August 10, 2006
Network effects: the laws explained
Nice article here from last month's IEEE Spectrum about Metcalfe's Law and other network value laws.
Personally I think whle there is some value in a log(n) function as the authors suggest, the underlying detail may be related to Milgram's six-degrees and subsequent identification of clusters and supernodes in social networks, which would suggest an uneven (i.e. non-quadratic) pattern of added value when adding members to a network - many new members form a tight network cluster with only a few known friends, while a rarer type has a social explosion closer to the Metcalfe figure.
In any case, I'd take Metcalfe's n-square law as being about the potential, rather than actual, value of a network. How close you can get to the Metcalfe line depends on the way the network operates. I think something like file-sharing P2P gets pretty close as it doesn't depend on forming relationships but on adding connectivity and resources. However, IM and email are much more likely to look like log(n).
A big thanks to Johannes Ernst for spotting this one.