on Thu, 16 Apr 1998 06:55:13 +0200 (MET DST)

[Date Prev] [Date Next] [Thread Prev] [Thread Next] [Date Index] [Thread Index]

<nettime> pgMedia/Name.Space Replies to Paul Vixie

In the May edition of "The Cook Report," it was reported that
pgMedia/Name.Space's comments to the National Telecommunications and
Information Administration  (NTIA) on NTIA's domain name Green Paper
were "technically flawed."  This position was apparently based upon a
contemporaneous interview with Mr. Paul Vixie, a well-known expert on
the Internet's domain name system (DNS). Even a cursory review of the
Vixie interview, however, reveals quite clearly that Mr. Vixie does
**not** disagree with the technical feasibility of pgMedia's  proposal
-- namely that there is no technical limit to the number of global
top-level domains (gTLDs) that can be supported by the Internet "root
server" -- but rather raises a single, debatable, policy concern.

pgMedia welcomes the opportunity to discuss and debate the merits of
its ideas and posts this response to Mr. Vixie's interview . [This
reply is structured in context -- first the Excerpt from pgMedia's "ex
parte" filing with NTIA; second Mr. Vixie's comments; followed by
pgMedia/Name.Space's response.]

Background Links:

NTIA Green Paper --

pgMedia Comments --

Ex parte Excerpt No. 1:

>RE:Ex Parte Meeting of March 13, 1998
>. . . this letter is being filed to document the ex parte meeting that
>took place in the office of Becky Burr at 10:30 AM on Friday, March
>13, 1998 regarding the Department of Commerce's ("DoCs") proposed rule
>on management of the Internet Domain Name System. In attendance at the
>meeting was myself and Glenn Manishin of Blumenfeld & Cohen -
>Technology Law Group, Paul Garrin, President of Name.Space, a pgMedia
>Company, as well as you and Ms. Burr on behalf of NTIA.
>During the meeting, my colleagues and I described the nature of the
>services and technology provided by Name.Space. In particular, we
>detailed how Name.Space provides Internet users with a wide variety of
>generic top level domain names (gTLDs), or domain names that reside
>after the dot (e.g. ".sports" or ".fun").
>We also indicated that pgMedia, the parent company of Name.Space, will
>be filing comments regarding DoC's proposed rule. Much of our
>conversation centered around the nature of those comments, which can
>summarized as follows:
>A.The Internet root servers (the "dot") function as the "traffic cop"
>of the Internet, allowing all ame service providers to look to
>authoritative servers for the "toplevel zone" files necessary for a
>seamless Internet

Vixie Comment No. 1:

>Vixie: it's not a "traffic cop" it's a "central database". pgMedia is
>trying  desperately to cast this problem in terms of "any set of
>cooperating name servers could coherently serve dot" which is also Jim
>Fleming's point of view. The end goal is that anyone who runs any such
>server can add names, and anyone who runs any other server who does
>not add the same names pointing to the same places is defined as
>"uncooperative." That is my fundamental point of opposition to their
>proposal: I don't believe that anyone who wants to add names should
>simply be able to create a "root"-like name server and add them and
>then whine like crazy until the other "root"-like name servers also
>add them.

pgMedia's Response No. 1:

pgMedia does not disagree that the is a file.  "Traffic cop"
is simply a common-sense metaphor to explain, in, non-technical prose,
the function that this file performs. Mr. Vixie is correct in stating
that pgMedia believes that any registry that meets the basic technical
requirements for running a registry should be allowed to add new TLDs.
 However, his assertion that "whining" could or should be used to
achieve universal resolvability does not accurately represent
pgMedia's proposal.  It is pgMedia's position that there must be a
standard set of protocols used by any registry wishing to be
recognized at the root level -- see pgMedia's comments to the NTIA to
assure network stability.  Unilke Mr. Vixie,  pgMedia does not believe
that root level recognition is a privilege, but rather -- at least so
long as Network Solutions, Inc. (NSI) retains monopoly control of the
root server files -- a right.

