140 discussions were selected on the basis of having activity within the previous 24 hours.

What were the ten most popular threads?

To answer this, it is necessary to decide how to measure popularity. Two options come to mind: the gross number of views a discussion receives, and the rate (expressed as a percent) at which viewers respond by participating in a discussion.[1]

I required that a discussion have a minimum of 100 views in order to qualify it for the top ten list. The choice was arbitrary, but as it turns out the mean number of views in the sample is 92 and the median is 37.[2] I did this in order to prevent a situation where, for example, a discussion would be in the top ten list because it had two viewers and one response (i.e. a 50% response rate for an otherwise unpopular thread).

Table 1 presents the top ten discussions when ranked by total number of views. The mean number of views is 581 and the median number of views is 427. It is noteworthy that the mean and median response rates were 5.58% and 5.61%, respectively. In the overall sample, those rates were 16.60%/12.88%. This is to say that the threads with the highest number of viewers had a response rates that were well below the average for the sample. I would reckon that this is on account of the strictly finite number of activists on the forum, as a result of which - over a certain number of viewers - the percentage at which viewers respond will steadily diminish. Seven out of the top ten discussions measured this way have as their purpose the distribution of videos.

Table 2 represents the top ten discussions ranked by the rate of response. Notice that only two of the ten discussions (ID numbers 9 & 10) occur in both lists. Again, in the overall sample the mean and median number of views are 92 and 37, and the mean and median response rates are 16.60% and 12.88%. In the top ten discussions ranked by response rate, the mean and median number of views are 303 and 149 - many times more than for the overall sample, while the mean and median response rates are 11.91% and 12.08%, which is in line with the response rate for the sample as a whole. Six of the ten discussions ranked by response rate were video distribution threads.

Both these methods of measuring discussion popularity have their redeeming features. The total number of views measure may have more value in a case such as this, where all access to the site requires a membership, and both measures will suffer in the case of a site that is both well known to the general public and allows anyone to view discussions.[3]

Finally, Table 3 presents the various means and medians discussed previously.

--

[1] Either way the results will not be immediately comparable between sites, given the effect of access controls on each site (e.g. open to the public, partly open to the public, or exclusively for members in good standing).

[2] The selection of candidates could be further refined, for example by only considering discussions that had a duration of less than 5 days (discussions on jihadi forums typically cease to be active after 24-48 hours).

[3] A third measure would be response rates - not in terms of the number of viewers of a given thread, but as a percentage of the forum's membership. This forum had 4,337 members on the day in question. Note that such a figure ought to be adjusted for the rate of attrition in site membership, a rate which would be difficult to measure by anyone other than the forum administrator(s), and which may not be uniform across forums.