Name.Space.Info on Mon, 19 Jul 1999 20:01:28 +0200 (CEST)

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

<nettime> Vision and Wisdom by the Inventor of DNS, circa 1996

The vision and wisdom of Dr. Paul Mockapetris, the computer scientist who
invented the Domain Name System, forsaw the need to expand the number of
Toplevel Domains (TLDs) and to share and distribute the authority and
management of the Root and TLDs in his presentation to the Global
Information Infrastructure Conference in 1996 (around the same time that
the Name.Space project was proposed at the Next 5 Minutes Conference in

In a telephone conversation with Dr. Mockapetris in August, 1998 I was
informed that the proposal below was shouted down by "folks who didn't
want to see it happen".  As a result, Dr.  Mockapetris, then head if the
Internet Engineering Task Force (IETF) resigned, and quietly withdrew into
his role as chief engineer for @Home, and into his own private business

Now that we have the events of the past three years in perspective, it is
enlightening to look back at this visionary text and ask ourselves what
went wrong?  Why is there still a stranglehold over the DNS by Network
Solutions?  Why is the US Government shielding NSI and allowing them to
run roughshod over an entire industry, making millions of dollars while
preventing real progress, innovation and internet self-governance? 

In a recent article by Dan Goodin of C|Net ( it was revealed that
Network Solutions has been paying "shills" to promote their agenda in a
lobbying effort in DC and at the various "stakeholders"  conferences held
over the past two years.  It's no wonder that NSI will use every trick at
their disposal to delay any change to their lucrative status quo. 

    "Network Solutions has continued to use the services of its long-time
     lobbying firm, the <<>Dutko Group. NSI also
     has sent paid consultants active in the domain name controversy to
various meetings convened by ICANN to give input on the controversy. Most
recently, the Herndon, Virginia, company sent Jay Fenello, a frequent
ICANN critic, to its meeting in Berlin, Germany, NSI spokesman Brian
O'Shaughnessy said. 

    NSI also pays Tony Rutkowski and Richard Sexton, two other prominent
figures in the public debate, for consulting work on domain name issues.",4,239-39113,00.html?

Why is there so much resistance to decentralize the root when, as Dr. 
Mockapetris comments in his text, and as he stated to me on the telephone,
it only requires " a few months of effort by a competent computer

With the failure of NSI's proprietary, so-called "shared registry system" 
and the broken "whois" database delaying the introduction of new domain
registrars and lower prices, it is important to review the importance of
independent technological development that is in tune with the tradition
of the open nature of the internet. 

Name.Space has designed such a system of decentralizing the Root and
distributing the management of TLDs with the SINDI (Secure Internet Name
Data Integrator) project, now under development by a "competent computer
scientist" who is presently a new partner in the Name.Space project.
Name.Space intends to release SINDI as an OPEN SOURCE project in the
coming months, to enable for the first time, true bottom-up management of
the DNS.  SINDI is finally the realization of Dr.  Mockapetris' vision,
presented in the text below. 

-Paul Garrin

 Founder, Name.Space


>From the Conference:

Global Information Infrastructure

"National and International Initiatives for Information Infrastructure"
Symposium January 25-27, 1996

Harvard University Archive

                             A competitive DNS operational structure 

                                          Paul V. Mockapetris 




Domain names are a necessary resource for the operation of the current
internet and are likely to see an expanded role in the future. No
alternative is on the horizon, so we need an effective policy for
alocating them at an accelerating rate as internet technlogy is adopted
and expanded.

This leads to prioritized objectives: 

1. An unlimited supply of domains (the ability to create names "under" 
some point in the tree), and domain names (specific single names). This is
our prime directive.

2. The ability to experiment and develop new naming structures. 

3. The ability to allocate "vanity" names to all comers, as distinct from
operational, but non-mnemonic, names. (e.g. vs or



The recent actions by the NSF to create a more realistic structure for
domain registration should be recognized as appropriate short term measure
which will be less and less appropriate as time passes. Our goal should be
to use the "breathing space" which NSF has created to put in place a
system which can scale to global coverage. 

At the same time that we raise our design sights and aim for a more
capable infrastructure, we should also lower our regard for solutions
imposed in a top-down manner; attempts to force X.500 down the throats of
the internet for naming have all uniformly failed, whether they originated
from IETF working groups, the IAB, or even the government's GOSIP process.

The general method


The two traditional ways to govern a service are: 
regulation of a monopoly, 
market control through competition

The typical plan will contain elements of both of these. 

Most plans recognize the necessity for allocation of country-code domains,
e.g. .US, .FR to the respective national authorities. Each country then
gets to set its own policies within that domain.

Beyond country-based domains, we have "generic" domains, such as .com and
.edu. These serve two purposes:

They provide a home for organizations that are multinational.  They
provide a home for an organization which will not be recognized by its
country or chooses to be independent.

The country codes can be seen as the first level of distribution of
authority. Proposals such as the ISOC's centralize all remaining authority
in a single place and attempt to regulate the monopoly; this proposal
takes the opposite principle to heart and attempts to distribute authority
and control as widely as possible, and use market control through

The way forward

The DNS will continue to use replication for its databases, and will also
add other technical features: nothing in this proposal changes that. 

What we will do is add mechainisms for: 
1. Splitting the control of a domains's policy and registration. 
2. Distributing control of different domains. 

The first step is a technical one, involving a few months of effort by a
competent computer scientist.  For any domain, multiple registries can be
certified, and they register new domains by following anagreed domain
policy (DP), and then contending for names in a "mutual exclusion" server
(MES) ona first-come, first served, basis. The DP, including dispute
resolution, etc. may be created by an external organization. The domain
registries (DRs) may charge whatever they wish for the registration
service, and whatever they wish for long-term services (revalidation if
called for by DP). The MES is a low cost service which can be selected by
whatever means the DRs choose, so long as it is a disinterested party. 

At this point there is no longer any reason why we need a registration by
a single agent. We could have multiple registries for the COM domain, for
example. However all registries would be required to implement a single

The second step is to distribute authority for policy creation. The
domains of interest are: 

1. the root (i.e. the power to create new "top-level domains" (TLDs)  such
as country codes and .COM, .EDU etc.

2. The TLDs themselves. 

Most importantly, the general principle should be that authority for these
should be distributed. In particular, the orgaization which creates the
policy for the root should be precluded from authority over ANY TLDs. 
Similarly, policy authority for .COM should preclude policy authority for
any other TLD.

Distribution of authority and creation of model policies should be our

Immediate Recommendations


The policy control for the root domain, all country codes, and the generic
domains should be distributed as far as possible. The ideal should be that
no authority controls more than one of these domains.  The technology to
allow parallel "first-come, first-served" name allocation by multiple
registrars in a single domain should be tested on .COM.

The root domain policy should be structured to allow creation of new
generic TLDs on a metered basis, using the usual IETF public forum

An example division of authority might be: 

Domain Policy Registrars

====== ====== ==========



#  distributed via nettime-l: no commercial use without permission of author
#  <nettime> is a moderated mailinglist for net criticism,
#  collaborative text filtering and cultural politics of the nets
#  more info: and "info nettime-l" in the msg body
#  un/subscribe: and
# "un/subscribe nettime-l you@address" in the msg body
#  archive: contact: <>