I tend to be a fairly agreeable guy. For instance, I recently spent ten minutes discussing the astronautical prospects of boy band 'N Sync with a significantly younger female relative.
She was of the opinion that this band, which has evidently conceived the notion of going into orbit around the Earth in a shuttle, is qualified to do so, and furthermore, should be allowed to do so.
Well, I agree. I think 'N Sync should be shot into space.
Given this essentially amiable disposition, it was a bit of a shock to me some years ago to find myself disagreeing with Bob Metcalfe -- the creator of Ethernet, founder of 3Com and illustrious columnist -- on the subject of Metcalfe's Law.
Metcalfe's Law states simply that the power of a network grows in proportion to the square of the number of its nodes. You can see easily enough how Bob arrived at this idea. If each network node has a roughly constant amount of valuable data, connecting all nodes to each other should square the amount of valuable data available to the whole network.
However, it seemed to me that this was a bit of a simplification, because if you simply square the nodes, you include each node's connection to itself...which, needless to say, it already had.
So it's more mathematically accurate to say that for N nodes, the power of a network grows in proportion to N-squared-minus-N; you subtract out each node's self-connection as already having been established even with no network.
Now, Metcalfe's Law is interesting for many reasons, but among them is the fact that it's sometimes used as a justification for business-class, peer-to-peer (P2P) networking products. I've argued in this column before that that logic is essentially flawed because it was born in a freakish petri dish; that is, modern P2P was created to serve up a freakish type of data, which many people wanted, and many people had, but which wasn't mission-critical (i.e., pirated MP3 files).
This is contrary to the type of data produced by businesses, which is far more likely to be mission-critical, and which should therefore be professionally and centrally stored, administered and archived.
So, Metcalfe's Law really doesn't seem to apply to business-class P2P network power, because if a company doesn?t need a given file, it?s not adding power to the network to make it accessible.
Well, my previous column on this topic inspired a number of pieces of e-mail, which collectively reminded me of a folded chair as deployed by a professional wrestler.
And so, agreeable guy that I tend to be, I recently found myself wondering if there might be circumstances in which Metcalfe's Law might indeed justify P2P products. And I thought of instant messaging (IM).
Now, IM products (such as AOL Instant Messenger) are still most commonly associated with people like my 'N-fatuated female relative, not businesses, but the fact is that they're gaining strength as a corporate solution of real merit. For instance, the forward-thinking company where I work includes an IM product on every deployed laptop, and that policy is probably not all that far ahead of the typical IT curve. The fact is that IM products are superior to voice technology in many cases, and nobody disputes the business merit of voice.
Why should Metcalfe's Law apply to IM? Simple. With IM products, you don't have mission-critical data as a concern at all. Instead, the power of the P2P network is determined by the number of minds you can connect to each other. And if a company's human resources folks have done their jobs, presumably every mind in the company is mission-critical.
When I consider an IM-centric company like Jabber, for instance, I see serious business value. I see a firm that is taking IM seriously enough to develop open-source code, not just proprietary shrink-wrapped stuff. I see a carefully constructed extensible architecture designed deliberately for the exchange of business information. I see security taken seriously: IP addresses are hidden, SSL is embraced, and in the future, PKI will be integrated. This isn't kid stuff, and these guys aren't kidding around.
So to sum up, I think for some P2P systems, such as industrial-strength IM, it is actually reasonable to apply Metcalfe's Law and say that the power of the network grows as the number of powerful minds using it grows. That is, it grows as N-squared-minus-N... minus the number of participating members of 'N Sync... should there be any.
Bad Packets: Ballmer and Jobs -- The April 1 interview