Theoretical Mathematics - Is there a Law relating Views and Replies?
Dear Covers:
Thanks to the warm welcome that I have received here, since I swam ashore on July 28, 2004 from a stormy sea, and the encouragement given by Craig, I am taking the plunge now with this submission. In my humble opinion, and quite coincidentally, this compliments what you will find in Steve Prevette's August 2004 article. I look forward to your comments. With my charmiest regards.
Charmed
Last edited by Charmed; 30th July 2004 at 10:42 PM.
Reason: spelling
A law is something that is always true. So, if I am right, and if we continue to make "observations" such as what I have done, we always find the relation, y = hx + c, over and over again. For this to be called a universal law, the same (or very nearly the same) value of h must apply at all times. Changes in the values of the constant c therefore explain why we get more or less replies, for the same number of views.
In this case, we are dealing with a process called transfer of "mental energy" from the poster to the viewer, who then chooses to reply or not to reply. Of course, I have been around for just a couple days. So, observations have been limited. But, it might be fun to see what this might reveal over a longer period of time. More important is the extension of this approach to similar data from other websites, and to the analysis of many other types of transactional data that we gather daily. Anyway, thanks for starting this off. I feel welcome, for sure. With my charmiest regards.
Now that I've looked at your papers, I'm really mad... (Tee hee hee!)
I browsed your two papers and that you have evaluated the numbers upsets me - That I didn't look at the relationship myself earlier on... So - My Kudos!
I will inject 1 aspect of the equation to consider... Views are measured by a 'page viewed' number which is a raw number. What I mean by this is that a 'view' will be counted every time the page is 'loaded' by a viewer. In addition, let's say I'm the thread author or a reply post author - every time I edit the thread or post another 'view' is chalked up by the software. So - If I 'subscribe' to the thread, for example, every time I come to see what others have entered another 'count' is added. If I started the thread, I may come back and view it multiple times and each time registers as a 'view'.
I'm only pointing this out so it is understood that there are aspects of the software which have to be accounted for. The failure mode is that the stats software is not counting individual viewers, but rather is a cume number.
But please keep on - I am anxious to see any analysis of the numbers. If you are not familiar with web site analysis - specifically with respect to vBulletin and this 'hack' stats program, let me know and I'll try to answer.
Oh - One off the cuff comment: Copyright 2004/ vlaxmanan/ Charmed/ ReplyView1.doc -- Nothing posted in these forums is copyright. This is an Open Source forum. I do not have any rights to posts or their contents nor does the poster.
I am not familiar with website analysis. Would love to discuss this further with you. Do send me an email.
Right now, I am just taking the "numbers" at face value. I agree with you about the "phony" nature of the number called "views". Perhaps, "replies" may be more representative. If we can "weed" out some of the uncertainty that you point out with regard to the "views" we can develop a more accurate analysis. I believe "views" can be related to the I. P. adddress of the computer and if the I. P. address is the same, it means the same person is "viewing" more than once. But, then, it might just be that his/her spouse, or children needed attention, and he/she came back - with genuine and bubbling - to finish reading the marvelous article posted on the Forum. With my charmiest regards.
Now that I've looked at your papers, I'm really mad... (Tee hee hee!)
I browsed your two papers and that you have evaluated the numbers upsets me - That I didn't look at the relationship myself earlier on... So - My Kudos!
I will inject 1 aspect of the equation to consider... Views are measured by a 'page viewed' number which is a raw number. What I mean by this is that a 'view' will be counted every time the page is 'loaded' by a viewer. In addition, let's say I'm the thread author or a reply post author - every time I edit the thread or post another 'view' is chalked up by the software. So - If I 'subscribe' to the thread, for example, every time I come to see what others have entered another 'count' is added. If I started the thread, I may come back and view it multiple times and each time registers as a 'view'.
I'm only pointing this out so it is understood that there are aspects of the software which have to be accounted for. The failure mode is that the stats software is not counting individual viewers, but rather is a cume number.
But please keep on - I am anxious to see any analysis of the numbers. If you are not familiar with web site analysis - specifically with respect to vBulletin and this 'hack' stats program, let me know and I'll try to answer.
Oh - One off the cuff comment: Copyright 2004/ vlaxmanan/ Charmed/ ReplyView1.doc -- Nothing posted in these forums is copyright. This is an Open Source forum. I do not have any rights to posts or their contents nor does the poster.
Yes, Marc. This is all public domain now. I would be very happy if you or anyone else finds this to be useful and cares to use it.
