IAHC-Draft David Collier-Brown Category: Informational Private Person Expires May 28, 1997 November 1996 Allocation of Domains: an Appreciation Status of this Memo This document is an IAHC-Draft. IAHC-Drafts are working documents invited by the Internet International Ad-Hoc Committee. IAHC-Drafts are draft documents and may be updated, replaced, or made obsolete by other documents at any time. It is inappropriate to use IAHC Drafts as reference material or to cite them other than as ``work in progress.'' Introduction This is the first of a short series of papers addressing specific parts of the general problem posed to the community. It is not a proposal as such, but rather a consideration of what factors affect a particular party of the decision and what courses of action might be considered. Aim This particular appreciation discusses the means of allocation of domain names to registries. Note that this is not about ``rent versus buy'': this is about models of how registries might obtain named domains. Situation There are a large number of prospective registries who would like to manage new top-level domains. This will be referred to as ``[many registries]'' in the rest of the document. There are a large number of possible domains they might like to manage, and they have in most cases specified which ones they wish to manage. ([many domains]) If, on the contrary, there is an limited namespace, some will not be able to obtain their preferred domains due solely to that limitation. ([namespace]) There are great differences in the financial return which registries might reap, depending on the popularity of the name. ([worth]) There is a proposed mechanism for applying, referred to hereafter as draft-postel, and registries which have applied for consideration under it or its successors. ([mechanism]) There are experimental top-level domains in use, notably through the Alternic, with experimental customers. ([experimental]) There are prospective registries who do not desire to use draft-postel, and presumably some who do not wish to use any IAHC-sanctioned mechanisms. ([uncooperative]) The enforcement mechanisms available are primarily moral suasion and the willingness of the Internet community to ``shun'' users of non-standard-conformant mechanisms. Courses of Action 0: Do nothing. This amounts to either non-allocation or allocations as per draft-postel. Draft-postel will be analyzed later, in section 3. If we elect to refrain from allocating new domains, we return to the previous situation, with a trademark-related scheme applying to a single TLD, ``.com'', with the majority of new applications occurring there. This addresses none of the above factors. (Note: course zero is conventionally ``do nothing''. This helps ensure sufficient coverage.) 1: California Gold-rush. The first scheme we might try is pure ``first come, first served''. This is very similar to the historical California gold-rush: anyone who can takes a ship to California and stakes out a claim. No two persons can have the ``same'' claim, but there is no standard about how big a claim one can have or if it can overlap another's, nor is there a body with sufficient power to enforce such rules. Instead there is ``frontier justice'': using whatever force one can muster to gain advantage, but with the limitation that blatant claim-jumping can lead to all one's neighbors banding together to hang you. In our context, this would be claiming first-come, first-served (or any other mechanism you'd care to name) is the allocation mechanism, but not having any means of enforcing it.[enforcement] This predictably will lead to conflict in U.S. and possibly foreign courts over ground rules, jurisdiction, infringement on trademark and dilution of trademark, without a clear body of precedent to guide the courts. Particularly obvious breaches of common law will be dealt with expeditiously by the courts (ie, claim-jumpers will be hung forthwith), but more difficult examples may drag on for years. Addresses: [many registries] [many domains] [uncooperative] [enforcement] Contradicts: nil Orthogonal To: [mechanism] [experimental] [worth] 2: Land Rush. The second course is to establish some ground rules and a means of enforcement, for at least part of the problem. This in turn is reminiscent of the Oklahoma land rush, which was organized by the United States to give prospective settlers in the ``Indian territories'' a means of competing with one another without quite as much lawlessness. It was still first-come, first-served, and it explicitly was a race to find and claim desirable territory. In our context, this might be defining first-come first-serve as an allocation scheme, but providing a contractual environment so everyone starts the rush ``fairly'' and ensuring that the first people there don't lose their land. Addresses: [many registries] [many domains] [uncooperative] [enforcement] Contradicts: [mechanism] [experimental] [worth] Orthogonal To: nil 3: Slow Land Rush. Next we have a ``slow'' land rush. Imagine that each few years, the U.S. government releases some small amount of land to claim. This is much easier to manage, and leads to fewer people trying to hang each other at any given time. This can be a good tactic if there are limited enforcement resources and high need for them at the time of the allocation. [enforcement] Alas, it assumes that there are sufficient resources to ensure no-one tries to sneak over the border and do their own allocation. In the case of the Indian territories, the former residents used to do this enforcement... This is substantially what draft-postel proposed: some number of domains are allocated per year, until demand drops off. Addresses: [many registries] [mechanism] [enforcement] Contradicts: [many domains] [namespace] [uncooperative] Orthogonal To: [experimental] [worth] 4: Homesteading. This was a kinder, gentler form of land rush, without any particular start time. For