Since we're now down to the wire on kos's 100,000th user registration, I thought I'd start occasionally doing an update until we reach that goal.
The following is the table in except tracking the number of registered users on daily kos since June 1:
The columns should be self explanatory, but I'll describe them below anyway:
Date
The date and time that the observation took place.
Hours
The number of hours that have elapsed since the first observation - in this case, June 1, 2006 at 9:00 PM
Members
Total highest registered member ID at the time of the observation
Difference
How many new members have joined since the first observation.
NM/H
How many new members are joining per an hour, on average, since the first observation.
Projected Date
Using the current number of registered users, and the average registration per hour calculation, the projected date and time that the 100,000th registration will occur.
The closer we get to the 100,000th registration, the more accurate this calculation should get. Below is a graph of the observations:
The rate of increase in membership appears to be rather constant, except for two notable occasions:
Between my observations on June 6 and June 14. In this timer period, 2,121 members joined over the course of 175 hours, or a little over 12 people per an hour. This period of time was when YearlyKos was going on in Las Vegas.
The other uptick appears to have taken place between the observations on July 23 and July 25 when 554 members joined in the span of 50 hours, or just over 11 members per an hour. Looking back on the site, I can't really see any explainable reason for this uptick, but it is there just the same (was that when Markos was interviewed on nightline or wherever?)
There are really three ways to calculate a projected date:
1) Calculate using the pace since I started observing. The pros on this is that the trend is more long term. The negative is that if there is a sudden drop or uptick in registrations at the end, the projected date is going to be off. However, this is the date I have in my signature - currently sometime in the evening on August 10th.
2) Calculate using the pace since the end of YearlyKos (the June 15th observation). Using that trend, the projected date actually moves out to sometime in the morning on August 12th
3) Calculate using data since the last significant variation. Unfortunately this gives me only the last 4 data points. This gives a projection sometime in the evening of August 13th (which, in this case, is basically the last week).
However, using all observations since June 1, there was a sudden drop in the projected date after YearlyKos which kept getting progressively later. However, in the past week, that date appears to have stabilized between August 8th and August 10th, so unless there there is an event that occurs that would make new registrations speed up or slow down, it will probably occur then.
Actually, my bet would be sometime on August 8th since that is the date of the Connecticut Senatorial primary, and registrations could very well pick up as people come here to get news.