Do domain names matter?
- Below is my essay on the state of domain names, and generally the idea
of naming entities online. You may recognize some of the text below
from some of the discussions we had here a few months back.
For a permalink (with links), go to
July 25, 2003
Is it just me, or are we paying less attention to the Domain Name
System than we used to? Seems like only a few years ago that the
tech-culture world was attuned to every new angle in the ongoing
struggle over the DNS' management. You couldn't read the front page of
Slashdot without catching one heavily commented-upon story on alternate
registries, trademark disputes, or the latest ICANN board meeting.
But today? Hardly a peep. Not because the problems have magically
solved themselves: The MPAA, for example, just sent a cease-and-desist
letter to a blogger with the domain name www.ratednc-17.com. But a
story like this won't draw the same attention it would have before. And
by the way, what ever happened to those new top-level domains, like
.biz, .info, and .name? Some of those are two years old and wide open
for business—homesteads desperate for homesteaders.
This could be simply a temporary development. If the economy picks up
we might see an uptick in the number of dot-coms suing hapless
webmasters, and our outrage might rise accordingly. Or maybe we've just
been exceptionally distracted of late: ICANN pales in comparison to the
new crop of acronyms—MPAA, RIAA, DMCA, TIA, USA PATRIOT—menacing us
But perhaps these trends obscure a deeper shift. At the beginning of
the boom, the vast quantity of people and organizations online
outstripped our ability to find them, and we pressed the DNS into
service to help fill that gap. But this usage of the centralized,
permanent DNS conflicted with the common-sense methods that people use
to name things in their everyday lives, and as the internet continues
to decentralize this dissonance only grows stronger. The conflict is
being alleviated not by technical or political reform at the center of
the network, but by innovation at its edges. As end-user applications
mature, they increasingly allow individuals to develop and share their
own naming systems—not to destroy the DNS, but to render it irrelevant.
JUST ANOTHER PYRAMID SCHEME?
The reasons that the DNS started to crumble under the pressure of
commercialization have already been well documented. Writing in 1998,
Ted Byfield noted that the DNS was never designed for that pressure in
the first place:
DNS was built around the structurally conservative assumptions of a
particular social stratum: government agencies, the military,
universities, and their hybrid organizations—in other words,
hierarchical institutions subject to little or no competition. These
assumptions were built into DNS in theory, and they guide domain-name
policy in practice to this day—even though the commercialization of the
Net has turned many if not most of these assumptions upside down.
One of the assumptions Byfield is referring to is the notion that name
collisions could be greatly reduced by dividing the namespace into
top-level domains and trusting that everybody would calmly accept their
place in that hierarchy. But as domain names became associated with
trademarks, common usage flattened these tidy divisions into one
undifferentiated sprawl. Corporations saw the web as one more front in
the battle for marketing and public relations—like television, only
with a keyboard—and accordingly they didn't care much for quaint rules
written by computer scientists. So when, say, Archie Comics sued a
California man for registering veronica.org in honor of his daughter,
arguments that the .org TLD didn't belong to companies fell on deaf
ears. (Online protest eventually succeeded where quoting RFCs had
failed, although today veronica.org redirects to SamsDirect.)
Many reformers aimed for a political solution, appealing to ICANN to
keep the DNS safe for bit players. They felt that in kowtowing to the
corporations, ICANN was bastardizing the simplicity of the system that
Jon Postel had managed until handing it off in 1998. (When the final
draft proposal for ICANN was finished, Wired called Postel "the
Internet's own Obi-Wan Kenobi"—a phrase that would attain an eerie
resonance with the Slashdot crowd when Postel passed away a month later
and ICANN revealed itself to be a bit of an Evil Empire.)
But would even a perfectly managed DNS have functioned in accordance
with its earlier hierarchical vision? The hierarchy made sense to the
users of the early internet, but the noisier commercialized internet
would have fit much less comfortably into such a scheme. Even if you
could've assigned every person, place, and thing its proper slot, most
people would not have bothered to learn what went where.
Take, for example, the .museum TLD, which has been open since 2001.
Most prominent museums have avoided using this TLD; the Whitney, the
Guggenheim, and the Museum of Modern Art all place their primary domain
names under .org. Conceptually, .museum muddies the waters because it's
not mutually exclusive with .org. And when it comes to marketing,
.museum is a disaster since it only serves to distract the user—who
ever heard of a six-letter TLD?—without making her life any easier.
Or to take a more high-profile example, look at the recent lawsuit that
forced the World Wrestling Federation to change its name to World
Wrestling Entertainment. The World Wildlife Fund had sued the
Federation for breaking the terms of a 1994 contract dictating who
could use the initials WWF, in what media, and how prominently. Now,
strictly speaking this wasn't solely an issue of domain names. The
Fund's spokeswoman attributed the suit to an "explosion" of the
acronym's use in three media: online, satellite TV, and cable TV. But
one of the major grievances in the Fund's suit was the Federation's
registration of the domain name www.wwf.com in 1997.
