Big twist as ICANN bans new gTLD auctions
ICANN is to ban new gTLD applicants from paying each other off if they apply for the same strings, removing a business model that saw tens of millions of dollars change hands in the 2012 application round.
But, in a twist, applicants will be able to submit second-choice strings along with their main application, allowing them to switch if they find themselves in contention.
While ICANN’s board of directors has yet to pass a resolution on private resolution in forthcoming application rounds, chair Tripti Sinha said in a letter to the GNSO Council (pdf) and blog post that there’s agreement on three principles.
“Private resolution of contention sets will not be permitted during the Next Round,” Sinha told the Council. The idea of permitting joint-venture resolution was also ruled out as impractical and open to gaming.
This of course means that where contention sets do occur, they’ll be resolved with a “last resort” auction where ICANN gets all the cash from the winning bidder.
Funds raised this way in the last round, along with a decade’s worth of investment interest, have been used to replenish ICANN’s reserve fund, to fund the current Grant Program, and may be shortly used to subsidize the Applicant Support Program.
Second, applicants will be able to submit at least one alternate string with their applications, allowing them to avoid a contention set and last resort auction.
This potentially makes the cost of acquiring a gTLD cheaper for the applicant while increasing the number of gTLDs that go live. ICANN might also have to issue fewer refunds for withdrawn applications.
ICANN thinks this measure might make gTLDs more affordable for less well-resourced applicants from the Global South, where ICANN is keen to diversify the industry, although the applicants may not get their first-choice strings.
Applicants would only be able to switch to an alternate string, which they will have to have pre-selected, if doing so would not create a new contention set or make the applicant join a different existing contention set.
They’d also only be able avoid a contention set of exact-match strings, and not sets subsequently created by the String Similarity Review or String Confusion Objection results.
So, to take an example from 2012, any of the seven .hotels applicants would have been able to switch to a second-choice string immediately after Reveal Day, but not after the similarity review placed them in contention with .hoteis.
The third point of agreement from the board is that the last resort auctions should keep the ascending-clock second-price method used for the 2012 round, deciding against lotteries or the Vickrey auction method.
The ascending clock method sees bids filed in rounds until all bidders but one had dropped out. The last applicant standing then pays ICANN the last price offered by the runner-up.
A Vickrey auction would have seen applicants submit their maximum bids at the time of application, not knowing who they were bidding against. Lotteries are legally problematic under California gambling law.
Sinha said the board intends to pass a resolution embodying these three principles “in the coming weeks”.
This is going to create some extra work for the GNSO, as ruling out joint ventures as a means to private resolution goes against community policy recommendations (and the board’s adoption of those recommendations).
The GNSO Council is set to discuss Sinha’s letter at its regular monthly meeting this Thursday.
I find this is genius from the ICANN text : “Ability to submit alternate strings: Applicants should be permitted to submit one or more alternate strings at the time of application submission. During a short window after reveal day, an applicant could then switch to an alternate string – if their initial string is identical to another applied-for string – without creating an existing or forming a new contention set.”
Well, this wouldn’t solve the problem which string a applicant prefer – this doesn’t prevent the community for ICANN harvest their money.
“The Board views the ability to submit an
alternate string at the time of application as a path to reduce the number of
contention sets…”
How did they model this? Or is this going to be another stellar example of non-data-based decisionmaking based on gut feelings?
If ten applicants are aiming for a word relevant to a particular industry/activity/interest, and choose an identical string, then it seems just as likely that several of them will have identical strings as their second choice – because the rank of preferable choices relevant to an industry/activity/interest are also obvious . So, starting from a contention set of ten applicants, if, say, six of them move to their “second choice” then you are most likely going to have at least one more contention set, if not more.
What models or simulations did the Board run to come up with the ridiculous idea that second choices among string contenders would not be likely to generate additional contention sets?
For example, let’s say that several entities with interests in the dairy industry apply for TLDs. What is the range of first and second choices among .dairy .milk or other alternatives. Obviously, if one is interested in dairy products generally, then something like .cheese is going to be under-inclusive, and the topline first and second choices are .dairy and .milk. Entities that would hypothetically switch to .leche .lait .milch. But if they were seeking a non-English registrant base, they wouldn’t likely apply for an English TLD in the first place.
Or, take the crypto crowd. If you have ten applicants for .crypto, what do you think the “second choices” are going to be in that bunch? It’s easily predictable.
Why not just adopt the year 2000 round method of having the board decide alternate strings on the fly, as they did by crippling the .air application to .aero, despite the common aviation industry branding convention of “[Word] Air”, merely because one board member didn’t like the idea of “selling air”.
The unexamined assumption of the Board is that TLD applicants want “a string”, any string at all, and that they choose their applied-for strings at random, such that an equally random “second choice” will move them out of contention.
The reality is that applicants want a string that is relevant to an underlying concept. If several choose the same string, then the range of “second choice” strings is constrained to strings that maintain as broad a possible relevance to the original concept, or as broad a subset as possible. The second choices, in other words, are more likely, not less, likely to generate additional collisions. Does anyone think that among ten .crypto applicants, there will not be multiple applicants whose second choice is .coin?
But this opens up an entirely different gaming strategy, which is for another time.
Secondly, by ruling out “joint ventures” then the initial strategy taken by Donuts makes sense. For each TLD application, form an individual corporate shell, which can itself be sold to another contender. There are second-order models in which the majority stake in the TLD applicant is itself another corporate entity, so that the owner entity can be sold without changing the ownership of the TLD applicant, and so on. The eventual rules will define the gaming necessary to defeat them.
I failed to note that if the idea is that one can switch to the “second choice” on the condition that it not collide with anything else, that’s not going to realistically happen very often, given the constraints on synonyms or suitable alternatives for a given concept in the first place.