"Global Attention: What We Watch and Why It Matters"
Notes for a talk at Pop!Tech 2004
Ethan Zuckerman

Slide 1
Slide 2
Looking at graphs that measured the growth of the Internet in the 1990s, you could be forgiven for speculating that it was just a matter of time before every person on the planet was online. The number of internet hosts served by DNS doubled (roughly) every year from 1993 - 2000; the number of Internet users increased from roughly 30 million worldwide in 1995 to an estimated 650 million in 2002. Follow these curves to their logical conclusions and it seems clear that all 6 billion people on the planet will have broadband access by 2010 or so. (Indeed, many of the business plans funded in the 1990s depended on this assumption being true.)

Slide 3
As slightly more critical minds looked that these figures and realized that there was no mathematical law at work guaranteeing universal internet access, it became clear that access to the Internet was increasing most quickly in wealthy nations. Specifically, access appeared to be increasing fastest in communities that were predominantly wealthy, white, and urban or suburban. The divide between communities with a high percentage of people online and those with low percentages began to be referred to as "The Digital Divide".

While this disparity isn't troubling if you consider Internet access a luxury - some have quipped that there's a BMW divide between these communities as well - it becomes troubling if this divide has economic consequences. It does - US Bureau of Labor statistics show that more than 50% of jobs require computer skills and that the vast majority of high-paying new jobs require extensive computer skills. A gap in internet access quickly looks like an opportunity gap for people living in underserved parts of the US.

So the US Department of Commerce sprang into action, producing reports like "Falling Through the Net" and holding meetings to discuss the problem. They launched a website and promoted "universal access to the Internet" through schools and libraries. It's now hard to find a community in the US that doesn't have some public, free access to the Internet. (Given that digitaldivide.gov is no longer online, it seems clear that the current US administration believes the divide has been bridged.)

Slide 4
While educators and labor economists began working on the digital divide in the US, it became apparent to people working in international development that the digital divide looked a bit more dramatic between Africa and the US than it did between poor and wealthy communities in the US. According to the UN's 2001 Human Development report, 54.3% of Americans were Internet users in 2000; 0.4% of sub-Saharan Africans were.

Slide 5
In 2000, I found myself getting interested in this problem. I'd spent the past five years working for one of the aformentioned dot.coms - Tripod.com - whose profitability required all of humanity to get online, build personal homepages and buy advertising on our site. And I'd gotten to witness, first-hand, how global the Internet was becoming. In 1996, we noticed that our fastest growing group of webpage builders were from Malaysia. We were baffled by this, but responded by sponsoring the Malaysian olympic badmitton team - which won silver in the men's doubles... which might have been a good marketing move, except for the fact that our Malaysian users were coming online to build homepages supporting Anwar Ibrahim, the reformist vice president later thrown into jail on false sodomy charges... which meant they probably weren't too moved by our support for the national Olympic committee.

Tripod was based in Williamstown, MA, a community similar to Camden, Maine, in the sense that we're a healthy drive from anything that could be considered a city, we both have a population where cows outnumber humans and both have lots of people who visit us to stare at our trees. A central part of the creation myth of Tripod was the idea that one could run a business in the Internet age from anywhere in the world.

Just prior to helping found Tripod, I had been living in Accra, Ghana, and I wondered to what extent was this actually true. Did the Internet mean that computer geeks in rural Massachusetts would do business with badmitton players in Malaysia and cocoa farmers in Ghana?

Slide 6
So I ended up founding one of the early organizations designed to bridge the International Digital Divide: Geekcorps. Geekcorps was based around an extremely simple set of ideas:

- To take advantage of the Internet, you need geeks
- There aren't many geeks in Africa
- Geeks beget geeks. (Specifically, they train them. IT is one of the last great apprenticeship industries)
- There's lots of geeks who would be happy to go to Africa and train other geeks if you'd pay for their plane tickets.

Using this sophisticated and well thought out business plan, we proceeded to recruit a couple thousand geeks from the US, Canada and Europe, find businesses throughout the developing world in need of geekery and send a couple hundred geeks overseas for a few months at a time. And if this were a talk about Geekcorps, I'd proceed to offer a litany of succesful projects accomplished and lessons learned designed to convince you to give me large sums of money or get on an airplane to teach C++ in Ulaanbaatar.

