Homophily, serendipity, xenophilia

There’s been a small but fascinating blog conversation going on surrounding the term “homophily”. Journalism and media critic Amy Gahran encountered the term in an interview I and Solana Larsen gave with Chris Lydon of Radio Open Source and explored the concept in an extended riff and a set of bookmarks. Tom, an educator living and working in Ankara, weighed in with a moving story about learning from a Guatemalan colleague. Michele Martin, an education blogger, worries that the internet as a whole is a source for homophily and may be making her (and all of us) dumber. (Here she’s pulling on some threads explored by Cass Sunstein in Republic.com and Infotopia. More on that in a bit.) And yesterday, my colleague and friend David Sasaki invoked the conversation in an important post on the difficulties of getting people to pay attention to voices from the developing world.

I love it when smart people join a conversation I was trying to get started. So here’s my attempt to flesh out a bit more of why I think homophily, serendipity and xenophilia are useful concepts, what little I know about the academic literature about them, and what I’m reading to learn more.

I was introduced to the term “homophily” through a post on Nat Torkinton’s blog titled “Homophily in Social Software“, which led me to an excellent piece by Shankar Vedantam titled “Why Everyone You Know Thinks the Same as You“.

“Homophily” is a remarkably useful term, a compact word that succinctly expresses the idea that “birds of a feather flock together” – that you’re likely to befriend, talk to, work with and share ideas with people who’ve got common ethnic, religious and economic background with you. It’s not a new word – it was coined by Lazarsfeld and Merton in 1954 in an essay titled “Friendship as a Social Process” – but it’s never quite caught on. A Google search for the term gives you 51,000 results, roughly as many as my favorite obscure, Greek-derived sociological term, “xenophilia”. (One of the high ranking results for “homophily” is an interesting question on Yahoo Answers with the unhelpful answer, “Homophily does not exist as a word although homophile (gay)does.”

I’ve been talking and writing about homophily as one of the concepts that helps explain the challenges and issues that surround Global Voices and my larger media attention work. It’s my contention that living in the 21st century requires understanding what people think, feel and want in different parts of the world, given that both the challenges and opportunities of next several decades are global, not local ones. (Understanding Iraqi attitudes towards a US occupying force and Shia/Sunni/Kurdish tensions better might have mitigated the disastrous invasion of Iraq. Understanding Chinese and Indian economic aspirations is probably a prerequisite to figuring out how to regulate carbon emissions while those nations embrace automobile ownership. And activists trying to change Chinese policy in Darfur would benefit from better understanding of Chinese pride, the concept of “face” and the power of nationalism.)

Historically, our understanding of attitudes and opinions in other cultures is a heavily mediated one. As Kwame Appiah elegantly outlines in Cosmopolitanism, it’s only in very recent times that most people have been able to directly encounter people from different parts of the world. And despite air travel, most Americans have an impression of Nigeria through newspapers, movies and 419 emails rather than from travelling to the country or spending time with Nigerians. We understand the world, for the most part, through what we hear about it, not what we encounter of it. To the extent that our understanding of the wider world is a poor one, it’s worth asking questions about our media is working correctly.

There’s no shortage of voices reporting a crisis in the world of journalism. In 2001, media critic David Shaw reported that foreign coverage had shrunk “70% to 80% during the past 15 to 20 years” due to economic and cultural factors. He quotes a 1998 study from UC San Diego which saw international news coverage in American newspapers shrink from 15% to 2% of total content. More recently, Alisa Miller – president of PRI – produced an elegant short video that graphically depicts the paucity of international news coverage on American television.

(News isn’t the only way we encounter other countries – movies, television and music shape perceptions as well. But journalism has an explicit public service function, a social responsibility to inform citizens so they can make political decisions. Some of the blame for an isolated, ill-informed citizenry has to fall on the news media.)

