Mike Best with evidence that ICT4D works…

If the bright lights of information and communication technology for development were rock stars, Mike Best would play sell-out shows and be followed by groupies. Instead, he gives great talks at Berkman and only a dozen or so of us are lucky enough to be in the room. (I held up a cigarette lighter for most of his talk.) Co-editor of the leading journal on ICT4D, assistant professor at Georgia Tech and fellow Berkman fellow, Mike has done both great fieldwork and theoretical work, trying to answer the question, “How can information technology improve lives in the developing world?”

Mike opens his talk with a quick overview of ways technology theorists have thought about the impact of ICTs on political development. He cites Cutright, who posited in the 1960s that voice telephony was a better predictor of political development than economic development; Skolnikoff, who postulates that, on balance, ICTs support greater openness in civil society; and Kedzie, who puts forth the “dictator's dillemma” – that an increase in connectivity neccesarily means a decrease in authoritarian control.

He spends a bit more time on Benjamin Barber's concept of “strong democracy”. (I got a very rough introduction to Barber two weeks ago, when he lambasted me and my colleages in Madrid.) Barber puts forward three ways technologies might work to strengthen or weaken democracies: the Panglossian, Pandorian and Jeffersonian options.

In the Panglossian scenario, citizens are primarily consumers and they blindly respond to the ideas put forward by stronger actors – most likely corporations, who keep them from political power by keeping them sufficiently entertained. In the Pandorian model, information tools become Orwellian tools for obervation and liberating technologies become tools for repression. Only in the most optimistic, Jeffersonian scenario do information tools reach their potential of allowing open, participatory democracy. (When Barber yelled at us a few weeks back, he seemed to be suggesting that the Internet was following the Panglossian path.)

Mike has decided to test a theory put forth by Kedzie, that “multidirectional, reciprocal communication technology like email” are conducive to democracy. Kedzie argues that a 1% increase in this sort of network connectivity leads to a 4 point rise on a 100-point democratization scale (adapted from Freedom House's 7/14 point scale.)

Analyzing correlations between changes in the Freedom House scale and other independent factors – connectivity, schooling, per capita GDP, life expectancy – Mike sees a strong correlation between connectivity and high levels of democratization. The correlation is stronger between connectivity and democratization than schooling or life expectancy – only the GDP to democratization correlation is stronger. Controlling for GDP, Mike still sees a meaningful correlation between connectivity and democratization.

Correlation is not causation, as has been famously noted (though, it's also been noted that correlation's a good clue about causation…) and Mike is not claiming that increased connectivity increases democratization or that nations that democratize become more connected soon afterwards. Adding an “arrow” to the correlation this way requires doing time series work – as connectivity increases, does democratization? Mike's work on this so far hasn't given him convincing data showing causality.

What he has found, by doing analysis of covariance over time, is that there's a “tipping point” around 1995 where connectivity is more tightly correlated to democratization than it previously was. It's possible that a number of countries hit a key level of teledensity or data density around 1995 where we start seeing the effects of connectivity on democratization.

Mike offers two data models – one shows a simple relationship between connectivity and political rights: adding 250 internet users per 1000 is correlated to a one-point rise in political freedom on a seven point scale. Alas, getting internet connectivity to the quarter of a nation's population is a big step… and the correlation Mike sees (R2=0.3) isn't all that strong. He's found another (very peculiar) data model that explains almost all his data variation:

y = -.84 + .30(internet Users) – .39(phones)

In other words, high internet usage and low phone usage correlates to increased democratization. This is a bizarre result at first glance, but there are some interesting possible explanations. In low teledensity nations, an introduction of one-to-many connectivity might be closely correlated to democratization, since open societies are less threatened by the Internet. Before concluding that phones somehow “cause” totalitarianism, remember that these models are descriptive, not causal. That effect might well be explained by authoritarian nations with heavy phone use, like China or Singapore.

After giving us hope that there might be a demostrable connection between connectivity and democratization on the macro level, Mike takes us to the micro level – the SARI (Sustainable Access in Rural India) project that he's been working on for several years. SARI provides internet connections to 50 villages in the Madurai district of Tamil Nadu state in India. The kiosks are set up in villages of 300-1000 households where average per capita income is about $0.60 per day.

The kiosks provide a wide range of internet services, everything from email, online training programs, entertainment (horoscopes, movies and games are especially popular), televeterinary services, telemedical services (notably cataract diagnosis) and e-government services. Mike is most interested in the e-government side of things. The systems allow citizens to apply for birth certificates, old age pensions, community certifications (evidence that someone is part of an “untouchable” caste, which has government benefits associated with it), income certificates (evidence that someone is below the poverty line), as well as voicing grievances about government services.

The numbers of people using the kiosks for e-government services look small, at first glance – a few users per month. Mike points out that most of these certificates are someone one applies for once – there's not a lot of repeat usage. And, when Mike compares the number of certificates applied for from wired villages to unwired ones of similar size, in the same region, the results are dramatic: citizens in wired villages apply for birth certificates five times more often and for old age pensions three times as often.

The reason for the increased usage is pretty simple. It costs lots less for citizens to apply for these essential documents online than it does to get them in person. To get papers in person, villagers need to spend one or more days in transit, which entails expenses, and often need to pay bribes to get the essential forms. The total expense for getting a birth certificate, including travel and bribes, is often more than a person's daily income. That becomes a powerful incentive to learn how to use the Internet kiosks.

Mike's research comes at an interesting time in the debate over ICT for development. There's a real backlash against the idea that ICT projects in rural areas have a meaningful, positive effect – the Economist dedicated a substantial portion of their last issue to an argument that cellphone penetration was far more important than rural Internet access, and that rural Internet projects had mixed impact, at best.

While I largely agree with the Economist – cellphones are critically important, and most rural ICT projects have been badly thought out and their impact poorly measured – Mike's offering a great argument that rural ICT can have a meaningful impact IF people are smart enough to build applications that have direct benefit to users in the developing world.

This entry was posted in Berkman, ICT4D. Bookmark the permalink.