I agree with Tim Wooller's comments. In reviewing a paper for another IEEE 
archival journal, I realised that 43 percent of the citations were solely 
URLs. The overuse of URL citations lowered my confidence in it, however 
good it was otherwise.

When I checked the cited URLs, 40 percent were erroneous, one bounced me to 
three further addresses, each time saying "Sorry, not here now." In many 
cases, I couldn't find the other URLs at the specified addresses, and had 
to look elsewhere. When I did find them, information regarding the subject 
was indeed on the site, but I asked myself whether this was the information 
I was supposed to read.

I understand --- and believe in --- the use and importance of the Web. In fact, 
when doing research, the first place I search is the Web. I use the Web, as 
many people do, as a world-wide encyclopedia. But I believe that I am only 
receiving about 30 percent of the total amount of information on the 
subject, because I feel that people are not adding their latest research to 
the Web.

The Web is a useful resource, but not a complete or accurate library. Thus, 
I cite URLs only tentatively. It is a question of amount: the more URLs in 
the reference list the less I trust the paper and the lower its archival 
relevance. It is useful (and unavoidable at times) to cite some URLs. I 
know, because I do cite them. But there needs to be a limit on the amount 
of changeable --- or in-the-future unavailable --- citations.

Referencing appropriate information is important when writing a scientific 
paper, as references support, develop, and validate the written argument. 
On my bookshelf I have a copy of David Lindsay's A Guide to Scientific 
Writing (Longman, 1995), which rightly warns against referencing too many 
university publications, personal communications, and sources that you 
cannot easily locate. URLs can be easily located, but, as Tim Wooller 
pointed out, the information may be easily altered or removed completely 
from the site.

Computers and the Web are rapidly changing. The printing press moved us 
from hand-copying manuscripts to mass-copying them. As publishing companies 
devote more resources to publishing on the Web, we will see (and need) more 
URL citations in our journals. Our confidence in citing URLs will increase. 
But for the moment we need to hold back on using too many URL citations.