April 4, 2003

Why Lotteries Are Better Than Auctions When Distributing New TLD Slots

This note is in response to sTLD Beauty Contests: An Analysis and Critique of the Proposed Criteria to Be Used in the Selection of New Sponsored TLDs by Karl M. Manheim & Lawrence B. Solum.  Other materials related to the issue of deploying new TLDs may be found on the authors' web page at http://gtld-auctions.net/

When I ran in the only open election ever held for an ICANN board seat my published platform set forth my preferred approach to the deployment of new TLDs.

In that platform I proposed that the the top level domain space be expanded not through the deployment of "names" but rather through the deployment of "slots".  The difference is subtle.  When I say deployment by "names" I mean that the character string that will be the actual domain name label for that new TLD is made part of the selection process.  When I say deployment by "slots" I mean that the character string is utterly irrelevant except to know whether that string is already in use.  A "slot" is a right or privilege to have a character string of one's choosing inserted into the DNS root zone (along with various NS records pointing to a suite of name servers.)

In other words, I prefer that we deal with "slots" rather than named strings, that we focus on the capabilities of the holder of the slot rather than on the semantics of the string.  There are courts a-plenty to handle the real or imagined slights of those who claim to have rights of one kind or another over a given string.  If we deal with "names" we are tacitly getting involved in those disputes, if we deal with "slots" we are explicitly saying that we chose not to be a forum for fights over character strings.

So I pose the question of new TLDs not as a question of names but rather of slots - who gets to have a slot in the root zone?

As Ross Rader points out, the most fundamental question is whether the applicant has the technical capacity to run a suite of name servers in accord with the applicable technical standards.  I emphasize the word "technical".  It is of no importance to me whether the applicant has any business sense, nor do I care about the applicant's motivations.  And, as mention in my note Thoughts on whois and privacy  I do not necessarily consider the applicant's ability to publish whois information to the world to be necessarily part of that technical competence.

Business acumen and financial strength are not among the criteria that ought to be used to chose between applicants.  Yes, we have a world in which the customers of a TLD build their brands on that TLD.  But I prefer that the underwriting of that TLD's continued operation be a matter between the customers and the TLD operator and not a regulatory matter imposed ex cathedra by ICANN or any other regulatory body.  (I would go so far as to say that perhaps ICANN ought to define good data preservation practices, such as registration data escrow formats.  I won't comment further here on ICANN's vanished effort to define escrow formats and impose them.)

So, how do we chose who gets to have a slot?

There are technical limits to the size of the DNS root zone - there simply isn't enough room to give a name to everyone.  An allocation mechanism is necessary.  And perhaps there also needs to be a garbage collection mechanism as well to reap dead allocations.

The paper sTLD Beauty Contests: An Analysis and Critique of the Proposed Criteria to Be Used in the Selection of New Sponsored TLDs proposes auctions.

Auctions are a means that guarantees the prize to the qualified applicant willing to pay the most money.  In practical terms this means that the wealthy will inherit the Internet, or at least the DNS top level domains.

For many, that is an acceptable outcome.

However, I feel that there are social values other than cubic money.  And those who live by those other values ought to have a chance to obtain TLD slots.

This is why I feel that allocation ought to be by a lottery system.  Everybody who buys a "ticket" has a chance.  The wealthy can, by buying more tickets, improve their chances to an arbitrary degree.  But no matter how many they buy, there is still is a chance that the small guy might win.

If we have lotteries for a non-trivial number of slots we can expect that at least a few will be won by the less well healed applicants.

I am, of course, ignoring the legal obstacles that face things called "lotteries".  There is no doubt that what I suggest is a kind of gambling.  But in the great scheme of things, that's a pretty small blemish on what otherwise could be a useful system.

I don't have an opinion at this time regarding the mechanics of the system - I don't know if a ticket ought to be good only for a single try for a single slot or whether it ought to remain in the system.

I do believe, however, that the tickets ought to be quite inexpensive - on the order of a few dollars.  ICANN has created a system that delves deeply into useless and irrelevant matters, such as the applicant's business plan.  A system that merely inquires whether the applicant is able and willing to abide by internet standards would be orders of magnitude less expensive and intrusive than what ICANN has done.

I also believe that we ought to dispense with the microscopic, and to my mind, useless distinctions between TLDs.  As far as I'm concerned the concepts such as "sponsored" and "unsponsored" are nothing more than false creations, proceeding from the heat-oppressed brains of those who have tried (and to an uncomfortable degree, have succeeded in their efforts) to transform ICANN into an overwrought bureaucracy.  (My apologies to Shakespeare for my paraphrase from Macbeth, Act II, Scene 1)


Please see Lawrence Solum's response in which he describes approaches that obtain the benefits of a lottery without some of the drawbacks.

Also see see the thread in ICANNWatch in which an interesting discussion has developed regarding of the various approaches and considerations of TLD allocation mechanisms.

Posted by karl at April 4, 2003 10:39 PM