"Tim's critique, while entirely valid, reminds me a bit of those famous quotes from the first few decades of computing about the upper limits on demand for computing."
He's probably referring to something that T. J. Watson might have said. While I do affect the curmudgeon pose around here at times, the cases are not comparable. If T. J. ever made the famous comment about '5 computers', it was a misapprehension about market elasticity, capital availability and technology learning curves. The specific issue that I called in Reed's Law is of a different nature: the limits on human time and attention. More hours in the day is one thing that famously cannot be bought. A 'Law' that assumes abundant time to even to find out if others are worthy of attention is ignoring this invariant, among other issues.
I also drew a somewhat snarky reply in e-mail from Dr. Reed himself. (OK, the original post was more than a little troll bait.) He refers me to Figure 4 and surrounding discussion in his original paper, which sure enough does talk about some of the self-limiting issues of saturation and competition. One small problem. The Law as circulating does not model these, it is monotonic increasing to say the least.
So why am I banging on about this, other than as an intellectual exercise? Two reasons. The first is the ability of formulations of this sort to enable delusional valuation arguments among the investing community. You might remember something of the sort happening a few years ago. Here's a good example of the kind of thing I'm talking about. It's precisely the "broad applicability of Reed's paper with the confines of Silicon Valley or even Sand Hill Road" that I'm worried about. I've no problem with KP or Benchmark putting down markers on this space, at whatever valuations they choose. If you haven't noticed, VCs tend to be an opinionated lot, and diversity in that opinion makes sure the space of possibilities gets explored. And just as well the big funds do so - those 'big' bets are relatively small to them, but will make sure theory gets tested. But reality is the entire social software / social networking space still has huge questions attached, from the reality of entry barriers and switching costs to problems of revenue extraction and product and service definitions. We are still in a very messy evolutionary phase of the medium - particularly as an investing opportunity - and need to be careful, not glib, about the values that are created so far. If this be curmudgeonry, make the most of it.
Second point, and the more central to my interests: This issue spun off from an ongoing discussion among Jeff Jarvis, Ross Mayfield, and now Fred Wilson about measurements that are appropriate for new media, particularly community or conversation oriented, with an eye to making them commercially viable (pun intended). We're not talking green fields here: We are inherently going to need to take time, attention and money away from old media for this to happen. It would be good to be able to talk about the dimensions of this competition, in a way that we might in due course be able to tie back to (mis)features of the media frameworks and how they enable conversations or community (for example).
Does the current formulation of Reed's Law do this? I submit not, neither in its formulation nor predictions. Case in point: You'll notice I haven't sniped at Metcalfe's Law, in spite of its similar simplistic, monotonic increasing nature. Applied in its original domain - layer 3 architectures - it predicts that all computer networks will collapse into one. Circumspice. Reed's Law makes an even stronger prediction of the same thing in the 'social layer' of networks: the winning social platform will take all, quickly. Some readers will no doubt have heard of the Usenet, the ur-social software of the infant net. It was all there: persistent memory, global scope, high penetration rate among users, ability to form new groups. Now it's a backwater. Why didn't it take all, and quickly? All of us who lived through those days have ideas, but O(2^N) provides no ways to talk about them, or to describe how the next Orkut, LinkedIn, Typepad, or whatnot will shape the value that individuals derive.
That's what I care about, because I think there are a lot of interesting questions to explore. I'll sample out a few:
Back in the day at Compuserve, we found that long term users participated in somewhat over two 'forums' on average. You could argue that was limited by the closed nature and limited market penetration of that system. You could argue that it might be comparatively high due to the strong identity model. What's the average online social community membership now, with wide Internet adoption? What characteristics of a platform increase or decrease it? What are the effects on 'real world' community participation?
Once upon a time the 'lurk-to-post' ratio on systems as diverse as mailing lists and the Well seemed to be about 10:1. Is it still that way on blogs with comments? Wikis? Does this number say anything about the relative attractiveness of platforms? Communities?
One of my Apple researchers once observed an 'iron law of mailing lists' that proposed they would collapse or fission when the number of active posters got much above fifty. How does this vary among systems? Again, what does it suggest about the utility of the platforms, or the cohesion of communities?
Is there a Power Law of online communities and conversation? Perhaps. How do characteristics of the platform affect the coefficients of the curve? Its practical limits on group size- both upper and lower? If there is such a thing as total social platform utility, a la Reed's Law, how is it affected by the size distribution of communities? Are there subpopulations of users for whom the optimal platform will be significantly different - enough so that collapse onto one platform will never happen?
These are all just top of mind, and all arguably relevant to dissecting the absolute and competitive values of "citizen's media". And the marvelous thing about the net is that much of this information is lying there exposed, to something like a Technorati crawler equipped with the right analytics models. Which I'm still looking for....