The "Serendipity Value" of Networks: Beyond Metcalfe's Law

Reviewing the syllabus and prepping my eBusiness course for MBA students at Baruch's Zicklin school of business, I again read the piece by Joe Weinman on GigaOm asking "What If Metcalfe’s Law Is Wrong?".

It, and a companion PDF linked from the piece, posit ways in which the network effect posited by Metcalfe -- that the value of a network goes up by the number of connections squared, or n(2), and the number of possible connections in one-to-one network is n(n-1)/2.

Weinman writes that while the theory of ever-expanding value may be theoretically true, there are real-world limits to humans' ability to capitalize on the network. We cannot all connect to all the hundreds of millions of people on Facebook, or the billions who have phones. Even with an average of 130 "friends" on Facebook, we tend to relate to only a few of them. We cannot possibly consume all the content offered; there's not enough time, not to mention interest or stimulation.

Yes, the true value of Metcalfe's law may be less than is sometimes supposed, and there are real-world limits. But something that seems to be ignored in the article is the value of what I might call "the serendipity factor".

Weinman assumes the value of connections is derived from the ability to communicate bi-directionally, or even consume information from that connection in one direction. But that abilty also begets new sources of information, and new connections to which one might not have been, likely would not have been, previously exposed. Just some examples:

- The Retweet on Twitter lets a user not only see something that someone else forwarded with minimal effort, but also something from a new source he might never have connected with or become aware of. That person can then connect with that new source. New value in two ways.

- Facebook walls allow similar kinds of serendipity. A friend of mine run content through her wall I never would have seen, or have a conversation with someone I never would have connected with.

- LinkedIn connections: You can reach into companies and organizations and to individuals through connections in a way that was not possible before -- you may not have even known that your "friend" was connected to this other person you were trying to reach, and via that connection you not only gain entree, but also authority. You skip boundaries.

- Mass sharing. There is surely value in the ability to let everyone on my network know at once that I have new photos to share, have an achievement, an article, etc. True, I may be actively connecting with only a few of them. But allowing more than those immediate connections to keep track of me, know something of me, gives both them and I more social currency should we choose to interact, need a favor, etc.

- Technological ease. Again, the network effect gives me the ability, combined with technology, to share content, information, sourcing, thoughts, etc, at a much broader rate and level than I could have otherwise. I benefit by gaining power from the power of network, a multiplier effect. I not only drop my content and ideas onto the network, but may gain new followers, have that content and ideas forwarded further, gain new inbound ideas and connections and inquiries, and further hone my thoughts and ideas.

1 comment:

Joe Weinman said...


I agree with you. There certainly is a "serendipity value," or perhaps "option value" to network connections, which you well articulate in a variety of contexts. For example, I've never communicated with you before, but Google search (a type of network connecting searchers to content) enabled me to identify you ("serendipitously", perhaps), and leaving a comment is a form of connectivity.

However, I did address this in my original Business Communications Review article. I pointed out in the opening paragraphs that one could rigorously define the value of a connection, based on expected value and discounted cash flow.

Such an approach doesn't change my conclusion, however. Consider playing roulette: while it is true that landing on any non-zero number MIGHT result in a payout of 35:1, the expected value of the payout from a one dollar bet is not a gain of $35, but rather a loss of roughly a nickel. The bottom line, literally, is that each potential outcome must be probability-adjusted to determine the expected value.

Similarly, while the virtue of a ubiquitous network is that it enables you to connect with anyone on that network, and it is true that there is always an ex ante non-zero probability of that connection, that doesn't mean that you can or will connect with everyone in such a way that you will derive meaningful value (as opposed to, say, the ego / status value of knowing that you can so connect.)

In the same way that you might count your chips at the end of the night to see how much you won or lost, an ex post valuation would show that, as I indicated, once networks get above a trivial size, only a few connections will have proven to have value, or that there will have been finite limits such as time or budget or bandwidth on consuming that value. This in turn leads to my result, namely that while the theoretical number of connections may be O(n*n), only a few will have non-zero, positive value, and thus the total value of the network may turn out to be only O(n).

Joe Weinman