Monetizing disorder
Remember Sitefinder? I still have my notes from a session four years ago that was led by Steve Crocker. VeriSign (the registry operator for .com and .net) introduced a change in the way .com/.net operated. A misstyped domain would return the address of a host that was one of VeriSign’s own instead of an error message. There was considerable pushback from the community. VeriSign quickly suspended the service.
The generic name for what VeriSign was doing was a “wildcard.” Before SiteFinder was implemented, if you typed the name of a nonexistent .com domain (say, scrrawford.com hadn’t been registered), you’d receive back a “name not found” answer. After SiteFinder, if you typed a nonexistent domain you’d receive back a referral to a server controlled by VeriSign. VeriSign claimed that they weren’t retaining or collecting any data in connection with this service, and pointed out that other registries (mostly country code top level domains) also used wildcards. (Internet Explorer does something similar. If you misstype a domain, you’ll get a page of suggestions. There’s money to be made from confusion.)
What was the point of SiteFinder? To send confused users to a potentially “useful” page — a page that would have (eventually) been populated with links to advertisements and other relevant materials. In the end, had the service stayed in place, VeriSign would have been able to monetize this confusion - making money from the pay-per-click advertisers that populated the potentially “useful” page.
Why was there uproar? Because not all domain record requests are made in connection with hunting for web pages - spam filtering email services got hung up on the wildcard service. There’s a commonly-used spam filtering rule that rejects messages with envelope sender addresses that do not resolve, and other protocol (non-web) requests didn’t react well to the redirection. Sitefinder didn’t break the internet, but it did cause a number of small problems for applications that rely on domain name queries. The DNS infrastructure is very sensitive to change, and SiteFinder was viewed as a “system expectation violation.” The plumbing was getting uppity, and that wasn’t appreciated.
In recent years, the market has filled with “parked” pages. If you misstype a domain name, chances are you’ll go to a page that someone has registered and has populated with a bunch of (somewhat) relevant links. Again, the aim is advertising revenue.
Over the last few days, Verizon has reportedly been doing something similar to its DSL and FiOS subscribers. According to ConsumerAffairs.com:
“When users misstype a web site address, they get redirected to Verizon’s own search engine page — even if they don’t have Verizon’s search page set as their default.”
One more time, the goal here is to make money from confusion. Verizon doesn’t want to be treated like plumbing - they’re anxious to make money from their users’ navigation failures. In order to get away from the Verizon default page, you’ll need to reset your DNS settings. Not something that the average user wants to cope with.
You can imagine this as a layered fight over confusion. First the DNS registry itself - VeriSign - tried to make money from misstyped domains. Then DNS registrants - the people who build parked pages - tried to get in on the value chain. Meanwhile, edge applications (like the browser) were doing the same thing. Now the ISP is inserting itself, trying to override the parked page and browser efforts.
Is this a violation of net neutrality? It certainly is a “system expectation violation.” We don’t expect ISPs to be filtering our web browsing requests and inserting themselves into the conversation. There’s some concern that the ISP could be doing more than presenting a response page, as we’ve seen from the Comcast flap. Although in a larger sense it’s just what all the other players in the chain want to do - make money from disorder - we want to avoid having the plumbing, the transport, do this without a user’s acquiescence.
I have a feeling that Verizon’s actions here will cause as much upset as VeriSign’s.
(Say “Verizon/VeriSign” quickly a few times.)
ICANN Thursday
Today - the public forum. Not enough time for public comments. Many many reports.
This afternoon - endless board huddling. All discussions are over before dinner, which doesn’t always happen.
One more meeting, and then a break for the evening.
Tomorrow: back to telecom policy again.
ICANN Wednesday
This morning, the GNSO Council voted neither to adopt OPOC nor to sunset the current whois clauses (all jargon unpacked in yesterday’s post). AP story here, WIRED story here. The report that the board gets about this result should be useful in making clear exactly what the obstacles were to changing the status quo.
This afternoon was taken up with board committee meetings, a meeting about the accountability/transparency of ICANN, and meetings about the strategic plan.
Tomorrow - more meetings. The public forum in the morning - webcast - and then the board huddles all afternoon and evening.