According to the neighborly rules of the pre-boom internet, this should
not have been a problem: The Federation got wwf.com and the Fund got
wwf.org. This solution works if you care about those tidy hierarchical
divisions. Most people don't. This is one of the reasons that the new
TLDs have been so underwhelming: People don't see the world as cleanly
divided into discrete categories, with the corporations in this corner
and the non-profits in that corner. It's all one namespace to them.
Mutually exclusive hierarchies are convenient, but they only work on a
small group of items. Once that group gets too big and diverse—a comic
book artist here, an airplane-parts manufacturer there—any hierarchy
that might reasonably hold that group becomes too cumbersome for people
to use. When people want to organize large groups of items they often
find it easier to use overlapping sets instead. That's why filesystems
have symlinks. That's why many of Apple's OS X programs, such as iTunes
and Address Book, let you drag-and-drop your MP3s or contacts into as
many groups as you want. Why bother fretting over whether you should
put Christine from work in your Friends group or your Coworkers group?
Put her in both and get on with your life.
SIX BILLION NAMING SYSTEMS AND JUST ONE INTERNET
A hierarchical, precise DNS is a perfect system for computers. Human
beings, however, prefer to rely on systems that make use of their own
technical strengths, such as the ability to adapt their language to the
preconceptions of your audience, and the ability to adapt their own
conception of the world to accommodate new knowledge. Common sense, in
other words. If, in the days when World Wrestling Entertainment was
still a federation, you used the initials WWF in a conversation,
chances are your listener would be able to figure out which one you
were referring to. Humans do this by drawing on the context of the
conversation to make the correct match. Are you talking about panda
bears, or Stone Cold Steve Austin?
In real life, people have almost no problem resolving name collisions—a
good thing, considering how often they happen. There are two types of
Dove bars you can buy in a supermarket: One is chocolate and the other
is soap. There are three famous Dres in hip-hop: Dr. Dre of NWA, Dre of
Outkast, and Dre of Dre and Ed Lover. Hip-hop fans know how to tell
It happens on a personal level, too. In my freshman year of college, my
dorm floor had four Mikes and four Daves. We resolved these name
collisions by settling on nicknames for everybody: Big Dave, Sophomore
Dave, Asshole Dave, etc. People who lived outside our floor didn't know
who was who, but they didn't need that system anyway. We did, and it
worked fine for us.
What we didn't do, however, was make use of last names, even though
they offer a more global, permanent method of differentiation. Last
names were less memorable to us than the jokey, college-guy
associations we could invent on our own. Clay Shirky wrote that the
aims of the DNS are to be memorable, global, and non-political. "Pick
two", he said, but in fact most of the time people only care about the
first: As long as names are memorable, people don't mind that they're
local and highly subjective. Techies are an exception, since they spend
much of their time crafting language for machines, and as such are
accustomed to treating language as a brittle, precise tool. But most
people like their language loose and contextual, thank you very much,
and the hierarchies of the DNS demanded a rigor that never seemed worth
the trouble to them.
THE MNEMONIC AND THE MEANINGFUL
Another source of pressure on the DNS was the system's shifting role
from one that was primarily mnemonic to one that was meaningful as
well. The difference is subtle, but important. Consider the phrase
"Every good boy deserves fudge", which music students sometimes learn
to help them memorize what notes correspond to the lines of the treble
clef. The phrase is helpful, but its content—boys deserving fudge—has
nothing to do with music. It's mnemonic, but not meaningful.
The two can co-exist, and originally the DNS was a mix of both. A
domain name like gandalf.cs.columbia.edu could give you important
information—namely, that this domain is administered by somebody in
Columbia's computer science department—but then, what does this domain
do? Is it a mail server? A MUD? Knowing that somebody in Columbia's CS
department likes Lord of the Rings is almost redundant.
Originally the purpose of a domain name was to be an address that was
easier to remember than an IP address. This changed during the boom, as
users and companies developed the notion that the function of a domain
names was to serve as a self-explanatory pointer to a discrete
real-life entity—a writer, perhaps, or a corporation or a museum or a
hacker's collective. Of course, this was never fully realized, and it
made little sense if you weren't swinging for the big leagues of global
name recognition. If your site was niche enough that you could make use
of an odd URL like http://c2.com/cgi/wiki (the first wiki, hosted on
Ward Cunningham's web server), then those awful domain name disputes
were somebody else's problem.
Today, internet services are becoming cheaper, more specialized, and
easier to use, with the result that every day more people and
organizations create a persistent online presence. And as the internet
takes shape as—to borrow David Weinberger's phrase—small pieces loosely
joined, the use of the DNS as a meaningful system is in further decline.