Ex parte Excerpt No. 2:

>C. Coordination of root servers does not require a single "root," but
>rather consistent technical standards for interoperability and
>symmetry of all servers so to maintain universal connectivity on the
>             1. pgMedia's technology can work to maintain root server
>                synchronization by implementation of SINDI
>                (Secure Internet Name Data Integrator)

Vixie Comment No. 2:

>Vixie: SINDI would be a way to distribute the _capability_ of adding
>names, simply by saying "anyone running a "root"-like server is in the
>club". If that were the goal, we could achieve it by having the new
>IANA nonprofit be required to add any name for anybody who asked, and
>then there's no new technology. pgMedia is just trying to recast the
>problem in a technical light so that they can come out as a "member of
>the club" which they aren't now since it's not a club.

pgMedia's Response No. 2:

Mr. Vixie is absolutely correct that addition of the names to the file would achieve universal resolvability of those names.
Had he taken the time to review pgMedia's antitrust lawsuit against
NSI, he would have  realized that that is all that pgMedia requested
of NSI prior to filing its complaint. SINDI is a proposal made by
pgMedia as to how the root functions could be decentralized while
maintaining the coherence and stability of the Internet.  Mr. Vixie is
correct that implementation of SINDI would allow any participating
registry to add names under any existing TLD; however, he fails to
mention that it would also prevent any single point of failure error,
like the one caused by NSI which brought down large portions of the
Internet over the summer.  Clearly, some new technical approach is
required if the Internet is to move beyond the current practice of a
single, non-diversified source for root administration. Mr. Vixie may
believe that there is no "club", and perhaps this myopic view is a
result of Mr. Vixie's membership in the very club whose existence he
denies.  pgMedia, on the other hand, wishes only to open-up the domain
name registration market, for its own benefit, and for the benefit of
any other company that wishes to enter the market.  If the marketplace
is a club, then yes we want to join.

Ex parte excrept No. 3:

>B. pgMedia has long supported shared gTLDs, and believes that gTLDs
>are a public resource that should not be within the commercial or
>competitive control of any one entity; in addition, pgMedia has
>developed technology to facilitate seamless, interoperable sharing of
>toplevel domains by registries using SINDI.

Vixie Comment No. 3:

>Vixie: Great.  I've got similar technology here which I call EMACS.
>your mileage may of course vary.

pgMedia's Response No. 3:

pgMedia is happy to hear that Mr. Vixie has similar technology. If he
would like to contact us to discuss the differences, and similarities
between our systems, we are always happy to discuss our developments.
Since some new technology is needed to meet NTIA's conclusion that a
"set of authoritative roots" should replace the current,
single-machine model, it is clear that the Internet community will
need to develop standards for coordination and interoparability of
multiple root servers.

Ex parte Excerpt No. 4:

>A. NTIA correctly adopted pgMedia's approach of "shared" registries,
>under which all gTLDs, including existing ".com" registry, will be
>open to competitive registration of second-level domains

Vixie Comment No. 4:

>Vixie: it's not pgMedia's approach [sharing], it's IAHC's approach and
>this correction should be made in the record.

pgMedia's Response No. 4:

pgMedia was only pointing out that Paul Garrin, President of pgMedia,
and the company have been advocating the sharing model as a matter of
public policy for some time.  In fact, we did so even before IAHC was
formed.  (See
However, if Mr. Vixie wants to take credit for the idea, we have no
objection.  pgMedia is only interested in implementing the sharing
model to the benefit of the Internet community; childish "bragging
rights" benefit no one.

