This is how AppDetex works
A small brand-protection registrar with a big friend caused quite a stir at ICANN 63 here in Barcelona this week, after accusing registrars for the second time of shirking their duties to disclose private Whois data to trademark owners.
AppDetex, which has close ties to Facebook, has sent something like 9,000 Whois requests to registrars over the last several months, then complained to ICANN last week that it only got a 3% response rate.
Registrars cried foul, saying that the company’s requests are too vague to action and sometimes seem farcical, suggesting an indiscriminate, automated system almost designed to be overly burdensome to them.
In chats with DI this week, AppDetex CEO Faisal Shah, general counsel Ben Milam and consultant Susan Kawaguchi claimed that the system is nowhere near as spammy as registrars think, then showed me a demo of their Whois Requester product that certainly seemed to support that claim.
First off, Whois Requester appears to be only partially automated.
Tucows had noted in a letter to ICANN that it had received requests related to domains including lincolnstainedglass.com and grifflnstafford.com, which contain strings that look a bit like the “Insta” trademark but are clearly not cybersquatting.
“That no human reviewed these domains was obvious, as the above examples are not isolated,” Tucows CEO Elliot Noss wrote.
“It is abundantly clear to us that the requests we received were generated by an automated system,” Blacknight CEO Michele Neylon, who said he had received similarly odd requests, wrote in his own letter.
But, according to AppDetex, these assumptions are not correct.
Only part of its service is automated, they said. Humans — either customers or AppDetex in-house “brand analysts” — were involved in sending out all the Whois requests generated via its system.
AppDetex itself does not generate the lists of domains of concern for its clients, they said. That’s done separately, using unrelated tools, by the clients themselves.
It’s possible these could be generated from zone files, watch services, abuse reports or something else. The usage of the domain, not just its similarity to the trademark in question, would also play a role.
Facebook, for example, could generate its own list of domains that contain strings matching, partially matching, or homographically similar to its trademarks, then manually input those domains into the AppDetex tool.
The product features the ability to upload lists of domains in bulk in a CSV file, but Kawaguchi told me this feature has never been used.
Once a domain has been input to main Whois Requester web form, a port 43 Whois lookup is automatically carried out in the background and the form is populated with data such as registrar name, Whois server, IANA number and abuse email address.
At this point, human intervention appears to be required to visually confirm whether the Whois result has been redacted or not. This might require also going to the registrar’s web-based Whois, as some registrars return different results over port 43 compared to their web sites.
If a redacted record is returned, users can then select the trademark at issue from a drop-down (Whois Requestor stores its’ customers trademark information) and select a “purpose” from a different drop-down.
The “purposes” could include things like “trademark investigation” or “phishing investigation”. Each generates a different piece of pre-written text to be used in the template Whois request.
Users can then choose to generate, manually approve, and send off the Whois request to the relevant registrar abuse address. The request may have a “form of authorization” attached — a legal statement that AppDetex is authorized to ask for the data on behalf of its client.
Replies from registrars are sent to an AppDetex email address and fed into a workflow tool that looks a bit like an email inbox.
As the demo I saw was on the live Whois Requester site with a dummy account, I did not get a view into what happens after the initial request has been sent.
Registrars have complained that AppDetex does not reply to their responses to these initial requests, which is a key reason they believe them frivolous.
Shah and Milam told me that over the last several months, if a registrar reply has included a request for additional information, the Whois Requester system has been updated with a new template for that registrar, and the request resent.
This, they said, may account for duplicate requests registrars have been experiencing, though two registrars I put this to dispute whether it fits with what they’ve been seeing.
The fact that human review is required before requests are sent out “just makes it worse”, they also said.
If you find this post or this blog useful or interestjng, please support Domain Incite, the independent source of news, analysis and opinion for the domain name industry and ICANN community.
Most registrars are providing much more information thru WebWHOIS (with captcha) than thru port-43, so by using port-43 the system is setting itself up for failure.
Maybe it depends on how thorough the staff is in reviewing the list of domains? With the amount of complaints one has to wonder how many brand analysts work there.