In the mid-90s, as internet adoption picked up steam, Nicholas Negroponte at the MIT media lab used to talk about the “daily me.”
Individuals would be able to create personalized filters to view a newspaper that contained only the articles they wanted to read. Social critics worried that the “Daily Me” would be the death of democracy. They argued that that this lead to a world where people lived in their own bubbles, only seeing the information that confirmed their own prejudices.
That world may have arrived. Valdis Krebs, who consults about social networks for a living, did some interesting analysis on link patterns in the “people who read this book also read” recommendation engine on Amazon.com.
He started with a single book that he was looking up on a recommendation, The Silent Takeover, and traced that patterns of recommendation that surrounded it.
Here’s a link to the pattern he discovered. There’s a set of books that seem to represent “left-wing” readers, with titles by Chomsky and Michael Moore and Tom Friedman. And there’s a parallel set of books that seem to represent “right-wing” interests, with books by writers including Ann Coulter and Patrick Buchanan.
The clusters of recommendations seemed to be mutually exclusive. Only one book appeared on recommendation lists in both clusters: What Went Wrong, a book by Bernard Lewis about Middle East history.
Does this mean that we’ve arrived in the world Negroponte saw in his crystal ball? In Valdis’ words, “once the propoganda gets into the echo chamber, you hear the same message continuously from many different sources, and you begin to believe that is how the world works.”
Valdis’ research isn’t conclusive proof. The methodology he followed was “snowball sampling”, as it’s called in network analysis circles. The links were selected by browsing, following a near-infinite set of links in for a finite amount of time.
It would be fascinating to do similar analysis with a larger data set, to create a more conclusive result. (Dear readers with a statistical background, I would welcome your thoughts about how to know whether the result is reliable).
If a search of Amazon’s entire virtual bookshelf revealed the same result, what would it mean?
It doesn’t tell us whether society has gotten MORE polarized than in the past; history is full of divisive partisan politics.
And it only tells us that the self-selected group of people who read political books have polarized opinions. We know that less than half of the eligible population votes. Most people tune out of political conversations.
As Valdis said by email, “The challenge is to create *bridges* so that diverse information and ideas can be exchanged (not just via hollering and arguing).”
We need to create a conversation where more people are talking and more people are listening.
Mapping What We Read
Adina writes about some informal research done by Valdis Krebs that resulted in this diagram, which you can see full size by clicking here or on this gargantuan thumbnail: This is a rough-and-ready map of the reading preferences of the political left a…
I’m not a statistician, but the networks of books shown by Amazon also-boughts, being a sort of highly connected social network, seem like a good case for a power-law distribution. I think this means that there are bound to be peaks and valleys of influence.
That said, I’m not sure what to make of Krebs’ chart. When I go to http://www.amazon.com/exec/obidos/ASIN/0195144201, I don’t see *any* of the links he shows in Amazon’s also-bought list for What Went Wrong. It could be that Amazon’s lists are very dynamic, which calls into question the validity of the chart.
Still, I’m convinced that there certainly will be circles of influence of ideas that don’t overlap much. Though they’ve always existed in social networks, they’re better defined by things like the also-read lists. Tools like that, because they weigh against random browsing, could be causing the power-law distribution to be more pronounced, i.e. we do become more tightly grouped in our communal views.
As Valdis said by email, “The challenge is to create *bridges* so that diverse information and ideas can be exchanged (not just via hollering and arguing).” But it’s so difficult to have an intelligent discussion with people who disagree with me, since they’re obviously not as smart, and are prone to hollering and arguing.
Steve, I added a link between two books IF EITHER book listed the other as a buddy. Because the books shown sell at different volumes/quantities and Amazon only shows the top 5 buddies we will have some ‘apparent’ mismatches. X, a bestseller, may not show Y, a medium seller, as a buddy. But Y will show X as a buddy.
Network analysts have various rules of thumb for showing links when the relations are reported from one side, the other, or both… much depends on what relationship/flow you are mapping.
Go to Amazon and look up “Clash of Civilizations” and then look up the 5 buddies/’also boughts’… you will find “What Went Wrong”. I just did[Noon EST, Jan 8th]… it has not changed in the last week.
Power law network? No, this network does not match the criteria.
It is more like the small-world networks described by Watts and Strogatz in their famous 1998 article… yet, not enough ‘short-cuts’ [i.e. bridges] to fit their criteria of an efficient[high-clusterng AND short path lengths via short-cuts] small-world structure. Ironically, the book network is like the ‘caveman’ graph Watts describes in a later article and his book. The current book, NEXUS, describes this work very well.
Mapping across networks is what everyvoice.net is all about. thinking about a conference on the spiritual dimensions of the online commons. seeing the religious dimension of protecting the common weal to use the old english term
Wetware Trendwatch: Week of October 20, 2003
Wetware related news from the week that just passed.