Not my participation ... yours! And your, and yours, and ..
Others have noted the growth of Daily Kos web traffic: the most recent figures from sitemeter predict 16.4 million visits in the next week alone. As close as I can tell from the sitemeter stats, this is a 3-4 fold increase from a year ago.
But how many people actually write diaries or recommend them? And how does that compare to the number of registered user ids?
I try to pull down a copy of every diary posted and (after the diary closes) the people who recommended it. I've been doing that since just after the last national election in 2004. In addition I've been tracking user id usage for a couple of months. As of last Saturday, the number of user ids was 72759.
Some only write diaries, others only recommend them (presumably after having read them), and still others both write and recommend diaries. These groups were enumerated weekly from November 6, 2004 to November 25 2005 as shown in the following table.
abbreviation | Type of participation | number |
D | writes diaries | 3159 |
R | recommends diaries | 6597 |
R&D | writes and recommends diaries | 7798 |
T | total | 17554 |
I'm pretty impressed that as many as 24% of registered users have participated in user diaries at one time or another by posting one or recommending one. But from week to week, the extent of participation is somewhat smaller. Here's a graph that makes that point. For each week from Nov 11, 2004 to Nov 25, 2005, it shows the total number of participants that week, as well as the size of three subgroups; those who write diaries (D), recommend diaries (R), or both (R&D).
In the aftermath of the election participation declined, particularly during the end of year holidays (Thanksgiving through New Year). In the first quarter of 2005 participation grew steadily, then flattened considerably in the 2nd quarter. With hurricane Katrina participation increased dramatically and appears to have settled at a new higher level. The Thanksgiving holiday saw a predicable decline, and a similar decline should be expected around Christmas/New Year holidays. The overall increase in weekly participation, from 2004 to 2005 is about 2000, not quite doubling.
But what accounts for the increased numbers? Are more people showing up, or more people sticking around? To get at that question I have another graph, that shows, for each week, the number of participants (total, same as T in first graph), those who had participated the previous week (stayed), those who had not participated the previous week (new), and those of the previous weeks participants who failed to participate (left).
It's interesting that there appears to be an alternation between weeks where those who stay newly arrived users outnumber those who left and vice versa, such that, over time, the two trend together. Most recently, the trend has been an increase in both people arriving and people leaving, though this is masked somewhat by the large increase post-Katrina in people staying from week to week.
There is a large discrepancy between visits per week, which are greater than 10 million, and number of participants, which are currently around 5000 per week. If each of those 5000 visits kos 10 times a day every day it doesn't account for even a million visits. So it is not inaccurate to say that the user community at Daily Kos in some ways is like the staff of a large newspaper or magazine, producing a media product which many more read. In fact, it would be fun to see what the ratio of circulation to size of creative staff is for some leading magazines and newspapers. Where would dKos 3200:1 ratio (16E6 / 5E3) fall?
I have no predictions for the future. I do admit to wondering whether, even though dKos is a virtual place, it may be just as subject to natural limits on size as a venue in the real world, and whether we are now testing those limits. This is the "too many sea lions and not enough rock" conjecture. If that is the case, what could we do about it? If nothing, are we doomed to continual cacophony and pie fights as natural consequences?
As a final note, I should point out that I don't include commentators in the count of participants, because I don't collect data on individual comments. In defense of this, comments are the area most subject to trolling and other sorts of bad behavior, and including them would complicate the analysis by introducing even more categories of participation. Still, there is no doubt that this omission underestimates participation, by my best estimate perhaps as much as two fold 20 percent.