Slide 7
But Geekcorps is mostly dead, through a combination of circumstances hard to discuss without offending my former employer, all of which I'm happy to discuss with anyone foolhearty enough to buy me a beer. So instead, I'm going to focus on different ways of thinking about digital divides.

Slide 8
Sometime shortly after 2000, rhetoric about "digital divides" started focusing on a new term: ICT4D - Information and Communications Technology for Development.

"ICT" reveals who's taken control of the debate. Geeks use the term "IT" to refer to their field; development people and philanthropoids use "ICT", reminding the geeks that, while TCP/IP networks are really fun, telephones, radio and television are technologies that are vastly more likely to be used by people in developing nations. When you're in a room full of IT and development people, you can utter the term "ICT" and tell everyone's affiliation by whether they look at you with recognition or bafflement.

Slide 9
"For Development" is an interesting clause. It asserts that we're not bringing ICT into developing nations because it's important in and of itself. We're doing it because it's got some sort of impact on human development - it will help people get health information, or grow more crops, or make money selling handicrafts to the rest of the world.

This distinction comes out of a debate that has plagued anyone who's worked with technology transfer: focus on technology to help people's lives improve in the long term, or focus on pressing life or death issues in the short term - food and water supplies, vaccination, etc. The only possible humane answer to the question is "both". Unfortunately, money for these projects isn't generally coming from tax revenues - it's coming from overseas development aid. This aid is insufficient to meet basic development goals. Current estimates suggest that it's going to require about $100 billion a year, until 2015, to reach the UN's millenium development goals. Current giving is around $50 billion a year, and unlikely to rise much higher. Reconstruction in Iraq and Afghanistan is taking money from projects elsewhere in the developing world. As the Center for Global Development puts it in their annual report "Ranking the Rich":

"The United States gives little development assistance for its size, ties much of it to the purchase of U.S. goods and services, and allocates it to countries generally richer or more corrupt than recipients of development assistance from other donors."

As for all the rhetoric about a "Marshall plan" for development in the middle east, or anywhere else in the world? The US currently spends about a tenth - in percentage of GDP terms - of what we used to spend on international development: 0.1% versus 1%.

To win some of the scarce development dollars to work with technology in the developing world, it's critical to do two things:
- Ensure donors that you're working on a project that's got an explicit development goal
- Ensure donors that you're working with the neediest population you can identify.

Follow those criteria and you'll get a project unlikely to offend a funder... and extremely unlikely to succeed, generally a project designed to discover development uses for a rural cybercafe. You end up taking on the hardest possible digital divide problems. You pretty quickly discover four problems that make these sorts of ICT4D projects extremely hard to accomplish:

Slide 10
- The electricity divide. One of the reasons the Internet spread so rapidly in the US was that almost every US household has electricity. This didn't happen overnight - from 1935 to 1994, the Rural Electrification Administration used cross-subsidies to support the wiring of rural communities. Most rural areas in Africa are off the grid - any electricity is provided by diesel generators... an added cost to any rural digital divide project.

Slide 11
- The telecommunications divide. In 2002, Ghana had 12 land phone lines per 1,000 people. (That teledensity rises to 5.4% in Accra, which tells you something about the corresponding rural teledensity.) The US had 700... in part, the result of our universal service efforts, in place since 1934. While cellphones are critically important to the developing world, they're a poor way to provision data. So we end up using satellite connectivity for most rural digital divide projects, which is inefficient and expensive.

- Language and literacy divides. The Internet is a written medium. Specifically, it's written in English. A 2000 survey estimated that 78% of websites are in English. Roughly 40% of sub-Saharan Africans are illiterate, (World Development Indicators, 2002) and a large additional percentage are functionally illiterate. It's a real challenge to find content that's going to be accessible to target populations in rural areas.

- The relevancy divide. What content on the Internet is going to help a subsistence farmer in sub-Saharan Africa increase crop yields or market prices? Relevant projects require not only access to the Internet, but creation of the content to make the Internet useful - without the hard work to identify or create relevant information, it's very hard to get projects to sustain themselves, as people are not willing to part with scarce income for information they don't need.

As a result of these challenges, there's a long track record of "pilot projects" in poor communities which die shortly after funding ends, and an increasing perception that the digital divide is going to be much more difficult to bridge in the developing world than in the developed world.

