Problems validating the addresses of .uk domain registrants, which caused one registrar to dump the TLD entirely, are broader than I reported yesterday.
Cronon, which does business as Strato, announced last week that it has stopped selling .uk domain names because in more than a third of cases Nominet, the registry, is unable to validate the Whois data.
In many cases the domain is subsequently suspended, causing customer support headaches.
It now transpires that the problems are not limited to .uk second-level names, are not limited to UK registrants, and are not caused primarily by mailing address validation failures.
Michael Shohat, head of registrar services at Cronon, got in touch last night to clarify that most of its affected customers are in fact from its native Germany or from the Netherlands.
All of the affected names are .co.uk names, not .uk SLDs, he added.
And the validation is failing in the large majority of cases not due to Nominet’s inability to validate a mailing address, but rather its inability to validate the identity of the registrant.
“This is where the verification is failing. The database they are using can’t find many of our registrants’ company names,” Shohat said.
“So 30% of our registrations were being put on hold, almost all of them from [Germany] and [the Netherlands], and 90% of them because of the company name. We checked lots of them and in every single case the name of the company was correct, and the address as well,” he said.
Michele Neylon of the ICANN Registrar Stakeholders Group said that Cronon is not the only registrar to have been affected by these issues. Blacknight Solutions, the registrar Neylon runs, has been complaining about the problem since May.
According to Neylon, the Nominet policy causing the issue is its data quality policy, which covers all .uk and .co.uk (etc) names.
The policy itself is pretty vague — Nominet basically says it will work with each individual registrar to determine a baseline of what can be considered a “minimum proportion of valid data”, given the geographic makeup of the registrar’s customer base.
Domains that fail to meet these criteria have a “Data Quality Lock” imposed — essentially a suspension of the domain’s ability to resolve.
Earlier this year, Nominet did backtrack on plans to implement an automatic cancellation of the names after 30 days of non-compliance, following feedback from its registrars.
“It’s disappointing that Cronon have taken this step; we hope they will consider working with us to find a way to move forward,” a Nominet spokesperson added.
She said that the registry has over recent years moved to “more proactive enforcement” of Whois accuracy. She pointed out that Nominet takes on the “lion’s share of the work”, reducing the burden on registrars.
“However, our solution does not include non-UK data sets to cross-reference with, so it is possible that some false positives occur,” she said. “Registrars with a large non-UK registrant bases, who are not accredited channel partners, would be affected more than others.”
An Accredited Channel Partner is the top tier of the three Nominet offers to registrars. It has additional data validation requirements but additional benefits.
While .co.uk domains are not limited to UK-based registrants, all .uk SLD registrants do need to have a UK mailing address in their Whois for legal service.
The company’s inability to validate many non-UK business identities seems to mean .co.uk could also slowly become a UK-only space by the back door.
German registrar Cronon, which retails domains under the Strato brand, has stopped carrying .uk domains due to what it says are onerous Whois validation rules.
In a blog post, company spokesperson Christina Witt said that over one third of all .uk sales the registrar has been making are failing Nominet’s registry-end validation checks, which she said are “buggy”.
With the introduction of direct second-level registration under .uk, Nominet introduced a new requirement that all new domains must have a UK address in the Whois for legal service, even if the registrant is based overseas.
According to its web site, Nominet checks registrant addresses against the Royal Mail Postcode Address file, which contains over 29 million UK addresses, and does a confidence-based match.
If attempts to match the supplied address with a UK address in this file prove fruitless, and after outreach to the registrant, Nominet suspends the domain 30 days after registration and eventually deletes it.
It’s this policy of terminating domains that has caused Strato to despair and stop accepting new .uk registrations.
“Databases of street directories or company registers are often inaccurate and out of date,” Witt wrote (translated from the original German). “The result: addresses that are not wrong, in fact, are be found to be invalid.”
Nominet is throwing back over a third of all .uk names registered via Strato, according to the blog post, creating a customer support nightmare.