land that wasn't as attractive as the Indian territories, one could simply move in, put up a house and start farming. The limitation was that you had to register your intentions, had to actually live there, and had to ``improve'' the land by building a house and plant crops. If we were allocating domains that weren't particularly desirable, this would be a plausible course of action. Non-desirable domains might be ones with obscure names (``.aa2''), small numbers of prospective customers (``.reptiles''), or where the cost is high and the benefits low for operating a registry at all. [worth] Addresses: [many registries] [many domains] Contradicts: [mechanism] [uncooperative] Orthogonal To: [experimental] [enforcement] [worth] 5: Surveyors First. This was the Canadian approach to settling the west: first they send the police force, then the surveyors, then the road builders and finally the settlers. In fact, there was also a steady stream of homesteaders and whiskey-traders leapfrogging ahead of the surveyors and annoying the police and indians. In this model the land was surveyed first, by an organization that did not stand to profit from it, the road allowances set out, the church and government reserves defined and some of the more obvious town sites surveyed as well. This is equivalent to defining the domains that would be made available, reserving certain of them (say, .museum, .park) for particular purposes, and setting the permissible forms of name for the rest, then making them available according to some ordering. This in particular would avoid two registries trying to create two domains with the same name spelled differently (.biz, .bus) then accusing each other of claim-jumping. Similarly it would reduce the cost of correcting errors, such as a for-profit organization claiming .park, an arguably non-profit domain, by allowing the presetting of ground rules. Addresses: [many registries] [namespace] Contradicts: [mechanism] [experimental] [uncooperative] Orthogonal To: [enforcement] [worth] 6: Ellis Island. This was the U.S. response to massive integration in recent memory. They set up a quarantine station off the shore of new York, and started interviewing prospective immigrants. If you didn't meet their criteria, you didn't get off Ellis Island. If you were ill, you stayed until you were better, or died. The order you arrived in was a consideration, but you only got ``in line'' if you qualified. This is equivalent to setting substantial technical and public policy constraints on being a registry. The technical constraints are similar to the medical ones, the public-policy constraints are similar to the immigration criteria. An example public-policy criteria might be one I've been assuming thus far: that only one registry could claim one (exclusive) domain. This is similar to U.S. immigration only letting one ``Smith family'' use a particular set of supporting documents speaking to the desirability of the family as immigrants. There is a broad range of behavior possible here: just as U.S. immigration applied different criteria at different times, there is a broad range of public-policy issues which might be part of the qualification process (ie, service standards) or which might be specifically excluded from the process (conflict resolution policies). Addresses: [many registries] [many domains] Contradicts: [mechanism] [experimental] [uncooperative] Orthogonal To: [enforcement] [worth] 7: Sale/Lottery. Next, one might sell or raffle off land or registries. This is actually done by immigration departments: Canada sells the right to immigrate, and the U.S. gives away green cards in occasional lotteries. These are primarily means of reducing the number of applicants, and of selecting applicants with substantial financial reserves. This might be done for domains which are in short supply [namespace], or which are inherently capital-intensive [worth] to operate, respectively. Addresses: [many registries] [namespace] [worth] Contradicts: [uncooperative] [enforcement] Orthogonal To: [mechanism] [experimental] 8: Fiat. And of course, one could just pick someone and bless them. If the business had few applicants, and there was little to distinguish them from one another, this might be worthwhile. Addresses: nil Contradicts: [many registries] Orthogonal To: [many domains | namespace] [mechanism] [experimental] [uncooperative] [enforcement] [worth] (The last one or two courses of action in an appreciation are often deliberately too extreme to consider, in hopes of ensuring full coverage). 9: Combinations. Of course, combinations of the above are possible. For example, Canada had an Ellis Island and also sent its surveyors out first. Conclusions For most, but not all, possible combinations of kinds of domains and registries, the author is of the opinion that this is almost unavoidably some form of land rush, and that 1. limiting the speed (3) of the rush is desirable, if possible, 2. an Ellis Island (6) of limited difficulty is a suitable filter for seriousness and probity, and 3. send the surveyors first (5) is the single most desirable factor in allocating domains to registries in the long run. Addresses: [many registries] [namespace] [mechanism] [enforcement] Contradicts: [many domains] [uncooperative] Orthogonal To: [experimental] [worth] This is not the only useful combination. My preference is just that: a preference. For those looking for the background to the preference, I commend the Yukon gold rush to your attention: the (mostly American) miners arrived at the pass to the Yukon to be greeted by the Royal Canadian Mounted Police with a list of equipment and food they would need and rules that they would be expected to obey. If they didn't like them, they could go home. Attending the Yukon gold rush was perilous and exhausting, but not necessarily life-threatening. And it was very much worth the effort. Dave Collier-Brown / davecb@hobbes.ss.org