I should pause for a moment and mention that Geekcorps was, in some circles, quite controversial. This had less to do with our odd name or peculiar founders, but for the simple reason that we systematically broke the cardinal rules of the ICT4D field. We worked almost exclusively in urban areas with for profit businesses. A consequence of this: our "beneficiaries" were middle or upper class citizens, rather than the desperately poor. And very little of our work had a direct development impact.

Slide 12
Instead, we focused on a philosophy we called "Digital Independence". Our goal was to develop sufficient technical talent in a country that there would be a community of programmers, designers, network engineers, system administrators and others who would be able to take on whatever IT challenges arose. As governments decided to computerize, they'd be able to use local talent, rather than exporting hard currency to US-based IT firms. And as creative ICT4D ideas came about, local firms would be able to execute the projects.

A result of the digital independence philosophy: we had a very different view of the digital divide than many of our colleages. We found ourselves working with a number of companies that were making lots of money bridging the digital divide, and creating original technologies in the process. Two quick examples from Ghana:

Slide 13
Busy Internet is the largest cybercafe in Ghana, and likely one of the largest on the continent. It's the brainchild of Mark Davies, a British net entrepreneur, who partnered with Ghanaian businesspeople and funders to turn a former gas bottling plant into one of the contemporary landmarks of dowtown Accra. There are approximately 150 flatscreen terminals in three rooms, usable for about 80 cents an hour... and the business is full 24/7/365, usually with a long waiting list to use the terminals.

Busy figured out how to target two markets that no one thought existed in Accra. About a third of the bottom line for Busy comes from entrepreneurs who rent space within the building. They're attracted by good connectivity, reliable electricity, a prestigious address and a community of other technical people. The remaining revenue comes from average Ghanaians who have enough disposeable income to go online. There turn out to be many thousands of people who've got that much disposeable income, even though GDP per capita in Ghana is under $1000. One of the smartest ways to price cybercafe access is to price an hour's access at the cost of a bottle of beer. Beer's a commodity with a middle-class audience - one of the ways we know cybercafe and cellphone useage in Ghana are growing is that they're cutting into beer sales.

Slide 14
IDN is the third-largest ISP in Ghana. It's founded by Francis Quartey, who geeked for AT&T in New Jersey for a decade before returning to Ghana. In building an ISP, Francis realized that he couldn't rely on the phone system. So he took cheap WiFi technology, combined it with a customized Linux distribution and directional antennas and built an ISP that services a few hundred corporate clients... wirelessly! We're just starting to see wireless ISPs (WISPs) in the US - IDN has been running a WISP since 2000.

Slide 15
Two things have become clear to me after five years of working with companies like Busy and IDN:

- There's a ton of money to be made by bringing the next billion internet users online.
- The companies that bring the next billion people online will be innovative companies in middle and low-development nations.

It's not that multinational companies couldn't develop good products for the developing world - it's largely that we choose not to pay attention to the developing world - an issue I'll be focusing on in the second half of my talk.

This is an issue that should be of profound interest to US tech companies. According to market research firm IDC, tech markets in the US are going to grow at 6% a year. Emerging markets, like China, Russia, India and Brazil are expected to grow at 11% a year, as hundreds of millions of middle class consumers purchase cellphones, PCs and other electronics. There's no guarantee that the leading global tech brands are going to thrive in these countries. There's domestic competition, and that competition has the great benefit of knowing its market far better than multinational companies do.

Slide 16
The ilkone 800 mobile phone is a great example of what happens when local companies design for the local market. Designed by a Dubai-based firm for the Islamic market, it features a call to prayer timer, sunrise and sunset timers for Ramadan, a Mecca-sensing directional feature and the full text of the Quran. As the company's marketing documents advertise: "ilkone is a mobile phone that gives you access to your belief wherever you are".

American firms are at a huge disadvantage as we try to design technology for these markets. To meet the needs of the market, we need to understand the consumers we're trying to sell to. And, basically, we're idiots about the developing world.

This is not entirely our fault. Our media's badly broken and, as a result, developing nations are systematically undercovered.