Its affected registrants are also confused about the verification emails they receive from Nominet, a foreign company of which they have often never heard, Witt wrote.
I don’t know how many .uk names the registrar has under management, but it’s reasonably large in the gTLD space, with roughly 650,000 domains under management at the last count.
If Strato’s claim that Nominet is rejecting a third of valid addresses (and how Strato could know they’re valid is open to question), that’s quite a scary statistic.
Nominet seems to be using an address database, from the Royal Mail, which is about as close to definitive as it gets. And it’s only verifying addresses from a single country.
I shudder to imagine what the false negative rate would be like for a gTLD registrar compelled to validate addresses across 200-odd countries and territories.
The latest version of the ICANN Registrar Accreditation Agreement requires registrars to partially validate addresses, such as checking whether the street and postal code exist in the given city, but there’s no requirement for domains to be suspended if these checks fail.
[UPDATE: Thanks to Michele Neylon of the Registrars Stakeholder Group for the reminder that this RAA requirement hasn’t actually come into force yet, and won’t until the RrSG and ICANN come to terms on its technical and commercial feasibility.]
Where the 2013 RAA does require suspension is when the registrant fails to verify their email address (or, less commonly, phone number), which as we’ve seen over the last year leads to hundreds of thousands of names being yanked for no good reason.
If Strato’s story about .uk is correct and its experience shared by other registrars, I expect that will become and important data point the next time law enforcement or other interests push for even stricter Whois rules in the ICANN world.
DreamHost, a web hosting provider which says it hosts over 1.3 million web sites, has been hit with a lengthy ICANN compliance notice, largely concerning alleged Whois failures.
The breach notice raises questions about the company’s popular free Whois privacy service.
Chiefly, DreamHost has failed to demonstrate that it properly investigates Whois inaccuracy complaints, as required by the Registrar Accreditation Agreement, according to ICANN.
The notice contains numerous other complaints about alleged failures to publish information about renewal fees, its directors and abuse contacts on its web site.
The domain highlighted by ICANN in relation to the Whois failure is senect.com
ICANN sent three compliance notices to DreamHost concerning a Whois inaccuracy report for the domain name
and requested DreamHost demonstrate that it took reasonable steps to investigate the Whois inaccuracy claims. DreamHost’s failure to provide documentation demonstrating the reasonable steps it took to investigate and correct the alleged Whois inaccuracy is a breach of Section 3.7.8 of the RAA.
Weirdly, senect.com has been under private registration at DreamHost since the start of 2012.
ICANN seems to be asking the registrar to investigate itself in this case.
DreamHost offers private registration to its customers for free. It populates the Whois with proxy contact information and the registrant name “A Happy DreamHost Customer”.
DomainTools associates “A Happy DreamHost Customer” with over 710,000 domain names.
As an accredited registrar, DreamHost had over 822,000 gTLD domain names at the last count. According to its web site, it has over 400,000 customers.
The breach notice also demands the company immediately start including the real contact information for its privacy/proxy customers in its data escrow deposits.
ICANN has given the company until November 21 to resolve a laundry list of alleged RAA breaches, or risk losing its accreditation.
Over 800,000 domain names have been suspended since the beginning of the year as a result of Whois email verification rules in the new ICANN Registrar Accreditation Agreement.
That’s according to the Registrars Stakeholder Group, which collected suspension data from registrars representing about 75% of all registered gTLD domain names.
The actual number of suspended domains could be closer to a million.
The 2013 RAA requires registrars to verify the email addresses listed in their customers’ Whois records. If they don’t receive the verification, they have to suspend the domain.
The RrSG told the ICANN board in March that these checks were doing more harm than good and today Tucows CEO Elliot Noss presented, as promised, data to back up the claim.
“There have been over 800,000 domains suspended,” Noss said. “We have stories of healthcare sites that have gone down, community groups whose sites have gone down.”
