Converging Counters...

As some of you may have noticed I have two counters on this weblog. The upper counter [server counter] was discussed in an earlier post. The lower counter was placed on this weblog shortly after the first to give me an external reference. At the time the second counter was inserted it was set to the value of the first. These two counters quickly became out of sync, with the upper counter taking the lead...

On Saturday, August 7, 2004 with the start of the day, the upper counter registered 23 points higher than the lower counter. [This difference in the counter readings was a gradual evolution taking place over a period of just under 3 weeks.] By the end of the day the counters were 18 points apart. The convergence had begun.

On Sunday, August 8, 2004 I had a brunch date. Before I left, I watched as the lower counter ticked off about 5 clicks, without any movement from the upper counter. I know my server counter was not incremented because nothing was being logged. I confirmed this by bringing up this weblog and yes the counters were converging. The lower counter is a "freebee" provided by an external source. I can access a separate stats page for detailed information [this is how I know the counter was being incremented]. The upper counter is a server counter and is logged every time it is activated. Upon my return there was only a 2 point difference remaining of the once diverging counter readings with the upper counter still in the lead. This is where the count difference stands as of this posting.

Web page counters can be notoriously unreliable, however I have had the lower counter on this page since the 20th of July and this is the first time I have noticed this anomaly. The lower counter erased 16 points of an 18 point difference between the two counters with absolutely no response from my server in a matter of hours. Was this external counter somehow being incremented by an outside force.. I don't know... All I know is that I was not refreshing the page. In the past, I have watched as both counters have recorded 5 hits in short order with no problems.

The only other time I notice counter increment issues is when I am updating this weblog. I perform quick edits and the counters [oddly both] will stop responding as I continually refresh the page to see my changes. This may be in part, due to the browser that I use to verify these updates. The counters do not always suspend during these edit sessions. I cannot reset the external counter [lower counter]. I can reset the upper counter and did so on one occasion after one of these editing sessions. I knocked off two hits in hopes of accurately reflecting the number of web surfers viewing the page. I stopped this practice, because as I explained in the earlier post, the methodology used in arriving at the initial counter setting, was by means of a very conservative approach.

Can my server become so busy that it fails to log server hits and increment counters? I have asked my web hosting service about this and they claim everything is being logged. Periodic tests reveal the absence of logging at times [this was not one of my tests]. If ranking services [e.g. Alexa] were to rely, in part, on logging data for rankings how would they be able to get accurate results with the somewhat faulty logging issues outlined here? In my case this happens to often for me to ignore it...