Trigraph domain names: the end of a frontier.

Introduction

Figure 1: trigraph domain creation
In May 1999 I'd read many news articles about the market for domain names: some claimed that good domain names were becoming hard to find. Shadowy figures owning thousands of domains would occasionally get in the news because of technical glitches or trademark disputes. Although the same people would appear again and again, news stories would often omit their history -- it seems that large speculators keep a low profile to attain an information differential over domain buyers.

Wanting a direct source of information, I decided to study trigraph .com addresses, that is, any three letters with .com, such as abc.com or xyz.com. With a total of 26x26x26=17,576 addresses, the space of trigraphs is practical to extract from the WHOIS database. Trigraphs are commercially valuable: the trigraph space contains the names of major corporations such as ITT and CBS as well as people's initials. Web businesses are becoming aware that a short domain name makes it easier for customers to reach them. (Barnes and Noble, for instance, probably bought bn.com from an speculator around the time of their recent IPO since they found customers could remember amazon.com but couldn't remember if it was barnesandnobles.com or barnesnnobles.com.) With a tight market, it might be profitable to buy random trigraphs for later sale. People are already doing this.

We distinguish two kinds of domain holder who attempt to profit by selling domain names: domain name speculators and squatters. Domain name speculators, like other speculators, buy something with the hope that it will go up in value. Domain name squatters, on the other hand, take advantage of the two month grace period that Network Solutions allows before paying for a domain name. Squatters reserve a large number of domains and hope to sell them before they expire. Both speculators and squatters have been criticized, and some people use the words interchangably. Although speculators may raise the cost of running web sites, they do invest their own money, take risk, and support the cost of DNS infrastructure. Squatting, simply, is theft, and places a large load on DNS (when I bought tapir.org in Feburary 1999, my purchase was delayed for weeks since NSI's servers were overloaded by squatters attempting to steal expiring domains from other squatters.)

Legal

Because of people abusing DNS, NSI has recently begun to append the following to each WHOIS query:
You agree that you will not reproduce, sell, transfer, or 
modify any of the data presented in response to your search request, or 
use of any such data for commercial purpose, without the prior 
express written permission of Network Solutions.
This wording may prohibit this publication. However, everyone in the web business uses the WHOIS database for commercial purposes every day. Our results demonstrate that people are scanning the WHOIS database to support their speculation and squatting operations. The proper use of DNS can only be found if open information is available about abuse. In compliance with the above agreement, we will not transfer a copy of our database to you, although we may soon release our tool for mirroring the WHOIS database.

Methods

I mirrored a portion of the WHOIS database in a relational database, instantDB with a multithreaded Java application. A number of problems arose, such as parsing results that return multiple hits, an aproximate .01 system wide failure rate for queries, and a need for mulithreading to get acceptable performance on a slow net. On a fast network, my crawler encounters problems with InterNIC's throttling mechanism: the WHOIS database blocks queries from your IP address for about ten seconds if you issue queries too rapidly. My crawler slows down when it detects this condition.

Time Series

Figure 1 is a cumulative graph of the number of domains existing in May 1999, created previous to a certain date. It is related to, but not identical, to a graph of the number of domains occupied versus time: it does not reflect domains that have expired. It shows that nearly all trigraph domains are currently occupied. The most interesting feature of Figure 1 is a sudden jump on March 1, 1999. This is the result of a large acquisition by a single party of 1,300 trigraphs.

Figure 2:log of trigraphs created
Figure 2 is a log plot emphasizing the early history of trigraphs. The WHOIS database was inaugerated in early 1985, and the first trigraph registrant was bbn.com in April 1985. After a short spurt of growth, trigraph occupation grew exponentially with a doubling time of about 1.2 years from spring 1987 to the winter of 1995. The number of trigraphs in use increased a hundredfold in this time, with about 10% created by the end of 1995. A short spurt of growth happened then, when public interest in the internet peaked, and the space of trigraphs became quickly saturated. This behavior is similar to that observed in populations approaching a carrying capacity and in the adoption of social and technological inventions.