“I think we can safely say millions of internet users,” he said. “Those are real people just trying to use the internet. They are our great unrepresented core constituency.”
The RrSG wants to see contrasting data from law enforcement agencies and governments — which pushed hard for Whois verification — showing that the RAA requirement has had a demonstrable benefit.
Registrars asked at the Singapore meeting in March that law enforcement agencies (LEA) be put on notice that they can’t ask for more Whois controls until they’ve provided such data and ICANN CEO Fadi Chehade said “It shall be done by London.”
Noss implied that the majority of the 800,000 suspended names belong to innocent registrants, such as those who had simply changed email addresses since registering their names.
“What was a lovely political win that we said time and time again in discussion after discussion was impractical and would provide no benefit, has demonstrably has created harm,” Noss said.
He was received with cautious support by ICANN board members.
Chair Steve Crocker wonder aloud how many of the 800,000 suspended domains are owned by bad guys, and he noted that LEA don’t appear to gather data in the way that the registrars are demanding.
“We were subjected, all of us, to heavy-duty pressure from the law enforcement community over a long period of time. We finally said, ‘Okay, we hear you and we’ll help you get this stuff implemented,'”, he added. “That creates an obligation as far as I’m concerned on their part.”
“We’re in a — at least from a moral position — in a strong position to say, ‘You must help us understand this. Otherwise, you’re not doing your part of the job'”, he said.
Chehade also seemed to support the registrars’ position that LEA needs to justify its demands and offered to take their data and concerns to the LEA and the Governmental Advisory Committee.
“They put restrictions on us that are causing harm, according to these numbers,” he said. “Let’s take this back at them and say, hey, you ask for all these things, this is what happened.”
“If you can’t tell me what good this has done, be aware not to come back and ask for more,” he said. “I’m with you on this 100%. I’m saying let’s use the great findings you seem to have a found and well-package them in a case and I will be your advocate.”
Director Mike Silber also spoke in support of the RrSG’s position.
“My view is if what you are saying is correct, the LEA’s have blown their credibility,” he said. “They’re going to have to do a lot of work before we impose similar disproportional requirements on actors that are not proven to be bad actors.”
So what does this all mean for registrants?
I don’t think there’s any ongoing process right now to get the Whois verification requirements overturned — that would require a renegotiation of the RAA — but it does seem to mean demands from governments and police are going to have to be much more substantiated in future.
Noss attempted to link the problem to the recommendations of the Whois Expert Working Group (EWG), which propose a completely revamped, centralized Whois system with much more verification and not much to benefit registrants.
To paraphrase: if email verification causes so much harm, what harms could be caused by the EWG proposal?
The EWG was not stuffed with LEA or governments, however, so it couldn’t really be characterized as another set of unreasonable demands from the same entities.
An ICANN working group has come up with a proposal to completely replace the current Whois system for all gTLDs.
Outlined in 180 recommendations spread over 166 pages (pdf), it’s designed to settle controversies over Whois that have raged for 15 years or more, in one fell swoop.
But it’s a sprawling, I’d say confusing, mess that could turn domain name registration and the process of figuring out who owns a domain name into an unnecessarily bureaucratic pain in the rear.
That’s if the proposal is ever accepted by the ICANN community, which, while it’s early days, seems like a challenge.
It’s a complex document, which basically proposes rebuilding Whois from the ground up based on ideas first explored by George Orwell, Franz Kafka and Douglas Adams.
Having read it, I’ll do my best in this post to explain what the proposed Registration Data Service seems to entail and why I think it seems like a lot of hard work for very little benefit.
I note in advance as a matter of disclosure that the RDS as proposed would very possibly disenfranchise me professionally, making it harder for me to do my job. I explain why later in this post.
I also apologize in advance for, and will correct if notified of, any errors. It’s taken me a week from its publication to read and digest the proposal and I’m still not sure it’s all sunk in.
RDS would be a centralized Whois database covering all domains in all gTLDs, new and old, operated by a single entity.
