The Closing Window: A Historical Analysis of Domain Tasting (original) (raw)

I wrote this history and analysis of domain tasting for the ICANN Business Constituency membership. It’s by no means perfect but I thought I’d share it with those who would like a bit more color on the subject.

History

Present day “Domain Tasting” has its roots in 2001 and 2002 when a small group of ambitious domain registrants persuaded two registrars (IAregistry.com [now Spirit Telecom]) and DotRegistrar.com) to allow them to register large blocks of domain names for the purpose of establishing which names garnered type-in traffic and then subsequently deleting those which didn’t. These registrants were motivated to locate generic, defensible domain names with resonance; names which were compelling and obvious that people would use to navigate the web and to then monetize the traffic those names generated through paid-search advertising. It is important to note that old traffic from a former web site (link popularity) or traffic from a company trademark is entirely different from organic type-in traffic which comes to “generic names” for the keyword weight or gravity of the name alone. Names such as spacepictures.com or pandabears.com contain generic intent type-in traffic which could be sold to the first generation paid-search monetization companies of the day (Overture.com [now Yahoo.com] and AppliedSemantics.com [now Google.com Adsense for domains]).

The time-frame was immediately following the “.com bust” and stock crash of 2000/2001 when domain names (and anything Internet related for that matter) were perceived to be worthless. Many long-time individual registrants of the day were letting what would ultimately be recognized as valuable, generic domain names expire for non-payment under the mistaken assumption that the Internet was about to disappear with the stock market crash. Early domain tasters simply couldn’t afford to buy the millions of then-available generic names so these enterprising registrants utilized the 5 day grace period to register blocks of 10 or 20 thousand names from the available pool every few days, identify those names which contained type-in traffic and deleted those which weren’t generic and had no traffic. During that time, the registry root-zone files only updated every 12 hours and so a registrant would have to register names, make sure they all made the registry cutoff window to “go live on the web” and then watch the names for one or two days (there was less overall traffic on the net at the time so registrants needed a longer sample window to keep names live in order to identify the good ones). All-in-all from registration to deletion, registrants kept names 3 days or longer before jettisoning the cast-offs.

VeriSign knew about this activity in 2001/2002 and for several reasons adopted the view that the extra 20,000 or 50,000 daily registrations and subsequent deletions were abusive. VeriSign worked over the course of several months to stop the registrars fulfilling those transactions, going so far as to threaten them with litigation. The registrars begrudgingly acquiesced to the registry’s demands and the ‘tasting’ (called “Batch Testing” or “Autodelete Registrations” at the time) window closed.

Evolution

In late 2004, VeriSign changed the way the root-zone files updated allowing a domain name to be registered and go live in the zone nearly immediately. Again, those enterprising registrants (and some new ones) realized they could now register many domain names quickly, keep the names for a shorter period and subsequently delete them without getting the attention of the registry. Several other dynamics helped fuel this second wave (current wave) of the practice now generally known as “domain tasting”: The paid search market in 2004 and beyond was much more mature than in 2001, so higher advertising pay-rates drove down “keep” thresholds. Rather than keeping .01% of names, registrants were keeping significantly more. Higher type-in visitor (direct navigation) traffic counts increased the profit potential and allowed more registrants and registrars to participate in the game; significantly driving up registration volumes (kept registrations) at the VeriSign (and also PIR [.org]) registry in 2005 and 2006. This surprise increase in registration volume and profits caused the registries to tacitly support domain tasting this second-time around.

The problem with this evolution is that there are a finite number of generic defensible domain names that garner type-in traffic. As registrants threw wider and wider drift-nets and registered larger and larger blocks of search queries as domain names, many owners of corporate brands began to find variants of their brand-name(s) taken by tasting companies. No longer in the salad days of 2001, 2006 brought phishing con artists and the trademark monitoring services to guard against them. These new trademark monitoring companies would wrongly identify the tasters as ‘phishers’. By mid 2005 the monitoring companies saw scores of variants of their client’s brand registered by large-scale commercial registrants and wrongly assumed these parties were out to begin a concerted phishing attack. The truth was that large scale commercial registrants were trying to harvest type-in traffic from domain names by registering lists of popular search engine queries with spaces removed and .com or .net extensions appended. Unfortunately many search engine queries are for “trademark intent” search strings. Commercial registrants, unable to groom the batch lists for every trademark, unwittingly registered scores of trademarks for a day or two before deleting them. Certainly disquieting to trademark holders such as the banks, Ebay, paypal.