The big players

Figure 3:major trigraph holders
To identify major trigraph holders, I searched for the primary nameservers hosting the largest number of trigraph domains. There are 18 nameservers that host more than 100 domains. Major holders could also be found by looking at administrative, technical, billing or zone handles, but some large trigraph holders, such as the people who use this.infa.net as a nameserver, create new handles for each domain, possibly to make it harder to find out what they're up to. My strategy isn't foolproof -- many small domain name holders outsoure their DNS to other providers, so it's possible some server contain domains held by hundreds of unrelated people, but, this is not the case for the five holders discussed in this paper. A time series plot reveals the differing strategies of the top five trigraph holders, Figure 3. For instance, this.infa.net acquired over 1,000 domains on a single day, March 1, 1999 while ns1.namething.com acquired over 600 domains gradually between early 1997 and the present. Time graphs can be used to tell speculators from squatters. Since domains can be held for up to two months without payment, a domain name holder with a large number of young domains is a possible squatter. Although March 1,1999 occured more than two months prior to our sweep, there is a warning period before a domain is taken away, so I cannot rule out (but I cannot prove) that this.infa.net and www.transmogrifier.org are being used by squatters although I can establish that the others are (at least primarily) speculators.

www.infa.net: 1,316 domains
Infanet (INFA2-DOM)
PO Box 221
Fayette City, PA 15438
US
In a few days in March 1998, this nameserver, and the operators of www.transmogrifier.org acquired about 10% of all trigraphs. The connection between them, if any, is unknown. On contacting their representatives asking about a randomly chosen trigraph, I was told to make an offer. They were receptive to an offer of $1,200, considerably less than quotes from speculators.

ns1.namething.com: 632 domains
9921 Carmel Mt. Rd. #248
San Diego, CA 92129
US
Namething has a large portfolio (4% of trigraphs) acquired before November 1998. Namething openly sells trigraphs for the price of $4,000 US as well as digraphs for $7,500 US.

ns1.telepathy.com: 591 domains
P.O. Box 9911
Washington, DC 20016
US
Telepathy openly sells a number of domains with no specific mention of trigraphs. I contacted Telepathy and found their prices are similar to Namething's.

www.transmogrifier.org: 571 domains
3330 La Mesa Dr. #12
San Carlos, CA 94070 US Like www.infa.net, transmogrifier's porfolio was acquired rapidly in March 1999. It's possible that this orgy of buying was triggered by a crackdown that Network Solutions made against domain name squatters in late Feburary.

ns1.downtown.com: 331 domains
3330 La MesElie S. Abouzakhm
7656 shaughnessy suite 6
Montreal, Quebec H2A 1K4
CA
This registrant is certainly a speculator, featuring an old portfolio which was probably larger than it's competitors as of January 1998. However, unlike some other speculators, downtown.com keeps a low profile and doesn't appear to openly sell domains on the web.

Is it right?

ICANN is currently planning to change the domain registration rules to eliminate squatting. What about speculation? Although many of us have experienced frustration when a domain we wanted was available only at a seemingly ludicrous price, it seems that speculation is inevitable in a capitalist economy. Many domains clearly have a commercial value greater than the $35 /year registration fee, and, advocates of the market tell us that, if domains are traded freely, they'll be owned by those who can get the most value out of them. One problem is that we don't know what internet businesses are really worth (which is why internet stocks are volatile) so we can't agree on what domains should cost, which is why some domains are offered for five figure prices. Like many things in a capitalist soceity, a market favors those with money, making it harder for people with talent and worthwhile ideas but no cash to get a good domain -- however, speculators haven't yet bought everything so we still see sites like http://www.gwbush.com/.

Links

Igoldrush is a dated, but comprehensive site about domain name speculation. Bestdomains is a major broker. News stories about namespace issues from news.com or wired news. The people behind whitehouse.com own a portfolio of domains that end in "sucks". Our own portal, Positive Propaganda.
 

Paul A. Houle
Last modified: Mon Sep 20 21:17:19 CEST 1999