What’s in an RDS record?
Under the hood, RDS records wouldn’t look a heck of a lot different than Whois records look today, in terms of what data they store.
There would be some new optional elements, such as social media user names, but otherwise it’s pretty much the same data as we’re used to seeing in Whois records today.
The big difference is which of these elements would be visible by default to an anonymous internet user doing a regular Whois look-up somewhere.
Some fields would be “public” and some would be “gated” or hidden. Some fields would always be public and some could be toggled between public and gated by the registrant.
Gated fields would not be visible to people doing normal Whois look-ups. To see gated data, you’d need to be accredited to a certain role (cop, trademark owner, etc) and have an RDS account.
By default, much of the data about the “registrant” — including their name, physical address, country, and phone number — would be gated.
No, you’re not reading that wrong — the name of the registrant would be hidden from regular Whois users by default. Their email address, however, would be always be public.
There would also be up to six “Purpose Based Contacts” — an Admin Contact, a Legal Contact, a Technical Contact, an Abuse Contact, a Privacy/Proxy Contact and a Business Contact.
So, for example, a registrant could specify his registrar as his technical PBC and his lawyer as his legal PBC.
The admin, legal, technical and abuse contacts would be mandatory, and would default to the registrant’s own personal contact info.
A newly registered domain would not be activated in the DNS until the mandatory PBCs had been provided.
Each of these four mandatory PBCs would have different levels of disclosure for each data element.
For example, the Admin PBC would be able to hide their mailing address and phone number (both public by default) but not their name, email address or country.
The Legal PBC would not be able to opt out of having their mailing address disclosed, but the Technical and Abuse PBCs would be able to opt out of disclosing pretty much everything including their own name.
Those are just examples. Several tables starting on page 49 of the report (pdf) give all the details about which data fields would be disclosed and which could be hidden.
I think it’s expected by the EWG that most registrants would just accept the defaults and publish the same data in each PBC, in much the same way as they do today.
“This PBC approach preserves simplicity for Registrants with basic contact needs and offers additional granularity for Registrants with more extensive contact needs,” the EWG says.
Who gets the see the hidden stuff?
In order to see the hidden or “gated” elements, you’d have to be an accredited user of the centralized RDS system.
The level of access you got to the hidden data would depend on the role assigned to your RDS account.
The name of the registrant, for example, would be available to anyone with an RDS account.
If you wanted access to the registrant’s mailing address or phone number, you’d need an RDS account that accredited you for one or more of seven defined purposes:
- Domain Name Control (ie, the registrant herself)
- Domain Name Certification (ie SSL Certificate Authorities)
- Business Domain Name Purchase/Sale (anyone who says they might be interested in buying the domain in question)
- Academic/Public Interest DNS Research
- Legal Actions (eg lawyers investigating fraud or trademark infringement)
- Regulatory/Contractual Enforcement (could be ICANN-related, such as UDRP, or unrelated stuff like tax investigations)
- Criminal Investigation/DNS Abuse Mitigation
Hopefully this all makes sense so far, but it gets more complicated.
Beware of the leopard!
In today’s gTLD environment, Whois records are either stored with the registry or the registrar. You can do Whois lookups on the registrar/y’s site, or via a third-party commercial service.
As a registrant, you need only interact with your registrar. As a Whois user, you don’t need to sign up for an account anywhere, unless you want value-added services from a company such as DomainTools.
Under RDS, a whole lot of other entities start to come into play.
First, there’s RDS itself — a centralized Whois replacement.
It’s basically two databases. One contains contact details, each record containing a unique Contact ID identifier. The other database maps Contact IDs to the PBCs for each gTLD domain name.
It’s unclear who’d manage this service, but it looks like IBM is probably gunning for the contract.
Second, there would be Validators.
A Validator’s job would be to collect and validate contact information from registrants and PBCs.
