Re: [wg-c] .com vs. ccTLDs in Europe

[Dave Crocker <dcrocker@brandenburg.com>]

At 10:51 AM 10/20/1999 , Jonathan Weinberg wrote:
>area in which my lack of technical expertise is painfully obvious, so I'm
>...
>were 44,386 domains in .fr as of 8/99.  This seems to me to indicate that
>.com plays an important role in Europe, but not an overwhelming one.  Can
>somebody please let me know if I'm understanding these statistics
>incorrectly?

An example of what makes this sort of analysis difficult is the difference 
between measuring a snapshot of current allocations, versus 
"trends".  Current allocations carries all the baggage of past allocations.

Trend analysis shows the rate of growth and is a far better indicator of 
such things as adoption rate and currently-perceived benefit.

A few years ago, there were essentially no .coms to be seen in Europe or 
Asia.  Now, billboard advertisements that cite a .com address are common in 
Bangkok, Borneo, Hong Kong, Brussels, London, Milan, ...

In other words, you are quite right to be very cautious about this sort of 
analysis.  It really does take a statistical methodology expert to develop 
meaningful and accurate conclusions.  (It is not enough to know the 
algorithms; one must know the efficacies and limitations of their use 
within the real world.  Hence the "methodology" requirement.)

Many years ago, a researcher developing easy-to-use statistical packages 
for health researchers reported that she had gotten some very strong 
criticism from her users because she added a restriction to the correlation 
package.  The package required that there be at least two data points...

d/

=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=
Dave Crocker                 675 Spruce Drive        Tel: +1.408.246.8253
Brandenburg Consulting    Sunnyvale, CA 94086 USA    Fax: +1.408.273.6464
<mailto:dcrocker@brandenburg.com>            <http://www.brandenburg.com>