The shift from broadcast media to read/write media has the potential to shift this equation. Rather than encountering people through the filter of professional media, perhaps we can reach them directly through their blogs, videos, photos. (This isn’t always possible – the digital divide is very real, and I’d argue that many of the arguments I made about digital exclusion in “Making Room for the Third World in the Second Superpower” still hold.) We no longer need to wait for CNN to connect us with people and stories in Bangladesh or Brazil – the explosion of personal publishing means that someone is likely speaking up in those corners of the world.

The rise of the read/write web turns the problem of paying attention to the rest of the world from a supply to a demand problem. You can find Brazilian, Bengali and Bulgarian voices, but only if you bother looking for them, stumble across them or are led to them by creators and curators of content.

This new “digital disorder”, as David Weinberger describes it, requires new systems to make navigation possible. Some systems rely on trusted guides, editors who we trust to help us sort through the mess. (Think BoingBoing and the crew who man that ship, or any prominent site capable of driving traffic to smaller sites.) Others rely on collective intelligence of their users to suggest stories – Digg, Reddit, etc.

The systems that rely on network effects, as Torkinton points out, are deeply affected by homophily. (So, as it turns out, are the systems based on human editors.) Some of these systems ask you what stories you liked, then find others who liked those stories and recommend their favorite stories – this is a technique called “collaborative filtering“, and it’s become increasingly popular as a paradigm for navigating a complex, choice-rich world of media. You can see how CF could be a homophily trap – tell Netflix that you liked the film “Sneakers” and it will find you other people who liked “Sneakers” (most of whom are, like you, ageing computer geeks), and suggest other films they liked. The recommendations you’ll receive are likely to be good, but are less likely to be surprising and challenging.

Systems simpler than CF fall victim to homophily traps as well. A site like Reddit attracts a lot of young men who work technical jobs and lean to the left politically – rely on Redditors for your news and you’re unlikely to encounter many stories from the developing world, from the political right, from non-technical disciplines. (Yes, I’m writing in huge, sweeping generalizations here – comments pointing out a single Africa story on Digg to refute this aren’t especially helpful.)

Why is homophily a trap? Cass Sunstein argues that it can polarize us – in Infotopia, he cites a study he helped conduct that demonstrates that deliberation of political issues with like-minded people leads subjects to a more politically polarized stance. From this, and from a close reading of political polarization in the blogosphere, he argues that the Internet may make it easier for us to share information with likeminded individuals, and that in a political context, this could be a bad thing.

I’ve made a much less persuasive and elegant argument summarized by the aphorism “Homophily can make you stupid.” My argument, basically, is that it’s possible to miss huge trends, changes and opportunities by talking solely to people who agree with you. I use myself as an exemplar of this sort of stupidity – I found myself so baffled by the results of the 2004 US Presidential election that I invited Republicans to come have a beer with me to explain what they were thinking. (One did. Thanks, Ian.) If homophily is capable of misleading Americans about local politics, just imagine what we fail to understand about Egypt, Pakistan and Fiji by virtue of not consuming media recommended by people from those places?

Writing from an engineering perspective, Torkinton suggests that authors of social software need to first decide whether homophily is a feature or a bug. If the goal of an application is to broaden your information universe, homophily may be a bug, and designers may want to include “less relevant but also likely to be interesting” recommendations. He recommends framing this as a feature, searching for ways to deliver “serendipity”, which he defines as “pleasantly surprising the user”.

(“Serendipity” is a fascinating term. It was coined in 1754 by Horace Walpole, a prolific British novelist and correspondent. He referenced a Persian fairytale, The Three Princes of Serendip, referring to a set of characters who “were always making discoveries, by accident and sagacity, of things they were not in quest of.” While that definition – and, indeed, the story cited – don’t precisely map to current usage of the term, Richard Boyle offers a long essay tracing the coinage of the term and feels that Walpole’s invention was the definitive first use. “Serendip”, incidently, was a Persian name for Ceylon, now Sri Lanka. Always nice to discover that a concept like serendipity has global origins.)

