Saturday, August 21, 2004

What Is A Webalizer Referrer? -- The Deleted Post...

In an earlier post: [ Web Counters] I discussed ongoing issues with my logging. In the last paragraph of this post I discussed referrers. The following is the definition of a referrer as provided by this link -- [Webalizer Quick Help]:

Referrers are those URLs that lead a user to your site or caused the browser to request something from your server. The vast majority of requests are made from your own URLs, since most HTML pages contain links to other objects such as graphics files. If one of your HTML pages contains links to 10 graphic images, then each request for the HTML page will produce 10 more hits with the referrer specified as the URL of your own HTML page.

The above implies that graphics files and other "objects" are counted as part of a url's referrer tally. I remember reading on two separate occasions, information that specifically claimed the referrer was a "page" and not graphics or other requests. Only pages. Now I cannot find that reference and I am wondering if anyone else read that claim.

Webalizer reports a "direct request" under the Referrers section of the report -- for those pages accessed via the Java applet menu I use on the main pages of the site. [I ran a separate test to confirm this]. This applet menu is "graphical" in appearance ...

Because I have ongoing issues with logging I am trying to gather all the evidence that supports the claim my server stats are not correct.

As of August 19, 2004, the blog [yes this blog] has made over 6800+ referrals [referrers] to my server -- for the month of August and it has only been requested 1350+ times for the same time period. If Webalizer [report provided by my server] is counting the external objects that make up the index page of the blog then this figure is still out of whack because most of the graphics found on my blog are from the server. The only external files on my server are the search engine script, my QiSoftware logo, and my photo [other than the graphics I recently started adding to my posts -- discussion later -- not even the Wired Pages logo is not located on the domain]. The blogger profile has a photo too -- but the referrer in question is "" and not a url.

In addition, I do not have enough bandwidth coming off my server for the same time period to account for the number of graphics and other requests that the 6800+ referrers would suggest -- and in any case most of the graphics are not located on the domain. My bandwidth usage thus far is only 110 Megabytes --[through August 19, 2004]. As I stated -- I recently started adding graphics to the blog entries and the ratio of referrers to hits has always been about the same. These recent graphic additions to my posts are not enough to account for the ratio difference.

The really interesting "object" to study is the cgi server counter used on the blog. It is (1) a top url hit getter (2) a referrer and (3) part of the overall number of items referred to by the blog url [this according to the poorly worded definition above]? Why would Webalizer a program used by many -- count this object so many ways? This is illogical might even be illegal. My cgi server counter does it all... Can it walk on water?

One other tidbit -- the Wired Pages home page also uses a cgi counter. This url has 283 referrers thus far for the month of August -- [thru 8/19/2004]. It also has three programs that are served up from the domain -- each time this page is accessed 11 hits [this is right -- only ten if accessed directly] should be logged to my raw access files. Let's see -- 283 divided by 11 is only 25 times this page would have been accessed according to the above definition [wording a little shaky] of referrer. My webalizer reports that this subdomain [] -- which is an index redirection -- has been accessed almost 200 times this month [thru 8/19/04].

If I am right [about the definition of referrer], why is the ratio off -- my links do not open in new windows. I always open links in the same window -- and I still do not have enough logging to accomodate the news service page requests. Most of the links on my blog are links to my news service which only have a redirect index file on my qisoftware domain server. This would correlate with the low bandwidth. The news service programs [applications maintained on the qisoftware server] have very small foot prints [bandwidth wise] -- but are not being logged. This was the first problem I noticed --

Something is really out of sync... Anyone else have a similar problem? In my last post [recursive reference here -- something that cgi counter should not be able to do --- ], I showed why the only other way to account for the ratio [6800:1350] difference between referrers and requests is not applicable in this case.

By the way I never use redirection to access my news service pages -- I access the pages directly in hopes of keeping my logging [ domain] to a minimum. So although I may account for some of the hits to the blog, I am not clicking on links found on the blog -- thereby causing no change in the referrer count total. The only logging that should occur when anyone accesses my news service pages directly -- is program and script usage for those applications maintained by the domain and referenced by a page on my news service. [I use a lot of server side programs and AOL does not support server side applications and scripts on its platform.]