In the commercial world, companies ranging from small retailers to
leading credit card providers use third-party services to manage online
bill payment or e-commerce checkout. In doing so, they happily give up
part or all of their domain-name branding in exchange for technical
Among individuals, of course, the most significant relevant trend is
blogging. By some estimates there are already more than a million
bloggers, and Lord only knows what those numbers will be like after AOL
rolls out its blogging product later this year. Many of these bloggers
don't have their own domain names. Instead, they're contained in
subdomains (http://jwz.livejournal.com/), directories
(http://weblogs.mozillazine.org/hyatt/), and CGI arguments
IF YOU CAN CLICK ON IT, IT'S SOFTWARE
The meaningful DNS simply can't cope with a world of, say, 10 million
bloggers, but luckily we have other ways to make sense of the internet.
In the last five years, we've gained a number of powerful navigational
tools, and these allow the DNS to pull back to a less high-profile
role. The most obvious example is Google, which has done more than any
other dot-com to make it easy to find your way around the internet.
But Google is still a centralized service, and as such there are limits
to how much it can help. There is more promise at the edges of the
network, where end-user software makes it easier for individuals to
name, remember, and share URLs. Some preliminary examples:
* Subscribe to a blog's feed in your RSS aggregator and you might
never have to type that blog's URL again.
* Blogging tools decrease the amount of manual work that bloggers
have to do to pass links along. The beta version of Google's browser
toolbar even has a BlogThis button.
* Apple's web browser Safari integrates with its Address Book to
automatically bookmark the websites of your contacts.
* Almost all email clients and chat clients will automatically turn
URLs into clickable links, relieving you of the need to even
None of these innovations are groundbreaking. But taken together they
add up to an environment where users delegate to computers the dirty
work of handling URLs. Consider, for perspective, this 1999 article by
usability author Joe Clark:
A long URL works poorly in stationery, in articles in the print medium,
and in advertising (e.g., on TV, with its low resolution and the short
time a URL can be shown, or on radio, where it must be read out loud),
and is a usability disaster in one-to-one conversations. (That's
conversations, as in voice, as in getting together or talking on the
Today, this is still sensible advice if you're the webmaster for a
Fortune 500 company or a popular dot-com. But for an increasing number
of people, keeping URLs short isn't as important as it used to be. To
take myself as an example: I'm relatively tech-savvy, and I create some
sort of online content for a small, tech-savvy audience, so a short URL
is much less important to me than it was only four years ago.
Stationery? I write in my Handspring Visor more than I write on paper.
Print articles? My site isn't mass-market enough, and my target
audience probably reads most of its news online anyhow. Television
advertising? I'd love to have that problem.
And what about communicating a URL through speech? Personally, I find
that this happens much less often than it used to. Maybe somebody will
speak a URL out loud if she's referring to an easy domain name like
half.com. But just as often she'll offer to email you that link, or IM
A URL can be both text and a software component. You can write it out
longhand, but if you put it in an email client or a chat client, it's
as much a software function as the Undo command: Click on it, and your
computer responds. It's functionality that can be serialized into text
if that makes it easier to transmit. And if you and your social circle
are never far from computers with persistent broadband connections,
then it's simpler to treat that URL as functionality rather than text:
Rather than spell it out over the phone, email it or IM it.
Not that you should go making your URLs 400 characters long now.
Shorter URLs are still better, or else why would we have those services
that let you create a short URL to redirect to a longer URL of your
choice? Notice, however, that the main purpose of these services is to
facilitate the machine transfer of URLs, since some email clients get
confused when handling a long URL.
In fact, many of these services making the URL itself less meaningful,
since they don't let you choose which key to assign to your long URL.
Is http://tinyurl.com/6a2 a map to your friend's party? That PDA your
girlfriend is considering buying? The CIA World Factbook's entry on
Afghanistan? These short URLs—and, of course, the domain names they
contain—tell you absolutely nothing about what they point to. You'll
have to rely on context to figure that out. Your friend writes you an
email, says "Here's that restaurant where we're meeting for dinner on
Thursday," and includes a short URL below. The URL itself means
nothing. It takes its entire meaning from the conversation it's
MERE USER CUSTOMIZATION IS LOOSED UPON THE WORLD
If the DNS is fading in importance, it won't be a surprise to
everybody. Byfield, for one, wrote that "DNS's level of abstraction is
sinking relative to its surroundings." A year later, in 1999, Jakob
Nielsen predicted the same, and with pretty good timing to boot.
It is likely that domain names only have 3-5 years left as a major way
of finding sites on the Web. In the long term, it is not appropriate to
require unique words to identify every single entity in the world.
That's not how human language works.
Today, in 2003, this is what the future of the domain name looks like:
For the major players, the system will remain more or less unchanged.
There will always be a small cast of large organizations and companies
who will have domain names with household recognition: ebay.com,
But for the rest of us, we can increasingly rely on the fact that
software is allowing users to build their own naming systems around
their desktops, and then sharing and cross-pollinating those systems
within their social circle. If you use the OS X Address Book, you can
browse through your Safari bookmarks to find the link to, say, David
Johnson's website. Which David Johnson? The one you care about.
So as decentralization continues, we can largely ignore the frustrating
world of the DNS and focus our efforts on other ways to make
connections. We can work on establishing our own roles in communities
that are intimate and deep, not broad and shallow. And we can think
less about marketing, and get back to just communicating.