At some point, someday, this blog will do more than report meetings, but for right now that’s all I can manage.
ICANN Tuesday
Today is Constituency/Supporting Organization day. The board is trying something new — splitting up and going to individual meetings for a long period, rather than trooping around as a group to a series of hour-long meetings. I went to the gTLD Registry Constituency meeting. Later we’ll get back together and compare notes before meeting in a public session with the Governmental Advisory Committee.
The big issues for this meeting include new gTLDs (as I mentioned yesterday), domain tasting, internationalized domain names, whois, and structural reforms (of the GNSO, of the Nominating Committee).
There’s a public comment period going on until the end of today on whois. This discussion has been going on for seven years or so within the ICANN context. The basic issue: should it be a condition of registering a domain name that the registrant’s personal information be made publicly available? Tomorrow, the GNSO Council will vote on three motions that have been suggested with respect to the whois policy development process. This document (p.9) sets forth the three motions.
The first motion would adopt a proposal to allow individual registrants to designate an Operational Point of Contact whose information would appear publicly. Requestors needing the registrant’s personal information would be required to give a good reason (as defined under applicable national law) to get that information, but would be able to reach the registrant through the OPOC.
The second motion would initiate more studies.
The third motion would say “you have until the end of 2008 to come to an agreement on how to both implement whois AND allow for individual privacy (and compliance with applicable laws). Otherwise, because the whois clauses in the contracts aren’t supported by consensus, they will be removed from the contract.”
Many of the participants in this process are simply worn out. The question is whether “we’ve had enough, and we can’t agree” should lead to (1) maintenance of the status quo, or (2) removal of the clause requiring public presentation of individual data.
There are many comments here. Add your own before the day ends.
ICANN Monday
Today is the first day of the 30th ICANN meeting. I’ve been here since late Friday - I wanted to go to the all-day GNSO Council meeting on Saturday about whois, new TLDs, and other issues. Yesterday, Sunday, the GNSO continued to meet, the board committees met, and the board had an all-afternoon workshop plus a dinner.
This afternoon there’s a key discussion about the policies to govern the selection of new TLDs. I’ve been hoping for a standardized, objective process for new TLDs for a very long time, and my hope is that we’ll move forward on this. The GNSO Council has adopted a set of recommendations, and this workshop was designed to focus on them.
We’ve reached the most contentious area of recommendations — the processes for addressing worries about particular strings being chosen as new TLDs.
Recommendation 3:
Strings must not infringe the existing legal rights of others that are recognized or under generally accepted and internationally recognized principles of law. Examples of these legal rights that are internationally recognized include, but are not limited to, rights defined in the Paris Convention for the Protection of Industry Property (in particular trademark rights), the Universal Declaration of Human Rights (UDHR) and the International Covenant on Civil and Political Rights (ICCPR) (in particular freedom of expression rights).
Recommendation 6:
Strings must not be contrary to generally accepted legal norms relating to morality and public order that are recognized under international principles of law.
Examples of such principles of law include, but are not limited to, the Universal Declaration of Human Rights (UDHR), the International Covenant on Civil and Political Rights (ICCPR), the Convention on the Elimination of All Forms of Discrimination Against Women (CEDAW) and the International Convention on the Elimination of All Forms of Racial Discrimination, intellectual property treaties administered by the World Intellectual Property Organisation (WIPO) and the WTO Agreement on Trade-Related Aspects of Intellectual Property (TRIPS).
This is definitely tricky. International law doesn’t have anything to say about “morality” when it comes to gTLDs. And, of course, morality varies by country….and city.
Recommendation 20:
An application will be rejected if an expert panel determines that there is substantial opposition to it from a significant portion of the community to which the string may be explicitly or implicitly targeted.
Also very tricky. “Implicitly targeting” is defined as being in the mind of the objector - that objector can assume that there’s targeting “or that the objector believes there may be confusion by users over its intended use.” Whoof.
This is a six-hour workshop, and we’re moving into the fourth hour. Lots of challenges ahead. How will/can these policies be implemented? My top priority is making sure we have a standardized, obvious, objective process.
If you’re interested in this subject, read this for more information. All comments very welcome.