IP: a rather decent commentary on "Domain Name Registration Fees

[David Farber <farber () central cis upenn edu>]

Domain Name Registration Fees Underway
--------------------------------------
  by Glenn Fleishman <glenn () popco com>


  The National Science Foundation (NSF) changed the funding picture
  last week on one of the few remaining U.S. federally funded
  Internet projects. The NSF and the InterNIC's Registration
  Services division, which registers and maintains domain names,
  announced that beginning at midnight on 14-Sep-95, all new domain
  name registrations under its authority would cost $100 and include
  two years of registration. Yearly renewals for new and existing
  domains will be $50, due on the anniversary of the initial
  registration.


  Domain names are technically the "human-readable" form of an
  Internet address. Every machine on the Internet is assigned a
  unique number: an Internet Protocol (IP) address. The number is in
  the form: 0.0.0.0, often called a dotted-quad. With this number,
  you can directly identify a specific machine anywhere on the
  Internet. For instance, Apple's Web server www.apple.com is at
  17.255.0.64. All of Apple's Internet machines have names within
  the apple.com domain.


  The fees will eventually replace U.S. federal funding for domain
  registration; currently, Network Solutions, Inc. (NSI), operates
  the Registration Services division under a five-year contract to
  the NSF that was awarded in 1993 under competitive bidding. The
  bid included the possibility of eventually collecting fees for
  domain names. Conservative estimates suggest that the fees could
  bring in more than $5 million in 1996, assuming the fees cause a
  considerable drop-off in new registrations and renewals.


  The action was widely expected and has been discussed at length in
  newsgroups and such mailing lists as com-priv - a list which
  endlessly and post facto discusses Internet issues - as well as in
  print media that covers online issues. The move apparently came
  without advance warning to prevent a flood of last-minute
  registrations. (Some reports indicated that the NSF planned to
  announce the policy this week, but a leak caused the early release
  of information.)


  Because the NSF is so deeply involved in the Internet, they've
  placed piles of useful information about the decision, the fees,
  and the history of why they have the authority to do this at:


http://rs.internic.net/announcements/index.html


  Prognostications and explanations by third parties have poured out
  since the announcement, which rated front-page or front-of-
  business-page placement in major newspapers, including the New
  York Times and the Wall Street Journal. Before we wade into an
  analysis, let's discuss the different components.


**What is the InterNIC?** The NSF created the InterNIC (Internet
  Network Information Center) to provide information and
  registration services to Internet users, which - in 1993 - meant
  largely academic, government, and corporate organizations rather
  than consumers and small businesses. The NSF has, for several
  years, been in charge of big chunks of the Internet authority -
  the parts of the Internet that make the final decisions on how
  policy is transformed into real equipment.


  The NSF ceded a chunk of authority on 08-May-95 when it shut down
  NSFNet: the Internet backbone that existed before commercial
  networks effectively made it redundant. Without NSFNet, the NSF
  doesn't exert much control or influence on the day-to-day workings
  of the Internet, but the NSF does direct the Internet's evolution
  into an ever-faster animal. (The Internet Engineering Task Force
  [IETF] actually drives technological change on the Internet, but
  NSF entities drive the implementation.) For more on NSFNet, see my
  article in TidBITS-275_ and:


http://www.boardwatch.com/mag/95/jun/bwm1.htm


  [To learn more about the IETF, see Paulina Borsook's "How Anarchy
  Works" in the Oct-95 issue of Wired - unfortunately, it's not yet
  online. -Tonya]


  One of the InterNIC's main functions, run by its Registration
  Services division, is to register domain names. Domain names were
  developed as a way to more mnemonically identify and group
  machines. With a domain name, you can have any number of
  subdomains. Subdomains are separated by dots but read right to
  left, from the most general category to the specific machine name
  or service name. So a name like "bilbo.engineering.ufoo.edu" is
  read like this: "_edu_ is the educational top-level (farthest
  right) domain; _ufoo_ is the second-level subdomain under
  education indicating this is the University of Foobar;
  _engineering_ is a subdivision of the ufoo.edu subdomain; and
  _bilbo_ is probably the individual machine name in the engineering
  school."


  The InterNIC is responsible for domain names that fall into five
  top-level categories: Education (.edu), Governmental (.gov), Non-
  and Not-for-Profit Organizations (.org), Commercial (.com), and
  Network (.net). The domain name registration fees apply only to
  second-level domains which fall under those hierarchies. Military
  (.mil) organizations handle their own authority and the .edu and
  .gov hierarchies will continue to be subsidized by federal funds
  for now.


  An alternate hierarchy already exists in the United States: the
  .us top-level domain. Many service providers have adopted this
  use, which is geographical in nature. (Some criticism has been
  levelled at geographical organization, since the Internet isn't
  place-driven.) Others have registered in both the .com or .org
  category and the .us domain to cover both bases. International
  hierarchies abound, with dozens of countries having their own
  top-level domains (such as .au for Australia) and authorities. The
  NSF's announcement doesn't affect any of these hierarchies.


**Meanwhile, back at the fees...** The NSF ostensibly instituted
  domain registration fees because of the precipitous growth in
  demand for registrations and the concomitant increase in costs
  necessary to keep up. The original contract with NSI was for $5.5
  million over five years, which is clearly inadequate to handle the
  tens of thousands of existing domains, and the potential tens of
  thousands to come over the next three years of the contract. (The
  grant, incidentally, doesn't cover just service, but all the
  associated network, staff, and overhead expense.)