I got your PM. If there is something specific you want to discuss with regard to web site anaylsis I'l be happy to comment, but you mifght want to start a thread in the Forum Help forum.
I'm still in the swollen jaw category from dental work yesterday but the Percet has almost worn off now so I'll try to comment on this aspect of analysis.
First off, some of the numbers here are valid, others are 'reference'. That is, some can give you absolute numbers which should be 'good' (such as 'how many total threads' or 'how many total posts'). A 'reference' number is what I would call 'views'. It does tell you how many times the thread was 'called', but there are a lot of cavats. For example, if it is a long thread a 'view' doesn't mean every page of the thread was loaded and read by someone. The count will increase by 1 even if only 1 page of the thread is loaded. As I think you pointed out, if a person reads a thread and re-reads it later it will count as 2 'views' even though it's 1 person and even if there are no new posts to that thread. These are just a couple of things, but I think you get the idea.
A lot of people have asked the vB people to address a lot of different aspects of the different statistics - And other aspects of the software in general for that matter. Database size issues often come into play. Often what is wanted is not impossible, but is more of an issue of limitations - If, for example, a complex tracking routine was coded where every 'partial thread view' was recorded, and other aspects such as who viewed the thread and how many times, the arguement is that the database could become so large that many people would have trouble. A secondary factor cited is speed. If the software stores all these aspects, how long will it take especially when there are calls for that data? Another aspect is views by unregistered visitors. For example, AOL uses proxies and the way it does so often gives a different IP for each 'call' from a person. So - when you think about tracking unregistered users by IP the model fails. This is also an issue in dialup. If you log in today and read a thread and then you dial in again tonight and read the thread again, you will have a different IP. That said we cannot us the IP as an indicator of whether, or how many times, an unregistered (or registered but not logged in) individual user reads a thread.
I ran into both these issues (database size and board speed) on the servers I had before last December (these forums have been on several different servers over the years). I ended up switching to a dedicated server that is 'oversized' for the amount of traffic here and the database size.
In short, with some of the numbers here one has to understand the aspects which will throw off any serious analysis attempt. My problem in trying to address these issues in a post is difficult because I could write a book - You pretty much have to sit down and discuss these things or you end up with 'snippet' failures in understanding. That is, someone asks a question and gets a partial answer because there are so many aspects that it would take hours to write out and address each one where face to face much, much more information can be related. From a 'snippet' here a lot of assumptions can be made and many of them may not be at all valid.
You have thought of and mentioned so many points that are worth bearing in mind. Wow, all of this even after dental surgery!
At this point, I will simply wait for more input. As you can tell, I only have the numbers for views and replies. If anyone knows how to "improve" the numbers being used, I would certainly be very glad to take a look at this whole thing again.
However, something tells me the basic conclusions will not change significantly. The method of analyzing the data remains the same. All that changes is the numerical values of h and c in the law y = hx + c. The law, not the exact numerical values of h and c, is what I was trying to call attention to. We seem to find this simple law in the most unexpected of places.
As we proceed, and assuming this catches on, we could analyze many other types of "information", similar to replies and views here, that is now being captured by other websites to asses their own performance. We could apply it other types of manufacturing and service industry data.
A more scary thought, which I had mentioned in my PM, is the potential application for "intelligence" data. What would we learn from traffic on websites being monitored by intelligence agencies for national security purposes?
May be the idea of a "work function", which is simply another way of looking at the nonzero c in y = hx + c, is telling us more than we realize at this time. Since I had reason to expect a linear law (based on prior experience with many other types of business, social, and economic data analysis), I tried to see what was going on at Elsmar Cove Forums, using the same model.
Again, Marc, please accept my sincere thanks for the thought that you have given this first effort on my part.
Charmed
P. S. I say "scary" thought because the 9/11 commission has concluded that we (the U. S.) as a nation has lacked "imagination" in the analysis of intelligence data. What is intelligence data? This too can be converted to a bunch of "numbers" that can be analyzed using similar methods, statistical or otherwise. I am thinking of a model similar to that used to develop various "quality" ratings. I have analyzed the J. D. Power Quality data (Vehicle Dependability Study and Initial Quality Study) using the work function described here. That's a different matter altogher.
Last edited by Charmed; 31st July 2004 at 06:51 PM.
Reason: Add a comment
Theoretical Mathematics - Is there a Law relating Views and Replies?
What is a law?
Websote Analysis
This is public domain now
Linear Mode
