The respondents raise three broad objections to my analysis of ICT4D and also offer a springboard for discussing policy. So, I will interleave rebuttals with my own recommendations, which differ slightly from Evgeny Morozov’s. Although I agree with his abstract suggestion for realistic, context-sensitive, holistic approaches, I have even more direct recommendations for those considering ICT4D interventions.

Objection 1: It’s true that technology alone is not enough, but if it’s designed well, it has great potential.

I agree with Jenny Aker, Nathan Eagle, Archon Fung, and Christine Zhenwei Qiang that technologies must be designed appropriately. Tales of international development often feature rusting tractors or hospital equipment that were not appropriate for their context. But the question of whether to use a technology should precede that of how to design a technology. The best-designed educational technology, for example, will have minimal educational impact in failing schools.

Fung’s dream of technology designed to improve the lives of the poor is fantastic, and some engineers believe this is what they are creating. But is it realistic? By definition powerful technologies serve multiple purposes, and inevitably, the rich, skilled, and socially connected will make better use of them than the rest. Fung hits closer to the mark when he considers applying technologies to support public goods. That brings us to the second objection.

Objection 2: Technology for technology’s sake is pointless, but how about technology for an acknowledged development goal, such as education?

Computers can benefit good schools, but they can’t make up for poorly run schools and absent teachers. On this point, I disagree fundamentally with Nicholas Negroponte. No one prescribes innovative software for the employees of a failing business. So, why laptops for broken schools, or mobile phones for badly run rural health care?

Charging in headfirst with technology to repair human problems simply doesn’t work. Fung mentions logging public grievances electronically, but it doesn’t matter how many complaints are logged online if the government neither intends nor budgets to address them (the bane of so many e-government programs). The application of technology to progressive ends also assumes political commitment, but this is again, a problem of intent, not of technology.

Mark Warschauer, arguably the world’s foremost expert on technology in education, writes of U.S. schools: “The introduction of information and communication technologies in . . . schools serves to amplify existing forms of inequality.” He continues:

Placing computers and Internet connections in low-[income] schools, in and of itself, does little to address the serious educational challenges faced by these schools. To the extent that an emphasis on provision of equipment draws attention away from other important resources and interventions, such an emphasis can in fact be counterproductive.

I agree with Dean Karlan on the importance of impact evaluations of development projects. Leigh Linden, an economist who ran randomized trials of PCs in Indian and Colombian schools, found that PCs that supplement teacher-led instruction help; PCs that substitute for teachers hurt; and implementing a large-scale PC program has little effect on educational outcomes, apparently because teachers don’t incorporate computers into their curricula.

If they could and did, their efforts would be an example of the human capacity that technology could magnify. But the irony of technology for development is that precisely where we’d most like to have impact, reliable human institutions (both organizations and social norms) are most lacking, and therefore, technology’s potential is most limited.

So, to those who are committed to using technology for development—whether because of funding, job description, or temperament—my first recommendation is to seek out and understand those institutions that already yield positive outcomes, and then design technologies that magnify their force. Otherwise, be prepared to build the human capacity yourself.

Objection 3: But there are cases where technology has positive impact, and you can’t ignore billions of mobile-phone accounts! Technology must be good!!!

Aker, Eagle, Negroponte, Qiang, and Ignacio Mas all express this view, though with varying volume. I don’t deny that technology can have positive effects. Of course, it can. The wealthy, educated, and powerful regularly benefit from technology. So, too, do poor and marginalized people when the technology augments capable, determined champions, as sometimes happens in development projects.

Consider Aker’s description of the benefits of cell phones for adult literacy in Niger. She implies that cell phones themselves accomplished that feat. However, in her peer-reviewed paper on this project, she explains that the technology intervention supplemented an intensive eight-month literacy program run by a nonprofit, Catholic Relief Services (CRS). Sure, the mobile phones helped, but it was CRS’s efforts that they magnified. We shouldn’t conclude from Aker’s experiment that mobile phones enhance literacy. Rather, mobile phones helped an effective literacy program do better. This is a subtle but earth-splitting difference: the former implies that we could eliminate illiteracy by increasing technology penetration; the latter that technology is useless if it is not built to supplement effective adult-literacy programs.

Those critics who bring up the high cell-phone count in the developing world also ignore the fact that large-scale demand is not proof of value to society. Yes, mobile phones have been found in some contexts to increase economic efficiency, but the effects were limited, examples are few, and it is not clear how the positive impacts balance against the negatives. Understanding total impact requires the kind of diffuse, larger-scale analysis that is notoriously difficult to get right. The jury is still out on mobile phones.

Even when we see positive impacts from technology, we have to ask ourselves what really generated the outcomes. Looking deeper, we find positive human intent and capacity already in place. Mas himself observes that none of the four factors contributing to the success of M-PESA are “technology- related.” M-PESA amplifies an existing custom of rural remittances. Whether M-PESA and its extensions lead to better financial management among the people who use it will depend on whether the kernel for those behaviors already exists, and whether there is sufficient investment in non-technological financial education. Expecting an extension of technology to create intent is wishful thinking.

Thus my second recommendation: when deciding how to allocate resources between technology and human capital, invest first in the factor that is most lacking. There may be times when a technology investment makes sense, but for the world’s poorest countries, human capital, not technology, needs the boost first. And information technologies that amplify knowledge, in particular, can wait. Schools need better administration, clinics need more reliable medical staff, and individuals need more education and vocational skills. Especially now that mobile phones are everywhere, perhaps we can finally focus on human capacity.

Does my caution about technological solutions for global poverty mean that I think we should inhibit the market for technology? No, in this I agree with Negroponte. Allowing something, however, is not the same as advocating for it. Pro-choice is not pro-abortion. Technology is fine, but international-development efforts need not spur their dissemination, nor should development-minded people insist on technological solutions. We should focus on those advances that free markets have difficulty providing: the human and institutional foundation for universal health care, universal education, agriculture extension, better governance, and citizen empowerment.

One final note. In the last four decades, the United States has undergone a boom in information and communication technology. The PC and cell phone were invented; the Internet stormed onto the stage; and corporations such as Apple, Microsoft, Google, and Facebook penetrated every corner of our lives. If technology cured social ills, then we’d hope that during the golden age of innovation in a technologically advanced country, there would be some dip in the poverty rate. Yet in the same four decades, the rate of poverty in the United States stagnated at around 13 percent, embarrassingly high for the world’s richest country. Either today’s Americans have not prioritized poverty reduction, or the world’s best technology is running as fast as it can just to keep us in place. If so much technology didn’t dent poverty in America on its own, why do we expect anything more in countries with far less ability to capitalize on it?

It’s not that technology is powerless or irrelevant; it’s that technology is not the problem. Technology is just a tool; its impact depends on how it’s wielded. If tool after fancy tool doesn’t build a better house, maybe we should invest more in the carpenter.