The interoperability of social networks
cdixon.org
cdixon.org
"Mass exportation of email is not standard on most social networks — when a user friends someone they don’t then expect that person to be easily able to send that contact information to a third party along with hundreds of other addresses with just one click. In order to protect user privacy, we now exclude email addresses from the CSV export file."
http://techcrunch.com/2009/10/26/orkut-slows-hemorraging-to-...
This is almost exactly what Facebook says today, while Google PR fakes a stroke in front of every blogger who'll pay attention. What was different was that, in 2009, Orkut was still the network of choice in two important markets (India and Brazil); "openness" of the sort that Google is advocating for its competitors would have actually been expensive.
I'll call Google as wrong and bad here while they are correct in demanding interoperability for Gmail vs. Facebook.
I would hope that sunlight in these areas would force everyone to be open rather than allowing everyone to be closed.
Metcalfe's law assumes that when you add a new person to a network, he will communicate equally with everybody, so that he contributes 1 unit of value per existing node, leading to O(n^2) growth.
But that is wrong. Instead his communication is more likely to follow a Zipf distribution such that he contributes 1 unit of value to the existing node he communicates the most with, 1/2 unit of value to the one he communicates the second-most with, 1/3 to the third-most and so on. This means each new member contributes O(log n) additional value, leading to O(n log n) growth.
I don't think can call Google "open" or "closed" in the sense of interoperability with respect to page rank. No one else puts information into pagerank. Google's search software isn't open source but that's a different question than social-network/email interoperability.