Slide 17
Consider Japan and Nigeria side by side for a moment. They've got roughly the same population - 130 million for Nigeria, 127 million for Japan. They're both extremely newsworthy. Japan's got the second-largest economy in the world; Nigeria's got oil, plus ongoing ethnic and religious conflicts, which are always good for a news story. In the average week, Japan will appear in a US news source eight times more frequently than Nigeria. If you're getting your news from CNN, that ratio rises to 15 to 1.

I got interested in this issue a few years back when I discovered how hard it was to keep up with African news stories when I was in the US. About eighteen months ago, I began making maps of media coverage on a range of news sources - wire services like AP and Reuters, news websites like Google News, newspapers like the Washington Post and news channels like CNN and the BBC.

Slide 18
Here's a recent map of media attention for the New York Times. Countries in bright red are experiencing lots of media attention - 3% or more of the stories my scripts detected. Countries in blue are experiencing less attention - 0.1% or lower. You'll note that North America, Europe and the Middle East are bright red; Africa is a patchwork of blues and pinks, Central Asia is solidly blue.

This map is a little deceptive, since it doesn't consider population. For instance, Iceland and the Democratic Republic of Congo are both light blue. But there's roughly a quarter million people in Iceland, and over 50 million in the DRC. This next map adjusts for population - it basically assumes that everyone in the world is equally newsworthy - something I call the Andy Warhol hypothesis - and looks to see whether nations are over or under-represented in these terms.

Slide 19
Iceland has turned bright red - massive overrepresentation - and DRC deep blue. Most of Africa is blue, except for Sudan and Libya, both prominent in the news lately. India and southeast Asia, including the massively populous Philipines and Indonesia are blue, while red continues to sweep Europe and the Middle East.

As it turns out, population is a lousy predictor for media attention. I've checked for correlation between 50 different statistics tracked by the World Bank and media attention across a dozen sources. The best correlation - for all but one news source - is a nation's GNI - gross national income. The only other statistic that comes close is a nation's imports of foreign goods, which tracks GNI pretty closely.

Slide 20
So here's a map of coverage adjusted for GDP. You'll notice the vast majority of the map is cream-colored, suggesting that countries are fairly represented in economic terms. We're still overfocusing on the Middle East; we're now focusing disproprtionately on certain African nations that are very poor, or the subject of attention like Sudan. We're ignoring poor Japan and Finland, whose economic clout merits even more attention. But we're also undercovering Indonesia, Malaysia, Sri Lanka, Brazil and Poland, all emerging economies likely to be important in the next stage of globalizing the Internet.

Slide 21
While my research is fairly recent, the observation that the developing world is undercovered is not. In 1965, Johan Galtung and Mari Ruge wrote a brilliant paper, "The Structure of Foreign News", which looked at coverage in Norwegian newspapers of international news from Cuba, the Congo and Cyprus. They concluded, in part, that newsgathering choices involved conscious or unconscious judgements about "elite" and "underdog" nations and people - unsurprisingly, elite nations get reported on far more than underdog ones.

A decade or so later, Galtung and Ruge's research was used by a group of leaders of developing nations to argue, in UNESCO, that what the world needed was a "New World Information and Communications Order" - NWICO. Under NWICO, media in the developed world would help subsidize media in the developing world and would be required to provide more coverage of developing nations. Needless to say, newspaper and TV station owners in the US didn't care much for the proposal. Ultimately, the US, the UK and Singapore left UNESCO over the debate - the US didn't return to UNESCO until last year.

The temptation is to blame this attention gap on big media. After all, it's their fault that there are permanent news bureaus in major European capitals and single reporters tasked with covering all of Africa. If big media would only cover the world more fairly, we'd know more about Africa and be better positioned to invent the next Islamic cellphone.

This critique misses a critical point: the media in the US is driven - heavily - by market forces. It's greatly to the advantage of media companies to give Americans the news we want - or are percieved to want: news on the war in Iraq, ongoing conflict in Israel, occasional stories from Europe, and lots of domestic news. Ask the editor of a publication like Newsweek why the magazine has so little coverage of Africa and you'll hear, "Hey, we're the good guys - if we gave our audience what they wanted, you'd never hear about Africa".

Slide 22
One of the great things about the Internet - the world's largest data set - is that it's possible to test assertions like this. For the past couple of months, I've been identifying and mapping data sets that can be considered to be proxies for net user's interest in different nations. My favorite proxy at the moment is web searches. Overture, which sells ads on search engine results, makes it very easy to find out what people are searching for. This includes some interesting insights on what we think of when we think of certain nations.