While registrars and registries could also act as Validators — and the EWG envisages most registrars becoming Validators — this is essentially a new entity/role in the domain name ecosystem.
Third and Fourth, we’ve got newly created Accrediting Bodies and Accreditation Operators.
These entities would be responsible for accrediting users of the RDS system (that is, people who want to do a simple goddamn Whois lookup).
The EWG explains that an Accrediting Body “establishes membership rules, terms of service, and application and enforcement processes, etc., for a given RDS User community.”
An Accreditation Operator would “create and manage RDS User accounts, issue RDS access credentials, authenticate RDS access requests, and provide first-level abuse handling”.
Because it’s not complicated enough already, each industry (lawyers, academics, police, etc) would have their own different combination of Accrediting Bodies and Accreditation Operators.
Who benefits from all this?
The reason the EWG was set up in the first place was to try to resolve the conflict between those who think Whois accuracy should be more strictly enforced (generally law enforcement and IP owners) and those who think there should be greater registrant privacy (generally civil society types).
In the middle you’ve got the registries and registrars, who are generally resistant to anything that adds friction to their shopping carts or causes even moderate implementation costs.
The debate has been raging for years, and the EWG was told to:
1) define the purpose of collecting and maintaining gTLD registration data, and consider how to safeguard the data, and 2) provide a proposed model for managing gTLD directory services that addresses related data accuracy and access issues, while taking into account safeguards for protecting data.
So the EWG proposal could be seen as successful if a) privacy advocates are happy and b) trademark lawyers and the FBI are happy, c) registrars/ries are happy and d) Whois users are happy.
Are the privacy dudes happy?
No, they’re not.
The EWG only had one full-on privacy advocate: Stephanie Perrin, who’s a bit of a big deal when it comes to data privacy in Canada, having held senior privacy roles in public and private sectors there.
Perrin isn’t happy. Perrin thinks the RDS proposal as it stands won’t protect regular registrants’ privacy.
She wrote a Dissenting Report that seems to have been intended as an addendum to the EWG’s official report, but it was not published by the EWG or ICANN. The EWG report makes only a vague, fleeting reference, in a footnote, to the fact that the was any dissent at all.
Milton Mueller at the Internet Governance Project got his hands on it regardless and put it out there earlier this week.
Perrin disagrees with the recommendation (outlined above) that each domain name must have a Legal Contact (or Legal PBC) who is not permitted to hide their name and mailing address from public view.
She argues, quite reasonably I think, that regular registrants don’t have lawyers they can outsource this function to, which means their own name and mailing address will comprise their publicly visible Legal PBC.
This basically voids any privacy protection they’d get from having these details “gated” in the “registrant” record of the RDS. Perrin wrote:
the purpose of the gate is to screen out bad actors from harassing innocent registrants, deter identity theft, and ensure that only legitimate complaints arrive directly at the door of the registrants. It is also to protect the ability of registrants to express themselves anonymously. Placing all contact data outside the gate defeats certain aspects of having a gate in the first place.
The EWG report envisages the use of privacy/proxy services for people who don’t want their sensitive data published publicly.
But we already have privacy/proxy services today, so I’m unclear what benefit RDS brings to the table in terms of privacy protection.
It’s also worth noting that there are no circumstances under which a registrant’s email address is protected, not even from anonymous RDS queries. So there’s no question of RDS stopping Whois-based spam.
Are the trademark dudes going to be happy?
I don’t know. They do seem to be getting a better deal out of the recommendations than the other side (there were at least three intellectual property advocates on the EWG) but if you’re in the IP community the report still leaves much to be desired.
The RDS proposal would create a great big centralized repository of domain registrant information, which would probably be located in a friendly jurisdiction such as the US.
That would make tracking down miscreants a bit easier than in today’s distributed Whois environment.
RDS would also include a WhoWas service, so users can see who has historically owned domain names, and a Reverse Query service, so that users can pull up a list of all the other domains that share the same contact field(s).
