How Big Is The Network?

One of the things that venture capitalists do is try to decipher how big the network is when they look at startups.  Then, they try and figure out how big it really can become.  Peer to peer networks are replacing hierarchies.  It is important to understand how the operate and grow.  Each network is different, but there are similarities across networks.

Recently, Facebook ($FB) bought WhatsApp for $19 billion US dollars.  There has been a lot of discussion about strategy, why Facebook would do that, was it too much.  There has been a little discussion about the network.

How do you derive how big it really is?

Ron Burt a professor at the University of Chicago co-wrote a paper in 1994.  They try to explain how to measure the size of a network quickly.  Networks aren’t just about direct connections from person to person.  They are also about uncaptured relationships between people.  Call it the three degrees of separation if you will.

What is interesting about this paper is it covers the network of a 2000 employee electrical industrial plant.   The process of production is complex, not assembly line.  Even though it doesn’t cover networking via software, I think many of the principles are still the same.

The interviewed employee described the organization of his or her work – how the work is scheduled, where supplies
come from, who depends on the output and soon – then named other employees involved in the work and most important to completing it. Name interpreter items followed. How often do you speak with him or her? How long have you known the person? Most important, respondents were asked to evaluate the strength of connections between people where connectivity was…..

With WhatsApp, because it deals with groups, isn’t that just like the network connections in the factory?

How do they figure out how big the network is?  It requires survey, then some assumptions, and some math. They call it capture-recapture.  It is pretty involved, and they take a lot of pages explaining and illustrating it.  Here is the power of it. The plant had 2000 employees.  Of those 2000 employees, there were 616 strong connections, and 19,409 distant connections.  It had a global density of .14.  Global density doesn’t mean global as international, but a measure of how interconnected the network was inside the company.  In this case, the engineers were the most connected people in the company.   The 2000 employees actually created a web of slightly greater than 9x.

Here is how they looked at distant relationships.

If we did not capture the relation between Bill and John, we looked through the relations we captured for some third party whose relations with Bill and John are known. The third party is an intermediary that completes a connection between Bill and John. The indirect connection could be strong, if the third party is close to Bill and John, or it could be distant, if the third party is distant from either Bill or John. Bill’s relation with the third party times the third party’s relation with John is a measure of connection between Bill and John.

Distant relationships are the breadth of the network.  Breadth is important, but it’s the frequency of connection that is critical to the network working.  500+ connections on LinkedIn ($LNKD) doesn’t mean anything. It’s how often there is interaction between those connections.  Obviously, stronger connections have more interaction.

Here is what is interesting.  “Of the indirect relations through one intermediary, a large proportion, 73%, are distant connections. But the remaining 27% amount to 3556 relations, 3394 average strength connections with a value from 0.25 to 0.75, and another 162 strong connections with a value over 0.75.”   27% of the indirect relations in this factory have pretty strong connections for some reason.  Almost a third of the network.

Imagine if a third of the WhatsApp network was the growth part of the network.   Maybe that’s what Facebook was really betting on?

In the old days, strong relationships were defined as personal contact through conversation.  Social networking is redefining what that means today.  But, how strong is a “Like” on a Facebook page?  How many “Likes” does it take to equal a comment?  Are comments stronger than “Likes”?  On a comment platform like Disqus, is an upvote stronger or weaker than a reply?

These are things that are being sorted out when looking at networks.  At this point, we have suppositions, but nobody knows. Klout and other apps are trying to solve this problem, but I don’t think they have it licked.

Enhanced by Zemanta


The information in this blog post represents my own opinions and does not contain a recommendation for any particular security or investment. I or my affiliates may hold positions or other interests in securities mentioned in the Blog, please see my Disclaimer page for my full disclaimer.

You might be interested in:
blog comments powered by Disqus