In time of response order: There is Leo's reference to the not yet concluded RAA process, in which a para contains possibly relevant "registrar shall" terms. This is forward looking (the proposed RAA is not yet required by the Corporation) and may apply only to parties contracting with the Corporation for the right to provide "registrar services" to some, not all, registries, operated under some contract with the Corporation. It may, if read creatively, solve the problem for a "new registrar" offering registration services for one or more "new gTLD(s)", but that may be the extent of its applicability. If the creative reading fails, AAAA and DS may fall outside of these "registrar shall" terms. Next, there is Mark's observation, citing the same proposed RAA, that if the registrar provides a web interface (note well the "if"), and this web interface provides a means to edit A and NS records, there is no additional functional requirement for AAAA and/or DS. Mark observes that AAAA and DS updates require more from the registrant (also the registrar, when software, testing, staff (technical, support desk, and legal) training are not abstracted by a magic wand), and then observes that:
Maintenance of A, AAAA, NS and DS records are core functionality and need to be treated as such.
Here I personally differ. For those not paying attention to my slightest utterance over the past 15 years of NEWDOM policy and technology... I am sure that v6 matters to some, but not all, at least not in the manditory-to-implement-yesterday sense advocated by the v6 evangelicals (who have captured the Corporation on this issue). I'm also sure that DNSSEC matters to some, but not all, at least not in the manditory-to-implement-yesterday sense advocated by the DNSSEC evangelicals (who have captured the Corporation on this issue). Some 80% of the available-by-contract names in the namespace published by the US DoC through its contractors, Verisign and the Corporation lie in one zone, which became signed as recently as March 31, 2011 (see Matt Larson's note to the DNSSEC deployment list). Of those a very small minority are signed. v6 availability statistics for North America, where over half of the registrars possessing the accreditation of the Corporation to offer registration services for this namespace are domiciled, and by inference, a substantial fraction of the registrant domains are hosted, are similarly a very small minority. It seems to me, and I don't suggest that anyone else hold this view, least of all the v6/DNSSEC evangelicals, that it is possible for one or more registrants to exist who desire neither to sign their domains, nor to ensure their availability via v6. This registrant, or these registrants, would be well served by a registrar which did not offer AAAA and/or DS record editing services. It also seems to me, and again, I don't suggest that anyone else hold this view, that the number of such registrants could be sufficient to support a cost recovery operator of a namespace which is not signed, and for which no AAAA record, in the namespace published by the US Doc (through its contractors, blah blah) exists. Obviously, the converse view carried the day, though not (yet) for namespaces not operated under contract with the Corporation. Leo's follow-up on input valuable to the consultation would, I think, have scope limited only to "new registrars" offering registry services to "new registries". See the "very small minority" observations, supra. Finally, Bill points out that there are several contracts still applicable, and the rather turgid nature of the policy and implementation dialog(s) of the opposing parties around the proposed 2013 contracts. There are registrars operating under the pre-2009 and the 2009 contracts looking at forming distinct legal entities to enter into the eventual post-2012 contract, a reasonable scenario is trademark exploitation and exit, iterated across a series of unlikely to be sustainable product launches, and there are registrars that simply won't bother with future "landrush" sales any more than they bother with current "expiry" sales. The point being the "trigger" Bill mentioned isn't universal, it really is limited to those who's registrar business interest in the Corporation is brand extension, or are applicants for vertically integrated registries. Bill observes that the ATRT2 is a possible venue. This may be, but on the whole, the interest of the United States Government in the capture of its delegated rule maker by the regulated businesses is limited. There was one mention "... a group of participants that engage in [Corporation]'s processes to a greater extent than ..." in the AoC of September 2009. Subsequent public communications of the Government concerning Notice and Comment obligations, usually referred to as "accountability and transparency" by the Corporation, are not evident to me. Bill closes with an obvious recommendation -- pick a registrar that works for your definition of "work". Of course this is the only useful act in the continuous present, as the Corporation adopted ab initio, and retains, a de minimus and reactive approach to contracted party oversight (see Sightfinder, tasting, ...) and "caveat emptor" (aka "registrant choice") exercised through initial selection, and transfer (when possible), are the means for "regulation through the market" of the registrar function. Basically, its "good luck with that" for any notion of MAY or MUST language on functionality, sort of like the fate of BCP38 and similar. My mileage is zero. Eric