About Top Level Domain Longitudinal Statistics

Just for fun, I decided to see how well I was doing with "Web DXCC". That is, World Wide Web users from how many distinct "top-level domains" (TLD) have visited the NG3K Web pages.

The TLD is the right-most segment (segments are separated by dots) in an Internet address. Most TLDs correspond to a country. For example, in a Web server logfile address like this:

  nhm068.dokidoki.ne.jp

the "top-level domain" is ".jp" which corresponds to "Japan". There's a one-to-one correspondence between many TLDs and a DXCC entity.

Of course, some TLDs (.com, .edu, .gov, .mil, .org, etc) do not correspond to a DXCC entity, but on the other hand some TLDs (.uk and .ru, for example) correspond to more than a single DXCC entity. Also, addresses can be spoofed. So just because there's an Oman entry in the logfile doesn't tell you with certainty that someone from Oman actually visited your site. To complicate things even more, 20% or so of all requests received each month do not resolve to a named address and their geographic source cannot be easily determined. This means that one may have received visits from more TLDs than are reflected in the statistics. So the correlation between TLDs and DXCC entities is pretty loose.

Keeping those limitations in mind, I wrote a Perl program to go through my monthly and daily statistical reports and pull out/calculate the information I needed to see how I was doing.

If you're interested, have a look at:

  http://www.ng3k.com/stat/index.html

and you'll see how my inquiry came out. You'll find an index of all TLD that have visited (with links to monthly statistics for each), the total number of TLDs that have visited to date, and a list of the 15 domains most recently "visiting for the first time". Next to each entry is the (year/month) of the entity's first visit. The data start with June 2000 (200006), so most TLDs will show as having made their first visit that month.

Each top-level domain in the list is linked to a corresponding Web page that provides monthly access statistics for that domain. These "linked-to" files are generated by a separate Perl program, not the one under discussion here.

The page is updated once a day to include data from the previous day.

As a matter of possible interest, Analog (the statistics software I use to generate my monthly statistical reports from the source logfiles) currently recognizes 255 TLDs. So I've got a way to go to reach the "honor roll".

Randy, W5UE, WebMaster for the Magnolia DX Association has also created a page based on my Perl code. Visit his implementation too.

Bill/NG3K
March 2001