In the everyday world of streets and buildings, addressing isn't much of an issue. In the US, for example, each "block" is given a range of 100 addresses. Each building (or set of buildings) is given a unique number.
Blocks almost never exceed 1000 feet in length, so there is little chance of running out of numbers. In any case, further subdivisions are possible. Apartment and suite numbers, for instance, handle divisions within individual properties.
In the virtual worlds of telephone numbers and domain names, things aren't quite so simple. There aren't any hard limits on the number of electronic locations that can emerge in a given part of the name space. Consequently, we occasionally see signs of overcrowding.
Telephone companies are all too aware of this issue. US telephone numbers are limited to seven digits. This is a raw capacity of 10 million numbers per (three digit) area code. Unfortunately, things are not quite that simple.
Because telephone numbers are not allowed to start with either zero or one, the number drops to eight million. Add in factors like geographical quantization, forced delays in recycling numbers, and reserved prefixes (e.g, 411), and the actual capacity could well drop to 75% of the raw figure.
At the same time phone companies are being hit by an avalanche of demand. People want cellular phones, fax machines, modems, pagers, and more. Making matters worse, it is a common practice for businesses to reserve blocks of numbers in advance.
These factors can multiply the average number utilization by a rather large factor. Between my spouse, my one-horse publishing business, and myself, over a dozen telephone numbers are being used. If the average number utilization hits seven numbers per person, a city of a million would need an area code aoo to itself. The new area codes that keep springing up are a reflection of these sorts of pressure.
The IP (Internet Protocol) address space has also been showing signs of strain. CIDR (Classless Inter-Domain Routing) has replaced the old [ABC] classes, allowing much finer granularity, and hence efficiency, of allocation. A mixed-length (32/64 bit) scheme is also being developed, and will be deployed over the next few years.
It would seem that Domain Name Service (DNS) would not be subject to the same limitations as these numeric addressing schemes. Domain name labels (e.g., ptf) can have up to 63 characters (letters, numbers, or dashes).
In theory, this lets us create about 6.4e98 (37^63 + 37^62 + ...) domain names in COM alone. Unlike telephone numbers and IP addresses, however, domain names are supposed to be memorable. A name like w3th7p24cd.com may be legal, but I can't imagine using it on a business card.
Even with the constraints of memorability and pronounceability, however, there isn't any real name space problem. Between surnames (abernethy), words (acetaminophen), portmanteau constructions (adventureseekers), hyphenated expressions (build-outs), and letter-number combinations (4h), there are many millions of available domain names. This is probably a good thing, because some folks are grabbing names at an amazing rate.
All of the names listed in the last paragraph, along with several thousand others, have been registered in the COM domain by a single firm: Free View Listings. (An Info-Zipped list of the names can be retrieved from ftp://ftp.mindspring.com/users/centauri/freeview.zip.)
I was curious as to Free View's reasons for grabbing all these names. The current $100 registration fee makes this sort of thing a bit pricy. Looking at Free View's web site (http://www.freeview.com), I was directed to a site for MailBank (http://www.mailbank.com).
The text I found there is pretty self-explanatory:
Claim your name ~ Stake your claim.
MailBank is an e-mail service. We lease fantastic e-mail addresses for only US$ 4.95 per year.
We have thousands of domain names to choose from.
MailBank will either host your mail box on its servers, or it will automatically forward your mail to any other internet mail box. This is under your control and can be changed at anytime, as often as you like, without cost.
I was relieved to find out that Free View isn't holding domain names hostage for future resale, but I still find their actions disturbing. AOL and other ISPs register many names for their clients, but they only do so on request. Free View, in contrast, has grabbed thousands of names purely on speculation.
The good news with Free View's scheme is efficiency: Fred and Freida Frobozz can share frobozz.com with Phil, Phyllis, and hundreds of other Frobozz family members. Actually, Free View had better hope that lots of Frobozzes sign up; At $5 each, it will take an average of 10 users per name just to pay the annual domain name registration fee!.
The bad news is that Phil Frobozz can't register frobozz.com as a domain, even if he's the first Frobozz to want it. Worse, he can only get an email address on frobozz.com by sending money to Free View. The cost ($5/year) isn't excessive and the service is certainly useful. I just wonder if this sort of speculation is the sort of thing we want to encourage on the Internet.
I'm also curious why Free View thinks it's worth nearly a million dollars to glom onto all of these domain names. Is frobozz.com all that much more valuable than, say, frobozz.mailbank.com? Finally, there is the question of netiquette: the latter approach is in tune with the cooperative spirit of the Internet; Free View's approach isn't.
Free View's activities aside, the current set of domains needs a bit of expansion. COM is a fine category for Apple, PTF or even CFCL, but it isn't really approprate for personal use. I'd like to see some categories such as IND(ividual), FAM(ily), or PER(sonal). The net.gods are charging us $50/year for each domain name; they might consider blowing some dust out of the system...