I had planned to do a pretty lengthy write-up on this subject, but right now I’m in desperate need of sleep, so I’ll just let you guys see the graphs:
[ul][li]Two plots of our subscription progress (plot 1, plot 2), one starting at zero and the other at the first data point Ed gave us (1002 subscribers at 14 hours). A trend curve has been fitted over the data to project the final result at the end of the charter membership period. The data is current as of about 3 a.m. CST 2004/04/12.[/li][li]An estimated average of the number of registrations per day, going all the way back to 1999. The gap is the Winter of Our Missed Content, and the huge spike is when the LOTR thread got slashdotted. I should add here that I tracked the surges of registrations both right after the Winter and during the LOTR influx, and each tops out daily at about 130-150 new members, so this may be the maximum number the hamsters could handle, in spite of the (one would guess) much larger potential pool of registrants from the slashdot effect.[/li][li]A month-by-month view of SDMB membership registrations, with the subscriptions laid on top of it. To make the comparison more explicit, I made a separate graph of the ratio of subscriptions to “regular” registrations.[/li]No plot here, but one thing I noticed is that a not-insignificant number of members have made no posts but show recent activity – perhaps to run a search or to look at someone’s profile. I’m guessing that this could be a cause of the discrepancy between Algernon’s estimate of 3,800 active posters and Ed’s statement that there are 7,000 active members within the last 30 days. In other words, if the SD staff ran a database query (and I don’t know if that’s what happened or not) and assumed that “last activity” means “posting,” then this could skew the active member figures upward.[/ul]
This graph confuses me – who were the paying subscribers in 1999-2003? In what way were members and subscribers separate entities?
Somebody who joined in 2000 and subscribed now shows up as a subscription for 2000.
Yer keeping me awake, man. :o <- that’s a yawn.
Basically, the bar chart is a historical tabulation of how many people registered during which months. Then the people from that population who are now subscribing get plotted on the line graph.
Members = everyone who ever registered, who today would show up with the Guest status, unless they’ve paid – and the people who’ve paid are the subscribers.
I hope I’m making myself clear.
And, in my attempt at explanation, I get the bar chart and line chart exactly backwards. Maybe it’ll work better with an example:
In March 1999, 437 people signed up to be SDMB members, and this is shown by the red line and on the right-hand axis. Out of this population, 59 people have so far subscribed, which is the bar chart, with the units shown on the left hand axis.
Yep … perfectly clear.
From the fits on both of your graphs, it looks like we can extrapolate out to about 3.2 thousand posters by April 21. But this is likely to be an underestimate, since some folks may be waiting for the last minute. At a rough guess, 3500 might be within reach.
I find the monthly registered/subscribed percentage graph interesting because it shows that until the end of 2003 the general trend is a slight decline. I wonder what that could mean.
It is interesting. And what actually got me started on this little data-collection project is the prediction made by some people that going to a subscription model will kill off new blood and leave the board with nothing but the same old-timers around, yet some have recently observed that a lot of Charter Members are newbies – so I was trying to see if there’s anything to either claim.
In a way, I suppose both are right. My guess is that there is a direct correlation between the value of this board to an individual and the time that he or she has spent on it; this should explain why you’re more likely to see subscribers amongst those who registered in 1999 than those in 2003. On the other hand, there are probably a lot of lurkers who still want to retain the ability to search and figured this is their chance to do that at a discounted price, thus explaining the large spike after Ed’s announcement. What happened with those who registered during the first three months of 2004, though, I can’t explain.
Can I ask where this data comes from?
Great job, Earthling! 
It comes from lots of manual data entry. I looked through a lot of profiles – mostly during wee hours so as to (I hope) not clog up the pipelines. I don’t know of an easier way.
Which reminds me – Ed, if you’re reading this – maybe we can make the Member List available to subscribers? If that makes the hamsters cry, perhaps we can do it on a request basis?
Nice job Earthling. Having done a lot of this SDMB membership research myself I know how time consuming and tedious it is.
I first thought that it was perhaps an artifact of ever increasing registrations, but the registrations per day was in a continual decline throughout 2003 also. So, both the number of registrations and the percentage of those registering becoming subscribers steadily declined during 2003. As you say, it is interesting.
By the way, in case anyone is wondering about the huge spike in registrations in early 2003, it was due to a lot of people registering to post in the “What If LOTR Was Written By Someone Else?” 41-page thread.
no such thing on the World Wide Web! Thats when I’m surfing!
I think you misunderstood my question, Earthling - I meant how can you tell if a member is “active” here if they’re not posting?
In the user profile it shows the “last activity”. This gets updated whether or not someone posts. (To test this I went and looked at a couple of low-post-count Dopers and checked their last post against their last activity.)
Ya live and learn - thanks Algernon!
The data from the last few days is still on track. I’m actually pretty amazed at how well the curve and data fit each other. Maybe we are just a bunch of sheep, heh. But it’d be interesting to see what the last-minute rush looks like.
Hey, you’re my role model!
I see I did, and Algernon’s answer is correct. Just to provide some solid back-up, though, here is an example of someone who registered in 2002, has never posted, yet shows activity (log-in) on Jan. 22, 2004.
Any chance of releasing the raw data? I would be interested in subscription rate/day plots
Yes. Just hang in there…