Both services (commercially available via the likes of DomainTools already) would prove valuable when collating data for a UDRP complaint or cybersquatting lawsuit.
But it’s important to note that while the EWG report says all contact information should be validated, it stops short of saying that it should be authenticated.
That’s a big difference. Validation would reveal whether a mailing address actually exists, but not whether the registrant actually lives there.
You’d need authentication — something law enforcement and IP interests have been pushing for but do not seem to have received with the EWG proposal — for that.
The EWG suggests that giving registrants more control over which bits of their data are public will discourage them from providing phony contact information for Whois/RDS.
The RDS proposes a lot more carrot than stick on this count.
But if Perrin is correct that it’s a false comfort (given that your name and address will be published as Legal PBC anyway) then wouldn’t a registrant be just as motivated to call themselves Daffy Duck, or use a proxy/privacy service, as they are today?
Are the registrar dudes going to be happy?
If the EWG’s recommendations become a reality registrars could get increased friction in their sales path, depending on how disruptive it is to create a “Contact ID” and populate all the different PBCs.
I think it’s certainly going to increase demand on support channels, as customers try to figure out the new regime.
Remember, the simple requirement to click on a link in an email is causing registrants and registrars all kinds of bother, including suspended domains, under recently introduced rules.
And there’s obviously going to be a bunch of (potentially costly) up-front implementation work registrars will need to do to hook themselves into RDS and the other new entities the system relies on.
I doubt the registrars are going to wholeheartedly embrace the proposal en masse, in other words.
Is Kevin Murphy happy?
No, I’m not happy.
It bugs me, personally, that the EWG completely ignored the needs of the media in its report. It strikes me as a bit of a slap in the face.
The “media” and “bloggers” (I’m definitely in one of those categories) would be given the same rights to gated RDS data as the “general public”, under the EWG proposal.
In other words, no special privileges and no ability to access the registrant name and address fields of an RDS record.
That can’t be cool, can it?
Murphy, brah, why you gotta cuss in your headline?
Good question. I do use swearwords on DI occasionally, but only to annoy people who don’t like them, and usually only in posts dated April 1 or in stories that seem to deserve it.
This post is dated June 13.
I think I’ve established that the EWG’s proposal as it stands today is a pretty big overhaul of the current system and that it’s not immediately obvious how the benefits to all sides warrant the massive effort that will have to be undertaken to get RDS to replace Whois.
But the clusterfuckery is going to begin not with the implementation of the proposal, but with the attempt to pass it through the ICANN process.
The proposal has to pass through the ICANN community before becoming a reality.
The Expert Working Group has no power under the ICANN bylaws.
It was created by Chehade while he was still relatively new to the CEO’s job and did not yet appreciate how seriously community members take their established procedures for creating policy.
I think it was a pretty decent idea — getting a bunch of people in a room and persuading them to think outside the box, in an effort to find radical solutions to a a long-stagnant debate.
But that doesn’t change the fact that the EWG’s proposals don’t become law until they’ve been subject to the Generic Names Supporting Organization’s lengthy Policy Development Process.
Some GNSO members were not happy when the EWG was first announced — they thought their sovereignty was being usurped by the uppity new CEO — and they’re probably not going to be happy about some of the language the EWG has chosen to use in its final report.
The EWG said:
The proposed RDS, while not perfect, reflects carefully crafted and balanced compromises with interdependent elements that should not be separated.
The RDS should be adopted as a whole. Adopting some but not all of the design principles recommended herein undermines benefits for the entire ecosystem.
It’s actually quite an audacious turn of phrase for a working group with no actual authority under ICANN bylaws.
It sounds a bit like “take it or leave it”.
But there’s no chance whatsoever of the report being adopted wholesale.
It’s going into the GNSO process, where the same vested interests (IP, LEA, registry, registrar, civil society) that have kept the debate stagnant for the duration of ICANN’s existence will continue to try (and probably fail) to come to an agreement about how Whois should evolve.