Serendipity is harder than it sounds. It’s one thing to surprise someone – it’s another to surprise someone helpfully. It’s even hard to define – lately I’ve been arguing with David Weinberger about whether certain examples constitute serendipity. Looking on a library shelf for a particular title, discovering it isn’t particularly interesting, but discovering the exact book you need nearby? (Fortunate, but also a consequence of the power of a topical organizational system.) Finding a newspaper story you never would have searched for but found very useful because the editors put it on the bottom of the front page, in what Dan Gillmor has called “the serendipity box”? (Again, very fortunate, but hardly surprising that editors would drive readers to less-read content.)

The reason serendipity is important to consider is that it’s one of the few affirmative ways to get people to pay attention to news from the developing world. I’ve argued that there’s three basic paths to get people to pay attention to, say, Somalia:
– Fear: There’s no government there, and there are lots of angry Muslims. If we don’t pay attention, we’re ignoring the next hotbed of terrorism
– Guilt: People are dying there. If things get out of hand, you might see Ethiopia slaughtering large numbers of Somali. Let’s not ignore another Rwandan genocide.
– Greed/Opportunity: Sure, it’s a mess, but did you know that Somali Telecom is making a fortune in the north of the country?

These arguments always put me in the mind of Melissa Rossi’s book, “What Every American Should Know About the Rest of the World“, a book which seems to be designed solely for people who are embarrased at cocktail parties when countries they can’t find on a map become topics of conversation. My guess is that avoiding embarrasment is not an especially powerful motivator for engaging with international issues and opinions.

I’ve argued for some time that a different model exists – xenophilia. There are people in the world who are genuinely fascinated by the very breadth, complexity and difference of the world. Many of these people are “third culture kids”, people who were raised in one country but “from” another country. Others are people who live, work or love outside their home cultures. My colleagues at Global Voices are, for the most part, people identifiable as xenophiles. I think there’s an argument to be made that xenophiles are uniquely equipped to thrive in a globalizing world and that cultivating xenophilia should be both a personal priority and an aspect of a nation’s educational and diplomatic strategy.

But xenophilia’s hard. It’s one thing to say to oneself, “I really should pay attention to matters in Somalia” and another thing to do it. Joi Ito has talked about “the caring problem“, the difficulty of really caring enough about people in another part of the world to engage with news from that community. At Berkman, we’ve been discussing the problem in terms of broccoli and chocolate – you know you should eat broccoli because it’s good for you, but there’s just so much tasty chocolate out there!

Serendipity breaks the chocolate/broccoli paradigm – it gives you broccoli as you’re searching for chocolate, adn it turns out to be just the right thing at the moment. It doesn’t require you to identify as a global citizen – it just means you’re following your interest in sumo wrestling and find yourself discovering a debate about Japanese identity and Asian politics.

So… that’s more or less an outline of the ideas I’m wrestling with right now, trying to figure out if they might represent the outline of a book. Specifically on the questions of homophily and serendipity, I’m interested in these questions:

– Do people who are ethnographically/psychographically similar consume the same media? Where’s the causality in this? Do we both read the NYTimes because we’re both liberals, or are we liberals because we read the NYTimes?

– Does consumption of the same media lead to the polarization Sunstein sees in deliberation, or is that a feature/bug of the deliberative process?

– Is homophily in terms of bridging and bonding social capital? If so, is Putnam right that bowling leagues create bridging capital? (There’s a lot of middle aged white dudes in my bowling league.) What do institutions that create international bridging capital look like in an internet age?

– How do you create serendipity? Is this something that’s algorithmically possible in collaborative filtering systems or other recommendation engines? Is serendipity a function of breaking out of your existing, homophilic social circles, or is it better generated within those circles?

On my immediate reading list:
Birds of a Feather: Homophily in Social Networks, by McPherson, Smith-Lovin and Cook
Lazarsfeld and Merton’s essay introducing the term “homophily” in Freedom and Control in Modern Society, edited by Berger, Abel and Page, 1954
Putnam, both Bowling Alone and his recent, worrying research on trust and socioethnic diversity

Would love your suggestions, directions, critiques and feedback, both from people already in this conversation and those who’ve been watching from the sideline.