressing||White Space|
Minds + Machines
Afilias - Mobile & Web Services