February 22, 2004


I've heard Dean called the "dot-com candidate." Well, his spending may have been dotcommish, but his grassroots membership growth curve sure wasn't. Dot-com curves are not supposed to level off the way Dean's did.

I started collecting the Dean counter images off of Blog for America back in mid-September 2003. Why, I don't know. I guess I had a hunch something interesting might emerge from looking back on the numbers.

Well, the numbers are rather interesting, once graphed. I would argue there are three interesting moments in time on this graph: point A, point B, and point C.

Point C is easy: The perfect storm of Iowa, the "I Have a Scream" Speech, and New Hampshire. Enough to take the wind out of the sails, causing the dreaded leveling-off of the growth curve. Just watching this curve all during late January and the first week of February, it was clear something was very, very wrong: this was a rocketship without enough thrust to get into orbit.

Point B: I'm not so sure about: I'm guessing the buildup before Iowa. (I welcome other opinions.)

Point A is the one I'm wondering about. If whatever happened that first week of October hadn't happened the way it did, Dean might have 1,200,000 + grassroots members by now. What the heck happened?

Posted by brian at February 22, 2004 05:19 PM



When did Dean start the index page popup? It was in going into Iowa - wanted my email address, and obviously delayed site loading. Around point A?

Kerry had one when he started his announced campaign, but it was gone by Iowa.

Posted by: Forest at February 22, 2004 11:50 PM

I'm wondering if Point A is actually the emergence of Wesley Clark. He was getting a lot of favorable press in early October. Perhaps Clark siphoned off a lot of would-be Deaniacs?

Posted by: Brian at February 23, 2004 10:02 AM

There's also the issue of churn. I don't think the numbers had any way of reflecting attrition. They are always additive, so the rate of new people dropping is bad enough but even worse if it masks a general exodus of support. Interesting stuff, especially given the issues of bragging about numbers (the old page counter thing) vs. keeping them proprietary.

Posted by: xian at February 23, 2004 11:01 AM

perhaps its just an issue of saturation? The base of online politically active Dean inclined people might have reached its limit. There is only a limited people fitting this model. September marks the return from vacation, uptick in television watching. By the end of the month Dean might have filled up his "pure" base and shifted into a recruiting less inclined individuals.

On top of that the other candidates started pushing harder by then. More media, more ads, Clark in the race and both him and Edwards pushing online hard.

There also might have been an end to the curiosity sign ups too. I know I signed up for both Dean and Clark meet ups but never bothered to attend any. Also worth noting that meet up sign ups are most definitely not an indication of support for a candidate, just an interest.

B seems pretty easy too, Its the end of vacation and coinciding beginning of major attention being placed on politics. The start of the primary season so to speak. At this point a whole new audience is opened up, people into politics, but not enough to be thinking about it all the freakin time.

Posted by: Abe at February 23, 2004 12:42 PM

What a fascinating question! My theories: A is the impact of Clark (who declared his candidacy 3 weeks earlier) plus the distraction of the California recall, which drew a lot of activist Democratic attention to the west. B is the cumulative impact of Dean's missteps in mid-December in Iowa and the media pile-on.
Not that this is your job to do, but it would be fascinating to track the dates backward. It's a lot harder to go from 100 to 10,000 and 100,000 than it is to go from 100,000 upward, methinks. The old screen shots might be on the Internet Archive at www.archive.org.

Posted by: Micah Sifry at February 23, 2004 04:32 PM


I recommend that you calculate the second derivative of the curve at every point == the rate of change of ther rate of change. If positive, then growht is healthy. A value of zero is a flashing red warning light. A negative value is prety much like a dead canary in a mine.

The further back in time you go, the more interesting it will be.

You might calculate the second derivative of the MeetUp numbers and then compare. My suspicion is that the MeetUp members were stronger supporters [donors?] and that about half the 600K were simply tire kickers with little commitment or investment in the campaign.

Your thoughts?


Posted by: Jock Gill at February 24, 2004 04:23 PM

Point a... California Democrat Debacle... Wasn't Dean seen on national television... on stage... raising Gray Davis' hand... immediately before his crushing defeat by Arnold... endorsing the obvious loser... did Kerry join with Davis? I don't know... but it seems that this visual... was the beginning of the end... Dean wasted his political capital... on the obvious loser's futile attempt to hold power...

Posted by: Eric at February 26, 2004 07:38 AM

Just wanted to say thanks! Allus appreciate the work of a fellow "lay-scientist". (Dunno ANYthing about you, Brian, so don't mean offense if you ARE a Scientist.)

Comments above are good, especially the part that (although easier to measure) raw counts don't indicate future popularity with voters, just tire-kickin'.. And, as Dr. Shirky and others have posted, Net has re-defined what the counts and dollars indicate (probably by an order of magnitude).

Posted by: JayT at February 27, 2004 10:15 AM

brianstorms is Brian Dear's weblog. Non-spam email:

Be sure to take a look at these other fine websites:

Copyright 2002-2003 Birdrock Ventures. brianstorms is a trademark of Birdrock Ventures.