CGC in the Sands of Time by Kaustubh Adhikari

Download the pdf version of the article: data-analysis_kaustubh_edited-final_bg

“There are three kinds of lies: Lies, Damn Lies, and Statistics” (Often attributed to Mark Twain). So, this time I’ll refrain from telling too much lies (i.e. writing too much). Instead, let us have less of numbers, plots and tables compared to the mammoth report of last year, and indulge more in talking.

“A man may have six meals one day and none the next, making an average of three meals per day, but that is not a good way to live” (Louis D. Brandeis). Yes, the last year has brought some ups-and-downs to CGC. Amidst the varied reactions of the individual members, we can have a look how the combined behavior of members has been.

There are two components of any forum – one, the incoming lot of people who become members from visitors, and secondly, the current lot of members who indulge in all sorts of discussions through posting. Both are essential for the forum to grow. They depend generally on different factors, although both external and internal factors control them.

Let us dissect the components separately. First, the number of registered members across date, plotted by the blue line:

One can see, there are some periodic annual variations; but except that, there has been no effect till New Year to registration. That’s a good sign! The red dots are the major change-points in the data. The periods of waxing, as suggested by them, are mid-Feb to mid-May ’06 and Mar-May ’07. Rest of the years follows a fixed trend. But what do these periods correspond to? They correspond to the post-exam time and summer vacations in India, when people enjoy some free time and can spend more time on the net.

Next, we look at the daily posts. The problem here is that, there have been large scale post deletions in the history of CGC few times, to reduce the huge server load because of the ever-increasing database size; the most recent one being in Nov ’07. Let us compare the total daily posts, before and after deletion on Nov ‘07:

The red lines (data till March last year) denote before deletion, and blue lines after deletion. So a large amount, about ⅔ of the posts, has been deleted. But the basic post pattern remains same. Moreover, there is no marked change around Nov ’07 (deletion time), which means for getting recent trends the deletion does not cause problems.

So we can safely analyze the daily posts:

It also contains some change-points:

(i) The first one is at mid-Feb ’06, when the post rates had increased tremendously. We have seen that registration also increased in that period. And during that time, a large chunk of very active members (with high personal post counts) came in who posted actively, increasing the daily rate.

(ii) The rates fall somewhat in Aug to Dec ’06. And the reason for it is the-then turmoil in CGC, resulting in server hacks and bans.

(iii) They increase again from Dec ’06, when recently a lot of active people had joined, as well as a lot of discussions started about English sections, sharing issues and post restrictions to sections etc.

(iv) The post restrictions blocked all those people who just made posts like “thank you” and took the gifts. I showed in last report that it resulted in an increase in numbers of people who reached post counts 100 & 250 to gain eligible statuses. Still, many people were restricted, and it brought down daily post counts. Moreover, internal tiffs contributed to it.

(v) In Sep ’07, there were large collisions and collusions, as some people slowly separated from CGC. So the post rate fell. But since then, the rate has been nearly constant, even slightly increasing, which is another good sign.

I was talking just now about joining of active members. This will be shown in the CGC members’ timeline.

Here is the number of active members per day. It has the same pattern of posting, naturally. It directly gives the number of people who post daily, though it misses out those who come and read topics but do not reply. Anyway, note that among thousands of members, only a few are active.

The CGC members’ timeline is given bellow. It shows, corresponding to each day, the post rate of members joining on that day. People with high post rates or number of posts are marked.

Last time I did talk about poverty in terms of posts, participation of the female members, share of post among top members, popular sub-forums etc. This time, I feel they would not give the proper picture, because many high-post members have become inactive. I hope this report will give an insight into the recent affairs of CGC statistically. Also, it shows that there is no sign of decay, but stability.