Comments on: Networks nodes: handbuild one by one

By: srp

srp — Wed, 26 Mar 2008 21:21:16 +0000

The average would be constructed by Monte Carlo methods: randomly drop threads onto the buttons, measure the relevant parameter of network structure (% connected), repeat until statistical significance is gained. I believe Kaufmann showed that this connected happens almost all the time.

By: peter

peter — Tue, 25 Mar 2008 14:54:14 +0000

Andrew — I don’t agree with your (or Kauffman’s) conclusion in the button experiment. Perhaps the conclusion that the network is almost entirely intereconnected holds on average, but it would not necessarily hold in every case of random assignment of threads. For instance, if the same button was always selected when 4,999 threads were allocated, then one single button would be connected by 4,999 separate threads to 4,999 other buttons; in this case, there would still be 5,000 buttons unconnected to any other button, and I could not call such a network almost entirely interconnected.

However, even if the conclusion held on average, this would not necessarily mean it held for a majority of the networks obtained by this procedure (depending on just how an “average” outcome were to be defined).

By: Andrew Taylor

Andrew Taylor — Tue, 25 Mar 2008 12:39:15 +0000

Grant, well noted and well said. Perhaps the task of connecting people would seem less daunting if you made it less of a generic demand (I must connect more friends) and more of an opportunistic one. Whenever a friend or colleague raises and interesting question, article, or idea, for example, I can often think of another friend or colleague who might be wandering the same track. A simple e-mail to them both flagging the connection is all it takes to make the connection, and I’m saved from sitting at my desk considering every possible permutation of more general introductions I might make.

It recalls Stuart Kauffman’s wonderful thought experiment about complex networks in his book *At Home in the Universe*, in which he described 10,000 buttons cast onto the floor, and the random process of connecting two buttons at a time. Once you have half as many threads as buttons, the network is almost entirely interconnected (I describe the thought experiment in more detail here: http://www.artsjournal.com/artfulmanager/main/066774.php)

If we can bring slightly more than random connection to the process (in response to a thought, an idea, a whim), we can be far more productive and constructive than the friend-gathering systems now available on-line.

By: IshMEL

IshMEL — Tue, 25 Mar 2008 10:14:59 +0000

What the world needs is more Lois Weisbergs. ( http://www.gladwell.com/1999/1999_01_11_a_weisberg.htm )

By: peter

peter — Tue, 25 Mar 2008 08:51:54 +0000

Eric’s post reminded me of a colleague who worked in the R&D arm of a US telecommunications company telling me of a method his group used to generate new research ideas: Once a month, they invited a group of unrelated academics or reseachers to spend a day visiting with the telco R&D team. So, for example, a university linguistics department was invited to meet and talk about their research one month, while the R&D department of a merchant bank was invited another month, etc. Many of these meetings led to lots of new ideas and collaborations (fruitful for both parties), as well as to new friendships.

By: Eric Nehrlich

Eric Nehrlich — Tue, 25 Mar 2008 08:04:29 +0000

Great post, Grant. I love making those sorts of connections between people who would otherwise not meet. They don’t always work, but as somebody who tends to be a “boundary spanner” in social network analysis terms, I do think it’s my responsibility to bring networks together. I generally go with the party approach to reduce the awkwardness of the “you should meet” dynamic – one of my favorite parties was when I brought together people from four different social circles of mine to see what would happen.

By: peter

peter — Tue, 25 Mar 2008 05:53:09 +0000

While I agree with your general point, Grant, there is something which has always troubled me. Person A can be close friends with Person B, who in turn can be close friends with Person C. But Persons A and C may not get along at all, perhaps because they do not see in each other that which B sees in each of them, or perhaps because what each of A and C has to offer does not fulfill a need that each (respectively) fills for B. Friendship, alas, is not always transitive.

But I agree we should try to make these connnections, for when it works it’s wonderful!

By: communicatrix

communicatrix — Tue, 25 Mar 2008 00:33:57 +0000

I love this. I try to do this, but don’t do it enough proactively–it’s usually in response to a need of some kind.

My favorite story of this kind of “you should meet…” was how my father (who lived in Dallas at the time) met one of his best friends (an L.A. resident) in later life–a wonderful friend who has stayed close to my sister and me since our father’s death.

They used the same car service driver when they were each in New York, and they were both the kind of people who liked talking to pretty much anyone who was interesting. Which Artie clearly was, bless him…