Ex parte Excerpt No. 5:

>C. A market-driven system will allow for the greatest introduction of
>efficiency, innovation and price reduction into a registration
>services market that has suffered greatly from the stifling effects of
>1. Name.Space innovations to date include seamless "smart" Whois
>service; Instant, fully automated name registration with secure online
>payment options; Portable Address Management; Secure, realtime
>synchronization of toplevel zones
>2. Name.Space system is working today -- competitive gTLDs are a
>D. Shared gTLD registration is possible today in the Name.Space gTLDs,
>using the SINDI protocol that pgMedia has developed independently
>A.There are no technical constraints on the number of gTLDs that can
>operate simultaneously on the Internet, and absolutely no basis for
>any concern that expansion of the TLD namespace will contribute even
>to transitional instability of the Internet

Vixie Comment No. 5:

>Vixie: This is unfortunately true.  The caching model would just move
>up one layer. A million names in .COM is not fundamentally different,
>regarding caching, from a million names in ".". The IAHC people spread
>a bunch of F, U, and D about this, to the detriment of their
>credibility (with me anyway). There are however valid nontechnical
>reasons why polluting "." is a bad idea.

pgMedia's Response No. 5:

pgMedia is happy that Mr. Vixie at least recognizes the technical
soundness of pgMedia's model, and that there is in fact no practical
or  technical constraint on adding new gTLDs to the root.  While Mr.
Vixie may not like the idea of adding new TLDs (pgMedia does not wish
to respond to words like "polluting;" we expect that serious people
will recognize such invective for what it is), he is not a legislative
authority.  Most importantly, this concession -- that in fact a
million names can be added, without problem, to the "." root server --
completely undermines any claim by Mr. Vixie's, or anyone else, that
pgMedia's proposal for adding new gTLDs is technically flawed.

Ex parte Excerpt No. 6:

>1. The current "" or "dot" file which lists the currently
>recognized toplevel domains is simply a 75K text file, and updating
>the file for new gTLDs is technically trivial and transparent to the
>end users--new toplevel namespaces will simply become available and
>functional just as any new country code or new second level domain
>becomes transparently available as changes are made to the DNS
>listings on a daily basis without interruption of service or other
>noticeable  change
>2. The current zone file for "com." contains over 1.5 million entries
>and can scale indefinitely. The structure of all zonefiles is
>identical and used in the same way by the software--structurally, the
>"" file and the "com" file are identical, therefore the
>number of possible toplevel domains is equally as scaleable as is any
>other zone file in the existing DNS system
>3. Record before DoC from August comments does not support any
>technical reason substantiating NTIA's conclusion that the number of
>gTLDs in the short run "should not be so large as to destabilize the

Vixie Comment No. 6:

>Vixie: DoC's concerns from August were valid, for both technical and
>non technical reasons. Adding a million names to "." is not going to
>break the DNS model, but it will require a different root server
>topology than we have now. Note that the difference is a matter of
>provisioning and operations practice, not the addition of "SINDI".

pgMedia's Response No. 6: Mr.

Vixie will have to explain himself, because as stated this is patently
false. Certainly, the scale of the root file can at least be
comparable to the ".com" name servers without requiring **any**
network architecture revisions.

Ex parte Excerpt No. 7:

>B. Hence, there is no justification for:
>1. Arbitrary restriction of five new gTLDs between now and 9/98
>2. Role of non-profit corporation in setting policy for determining
>"the circumstances under which new top-level domains are added to the
>root system"

Vixie Comment No. 7:

>Vixie: Yes there is. Right now domain names can be identified by their
>structure rather than from their context. Someone can just put
>"" on a business card and it will be recognized as a
>domain name even without an "http://" in front of it. With a million
>names in "." this property disappears, and doing this unthinkingly
>would "destabilize the internet" and I'm against it for at least that

pgMedia's Response No. 7:

This is a policy consideration, not a technological one. Property
rights and potential customer confusion issues -- to the extent new
gTLDs would be any worse than the current system, with a nearly
infinite number of second level domain names -- do not undercut the
technical validity of our proposal.  Rather, these non-technical
concerns challenge the Internet Community to think about the future of
the Net rather than the past, and to develop a real directory system
for the Web, something that DNS was not intended to be and cannot

Ex parte Excerpt No. 8:

>C. Any gTLD administered by a technically competent registry should be
>included in the root server system.