The most popular search for Canada is a search for "canada pharmacy"; you'll note two themes (travel and sex) in the Brazil results. "Mike In Brazil", as it turns out, is a very popular .. ahem.. photographic site.

Slide 23
Looking at people's web searches, book purchasing on Amazon and mentions of foreign nations in weblog posts, there's no clear indication that our Newsweek editor is wrong - with some odd exceptions (Mongolia, Egypt) there's little difference between this map of last month's Overture web searches and the maps of media interest.

If we're not interested in the developing world, why should market driven media tell us anything about it? An interesting contrast - the only non-market driven mainstream media source I've studied is BBC, which uses license fees to generate revenue, and maintains a massive network of corresponents around the globe - they're the only media source I've studied whose coverage correlates more closely to population than to national wealth.

I began this talk discussing the digital divide - a divide that prevents people in developing nations from accessing the Internet and other communications tools. I'm now suggesting that we're the victim of a different divide, one of our own making. Despite technological infrastructure that makes it possible for us to know about events anywhere in the world, we're victims of an attention divide. Because we pay very little attention to the rest of the world - especially poor parts of the rest of the world - we have a very hard time paying attention to those parts of the world.

Slide 24
This is problematic not just because we're in a poor competitive position to design PCs for Chinese users or cellphones for the world's 1.2 billion Muslims. It matters because the world - post 9/11 - is a very scary place, and many of the scariest places are countries we pay very little attention to. You may remember, immediately after the World Trade Center bombings, a great deal of focus on the weak and failed states in Central Asia - there was a brief moment where you could find articles on Uzbekistan in the mainstream media. You may remember from a few moments ago that Central Asia was uniformly blue on all of my media attention maps. Islamic fundamentalists and failing states haven't dissapeared from the region - we're just not paying as much attention to them.

A bipartisan commission recently produced a report titled "On the Brink: Weak States and US National Security". It helpfully pointed out that the failed states of Sudan and Afghanistan allowed Al Qaeda to organize to the point where they were capable of striking the US, and that there are two to three dozen states likely unable to control their own borders to prevent drugs and arms trafficking. You'll be unsurprised to find out that most of the states the report considers "On the Brink" have extremely low media attention.

Slide 25
One of the bright spots in American media over the past year or so has been the rise of weblogging. Bloggers have helped bring down Trent Lott, kicked sand in Dan Rather's face, and are now trying to answer the question of whether the President gets his instructions from a hidden radio receiver in his ear. They've become a part of the media cycle, fact checking stories and focusing attention on details they find interesting. Many journalists want them to go away; others watch them carefully to see if they turn up interesting stories.

Slide 26
Some blog fans support the idea that blogs will replace media as we know it. We'll get our news from "citizen journalists" who cover stories in their communities from their own perspectives. While this sounds far-fetched, OhMyNews has become one of the most popular news sources in South Korea using just this model. The website and newspaper are coproduced by professional editors and citizen journalists.

There's not a lot of evidence that the blogosphere is going to change international news... yet. Most bloggers write about US politics and technology. Blogs are a selective amplifier for mainstream media - they reinforce some stories and drop others, and right now, they're mostly reinforcing domestic news.

But that's the nature of peer production. When open source software developers build software, they build the tools they need and want. When bloggers write, they write on topics they're knowledgeable about and interested in. When writers contribute articles to Wikipedia, the vast, million-article online encylopedia created by volunteer labor, they write what they know... which means that Wikipedia has more reference information on Middle Earth than on Central Africa.

Slide 27
The only way to correct this bias in peer production is to change the peer group. If you want bloggers to write about Africa, you need African bloggers... and you need American bloggers who are interested in Africa and willing to build bridges between their communities and other online communities.

And this, in turn, requires some serious work on the international digital divide. We need to approach the problem differently. We're not delivering critical bits to data-deprived nations - we're communicating. The magic of the Internet is that it's bidirectional. We can recieve as well as broadcast. We can hear as well speak.

We need the Internet in Africa not just because Africans need the Internet but because we need Africans on the Internet. We need our attention divide bridged just a surely as the developing world needs a bridge to the digital divide.