Criticism

In life there are good apples and there are bad apples. Some registrants found it expedient to keep ‘any’ domain name that got traffic without grooming the “keep-lists” for potential trademark intent. This “catchall typosquatting” by indifferent registrants is the action that has ultimately brought the bulk of criticism against “Domain Tasting”. Additional complaints come from large retail registrars such as GoDaddy who are concerned that their new retail registration business is stymied by “Drop Registrars” who organize in groups and stand at the expiring domain-name spigot. These groups arrange a phalanx of ICANN accredited registrars in the expiring name pool to snap-up domain names the moment they expire; only to “Taste” or “Kite” the entire volume of daily expiring domain names, retaining those which are generally accepted to be valuable or retain traffic from a previous site (link-popularity kiting). The concern is that the good (interesting, generic) names are getting creamed off the top without an opportunity for outsiders to compete for them.

Irony

The irony is that names which are kept, only get-kept because they have marginal traffic. If these names were not registered, the traffic would still exist, but rather than flow to the registrant of the domain name, the traffic would default to Microsoft’s Internet Explorer Browser or Google’s Firefox toolbar or some other search browser helper. Every day Microsoft and Google make money selling error traffic from inactive URLs that come in via browser application. It’s ironic that the companies which vilify domain registrants through typo identifying software (i.e. Microsoft’s strider tool) have the most to gain if the ‘nefarious domain registrants’ didn’t exist. It is also ironic that the majority of the revenue from most tasting portfolios flows to VeriSign and ICANN. The average tasting name generates pennies or a few dollars above registration price each year. The lion’s share of the gross revenue generated by these registrant whipping boys flows right into the hands of VeriSign and ICANN in the form of renewal and registration fees (hence the lack of urgency to slow the money train down).

Tasting Conclusion

Tasting has always come down to the intent of the party doing it. One person could deliberately register hundreds of thousands of trademark names for the purpose of catchall typosquatting. But in the hands of a well-minded entrepreneur, opportunities to pan the river for flakes of gold dust still exist. It is conceivable that even at this late stage anyone can register 100,000 domain names and keep 20 or 100 generic defensible names with organic traffic that have somehow slipped between the raindrops undetected until now. It is also conceivable that those names will have a secondary market value in the thousands or tens of thousands of dollars. That “gold rush” prospecting dynamic is what gave birth to domain tasting in the first place and what (at least partly) continues to provide motivation for the practice today.

If one thinks of VeriSign as a giant open-pit gold mine. The Tasters bring their heavy equipment to churn tons of ore trying to create an ounce of gold. In the process there is environmental damage downstream. Security and Trademark monitoring companies guarding against phishing have to filter the noise of Tasters from the root zone file. Likewise, domain registrars and expiring name auction providers use the root zone file to monitor names leaving the zone and have to filter Tasters out. You have to bear in mind that the Trademark monitoring companies, registrars and auction providers are paying a stipend (if anything) for the privilege of accessing the root-zone file to monitor domain names. One could argue that it is the Tasters who have been “footing the bill” adding millions and millions of permanent “kept” registrations in 2004, 2005, 2006 (most of which will be renewed forever). VeriSign and ICANN have been getting a significant cash-flow stream from all these kept registrations that ultimately would never have come without the Tasters.

I think the ‘law of large numbers’ has kicked in. It is not possible to harvest traffic names forever. At some point the cost of registering 1,000,000, 3,000,000 or 5,000,000 domain names “each day” only to keep a diminishing handful of generics will outweigh the benefit to the registries and ICANN. The potential for wholesale trademark abuse and other ills created by allowing the practice to continue will just get too large as latecomers pile-on.

We are probably at that point now, but VeriSign will have to close the window, just as it did in 2002. Only it may be harder this time because they have permitted the practice for so long. The best solution would likely be for VeriSign to send an email to all registrars announcing a restocking fee effective next-week. Even a 2 cent restocking fee (trivial for real errors) would be enough to curb tasting dramatically.