  Although there are currently 110,000 second-level domains under
  InterNIC's authority, the Internet has millions of email users,
  none of whom are affected. Organizations that use domain names are
  either service providers (doing dial-up or other Internet
  connectivity), commercial online services (America Online,
  CompuServe, et al), or corporations with their own feeds. AOL, for
  instance, uses the domain aol.com for all three million users'
  email, plus their corporate stuff. So this means that AOL will pay
  the whopping fee of $50 per year to continue to use that domain
  name.


  The real effect will probably be felt by service providers who
  charged little or nothing to register individual domain names for
  their users and now face thousands of dollars in yearly fees. In
  most cases, these domains are registered to individuals who
  maintain access accounts or leased lines with the provider, so the
  additional $50 a year can be absorbed if their bills are high
  enough, or tacked on in the case of simple email accounts. Savvy
  Internet providers probably have a provision in their user
  contracts for passing on new fees to the registrant of the domain.


**Registration fees are your friend?** Why should we celebrate
  being charged money? Many reasons, some political, some legal,
  some financial, some practical.


  First, this move establishes an independent source of funding for
  the InterNIC, independent of political vagaries. For now, NSF
  funding is probably safe given the importance placed on the
  Internet by House Speaker Newt Gingrich and Vice President Al
  Gore. But reducing or eliminating taxpayer support of a government
  plan is a rare and wondrous event.


  Second, and on the same theme, the NSF noted that a significant
  portion of the income (30 percent) will go to a legal fund to deal
  with the vast potential for lawsuits in the future. Under their
  existing agreement with NSF, U.S. taxpayers would pick up any
  legal bills, whereas the new agreement requires NSI to pay the
  bills from this fund. Remarkably, InterNIC has not yet been sued
  over domain name decisions.


  Third, InterNIC cannot keep up with the registration load given
  the current funding. More funding means more automation; the NSF
  said in one part of the announcement that InterNIC will clear
  through the backlog of registrations by the end of October, given
  their new ability to spend based on income derived from fees.


  Fourth, the Internet has always been about paying your own way,
  despite NSFNet and other subsidies. Even though Internet use
  always appeared "free" to users at academic institutions, money
  has always been involved. These days, any commercial Internet user
  is paying for an account, a leased line, an Internet feed -
  whatever. Having domain names be part of a U.S. governmental
  burden seems inconsistent.


  Fifth, this mechanism will reduce the proliferation of domain
  names to some extent, release unused and unwanted domains, and
  involve more accountability. The recent move by Kraft and Proctor
  & Gamble to register hundreds of domain names, some of them
  trademarks and others just English words, could have been
  prevented if enough time and staff were available at InterNIC.
  Many thousands of the 110,000 domains extant are probably
  inactive; why not reduce the administrative and technical burden
  in maintaining them?


**What about the naysayers?** Over the last year (and especially
  since the announcement) many voices of complaint have been raised
  at the InterNIC. Some have proposed starting alternative top-level
  hierarchies - possible only if the InterNIC and the international
  domain authorities agree. Without their cooperation, you would
  have a separate but unequal set of domains unreachable from the
  rest of the Internet without major kludges.


  Others have alleged that the bidding process was closed, there was
  no public discussion, and that NSI hasn't met its obligations. The
  NSF does a wonderful job answering these points, noting that the
  original free and open competitive bid in 1993 mentioned the
  possibility of fees; and that the most recent NSF review of NSI by
  an independent panel in Dec-94 (available to public scrutiny via
  the Web) showed that they had met their goals. The panel
  responsible for this review is a Who's Who of respected Internet
  experts.


http://rs.internic.net/NIC-support/nsf/review-toc.html
http://rs.internic.net/NIC-support/nsf/review-panel.html


  Critics who argue that there was no public discussion have
  disregarded the participation of NSI- and NSF-affiliated people in
  mailing lists, newsgroups, and other public forums in which these
  issues have been beaten to death. It's hard to imagine a more
  public forum than com-priv, for instance, in which the InterNIC
  has been a relatively active participant.


  Another important point has been missed in the discussion this
  last year. The InterNIC does a damn good job. Despite a lack of
  funding and geometric growth in registrations, the InterNIC last
  month implemented a one-day turnaround on all new commercial
  domain name registrations. I don't recall anyone complimenting
  InterNIC, unless it was "too little, too late." In fact, it was
  just enough and right on time given their load.


  The technical side of the whole megillah - the actual resolution
  of domain names (pointing requests for the second-level domains to
  the thousands of machines responsible for this on the Internet) -
  works like a charm. I can't recall a time since Aug-94, when I
  first got my own full-time Internet feed, that this has stopped
  functioning. The InterNIC isn't responsible for the design of the
  system (the IETF and many generous individuals are), but they do
  coordinate it and keep it running. And run it does.


  In the spirit of the Internet, I'd like to invoke the phrase: "If
  the existing system functions 100 percent of the time, if the
  mechanisms to perpetuate it indefinitely exist, and if it's
  constantly improving: keep it in place and help support it
  better." That's a high-tech version of "if it ain't broke, don't
  fix it."


  It's likely that this move will cause some lawsuits, lots of
  bellyaching, and not much dancing in the streets. But in the
  ultimate interest of the growth of the Internet, it's a good move.
  I'm putting my mouth where my money is: my company will be liable
  for as much as $2,500 in fees for existing domains over the next
  year as a result of this change. Some of this we'll absorb, and
  some we'll bill out to the domain holders. But since NSFNet was
  shut down last May, I've been waiting for the other shoe to drop.
  Now I can rest easy.


  [Glenn Fleishman has registered over 50 domain names for clients,
  friends, and relatives. He's a contributing editor for Adobe
  Magazine, a columnist for Web Developer (due out in November), and
  a freelance feature writer for InfoWorld. He also moderates the
  Internet Marketing Discussion List, one of the least rancorous
  mailing lists ever.]


http://www.popco.com/popco/glenn.html