The Shrinking World
Facebook got a lot of coverage yesterday for this blog post, which discussed two studies of its “social graph”. The headline finding - and the one used as lede by most of the coverage - was the idea that, as Mashable put it, Facebook has shrunk the “Degrees of Separation” between people in the world from “six to four”.
It’s actually more like - fans of mathematical rounding look away now lest yr faith in Mashable be forever shaken - “from 5.3 to 4.7”. And to say the least there are a lot of caveats. (For instance, while generalising from “Facebook” to “The World” is probably fine for a bunch of variables, ‘connectedness’ may not be among them.) But “Six Degrees” was always as much slogan as science so this doesn’t really bug me.
What does irk a bit is that neither Facebook nor any commentators gets beyond the gosh-wow idea that the world is ‘shrinking’ to ask what the practical, cultural or other implications of this might be. What good does it do us to know that the number of degrees of separation is now 4.7 not 5.3? How might it change the world?
As it happens, when I came across this story I was reading Mark Earls’ new book, I’ll Have What She’s Having (co-written with Alex Bentley and Michael J O’Brien). I’ll write a proper review of this at some point, but for now I was struck by a couple of findings he summarises which point to reasons the Facebook finding MIGHT be interesting.
They’re in his chapter on cascades, which is a very dense summary of a lot of 90s and 00s network, complexity and evolutionary science. I am not a scientist, and I may have misunderstood bits of what Earls and Co. are talking about: take this post as the hypothesising of an educated layman. But the book talks about two things which happen to information cascades when networks tighten and get more interconnected.
The first (via Stuart Kauffman) is that highly connected networks are more chaotic: from an evolutionary perspective their fitness landscape is shifting all the time, so there’s very little chance of agents in the network converging on a single ‘best’ strategy. And the second, related, point (via Duncan Watts) is that as the connections within networks increase, the chance of large scale cascades - so many people copying behaviour that it “sweeps” the system - actually shrinks. This is because an individual agent becomes overwhelmed by signals from their connections and it becomes much harder to decide which to copy.
Can we apply this to the real world and the Facebook case? Well, imagine you’re a marketer who hears that Facebook is shrinking the world. Your reaction to the press release might be “Hurrah! It will become much easier for my message to spread!” But take into account the findings I’ve summarised from Earls’ book and your reaction should instead be “Oh shit.” A more interconnected world means a shorter lifespan for your message and a smaller chance of large-scale behavioural change.
