As everyone from academics to engineers to tech reporters to Web 2.0 bloggers try to apply formulas like Metcalfe's Law and Reed's Law to social networking, I could not find a good timeline of actual myspace growth statistics.

Figuring out the value of nodes and hubs connected in an affinity network is a nice theoretical exercise, but the actual growth data speaks for itself. Its not exponential nor logarithmic, but follows a declining percentage model. All social networking startups (including this one) should take a realistic look at myspace data before projecting millions of users from viral growth...

Last update on 3/6/07:

Dec, 2006 - 138

Jan, 2006 - 145

Feb, 2006 - 154

March, 2006 - 161M

The data was taken from various sources like press releases, news articles, and blog entries. I tried to take data that was uniformly mid-month, but the granularity of the data is not exactly stellar. And the 3rd party data is as accurate as the sources themselves. Furthermore, different sources reported slightly different numbers for the same month. So the data should be taken with a grain of salt, but it is the best attempt outside of Fox Interactive Media :)

Plotting this data against the curves of consistent 20% or 15% growth shows the inaccuracy of an exponential curve-fit.

The curve-fit in light blue is what I'm calling the declining percentage model. If you take the top chart and calculate monthly percentage growth, you'll see that average growth was roughly 20% in the first 12-month period, 15% the next year, and 10% in the following year.

The declining % represents "true" viral growth - the kind of organic growth that happens after a marketing campaign where planned (and unplanned) distribution channels were already exhausted. I base this claim on 3 reports that documented Myspace's initial launch campaign. All reports agree that Myspace ended its aggressive marketing campaign to seed the network with its first couple million users in ~Q2 of 2004.

These numbers don't disprove or prove Metcalfe's Law. Metcalfe's Law is conceptual in nature and is expressed in the value domain as opposed to the numerical time domain.

Edit (10am on 11/21):

This article was on the front page of Digg for a few hours today, and some smart folks indicated that a logistic function should be used to model the growth. So I went back to the data, figured out the constants and derivatives, and applied a Verhulst equation

We now get a slightly more conservative projection and curve-fit.

Qualitatively, a logistic function reflects how the initial stage of growth is approximately exponential, then slows down asymptotically as competition arises or the product matures. The Verhulst formula, in particular, is often used to model population growth that takes into account the size of the population and the amount of available resources.

One can argue that US growth has saturated - ie no more resources (besides spammers). Some amount of growth is left in Europe, and a large untapped pool in China. It will be interesting to see Myspace numbers in late 2007 - one thing is for certain, the growth will no longer be exponential...

Edit (3/26/07): I've gotten a couple requests/questions to explain how I derived the constants K and r.  You basically have to do a polynomial interpolation with the raw data provided.  Taking different discrete data points will give you slightly different values, so you'll want to do some experimenting until the curve fit is about right.  Interpolation can be done with a graphing calculator or a simple script.  I'll leave the actual values of my K and r as an exercise for the reader to uncover :)