Vixie Comment No. 8:

>Vixie: OK, then I want ".FUC_".

pgMedia's Response No. 8:

To the extent Mr. Vixie is serious about this request, he undoubtedly
is aware that ".FUC_" would not be an allowable domain name because it
includes the illegal  character "_". (Only the US ASCII characters of
0 to 9 and A to Z and the "-" (hyphen) are legal characters for use in
domain names; and of course, the "." as a delimiter).  However, if as
is more likely, he is just attempting to be polite about the notion of
people attempting to register profane words as TLD, than we have to
ask the question, so what?  Although registrations under the domain
name .FUCK may be rude and tasteless, it is not the place of the
registrar, or any entity in the Internet community, to restrict use of
that or any other speech without compelling technical justification.
It is also unclear how .FUCK is any worse than or,
perfectly legal registrations under today's system, or even,
which is a hihgly popular site operated by Hot Bot. People of
different sensibilities are offended by different words; should we now
use DNS as a system of censorship as to what words can and cannot be
used to identify Web sites?

Ex parte Excerpt No. 10:

>D. Trademark concerns are irrelevant to DNS administration, and thus
>whether or not it would be "more difficult for trademark holders to
>protect their trademarks if they had to police a large number of
>top-level domains" is no excuse or justification for restricting gTLD
>1. No registry/registry trademark dispute resolution process should be
>required, and domain holders should not be compelled to agree to
>judicial jurisdiction in any specific location

Vixie Comment No. 10:

>Vixie: OK, then I want ".MICROSOFT" and I will sue any root name
>server operator who adds it without pointing it at me rather than Bill
>Gates' box.

pgMedia's Response No. 10:

If Mr. Vixie wants ".MICROSOFT," he may find himself being sued by
Bill Gates and not the other way around. It is pgMedia's contention
that Mr. Gates and Microsoft, as well as all other trademark holders,
are responsible for defending their trademarks, and that this burden
should not be shifted to domain name registries. Nonetheless, as a
responsible registry we have voluntarily taken the step of insisting
that applicants for gTLDs that obviously potentially violate
well-known trademarks make some showing of approval from the
legitimate holder of the mark before the application is granted.

Ex parte Excerpt No. 11:

>E. Even if transitional limits are believed necessary, the US
>government should not (and cannot, per section IV) sanction any system
>that does not allow unlimited competition for gTLDs
>1.Green Paper's "first-come, first-served" approach to the 5 new gTLD
>registries is unworkable and foolish that is certainly true. the IAHC
>plan should go forward and Ira Magaziner should stop trying to invent
>from behind.
>2.Non-profit corporation's charter must require that it permit gTLDs
>served by all technically qualified registries to be included in the
>root server system on a non-discriminatory basis

Vixie Comment No. 11:

>Vixie: This is ridiculous.  See above.

pgMedia's Response No. 11:

pgMedia is not exactly sure what Mr. Vixie is referring to; however,
if he wishes to clarify this remark we would be happy to respond.

Ex parte Excerpt No. 12:

>3. NTIA should support immediate movement to a fully market-driven,
>openly competitive system of gTLD creation and registration

Vixie Comment No. 12:

>Vixie: If this occurs, then Jim Fleming and Karl Denninger will start
>their own competing "." registries since they won't like the one DoC
>pushes, and we'll have balkanization. It's only the perception of
>strong central glue that keeps everybody's "root.cache" file pointed
>in the same place. Open it up too quickly and there will be multiple
>competing overlapping intersecting incompatible namespaces, and what
>you get from a URL will depend on where you are and on who  your ISP
>is. Bad bad bad.

pgMedia's Response No. 12:

pgMedia is committed to preserving coordination of the "." and is
opposed to any efforts that will cause splits in the coherence of the
"."   To this end, pgMedia has developed SINDI which functions to
coordinate the contents of the "." as well as to enable the complete
sharing of toplevel and other levels in the dns chain, as well as
eliminate the single point of failure that is inherent in the current
monopoly-based system which caused the massive failures of nameservice
in July, 1997
#  distributed via nettime-l : no commercial use without permission
#  <nettime> is a closed moderated mailinglist for net criticism,
#  collaborative text filtering and cultural politics of the nets
#  more info: and "info nettime-l" in the msg body
#  URL:  contact: