Geek Heresy (39 page)
Authors: Kentaro Toyama
19
.
This generalized conception of technology is not rare. Economists, for example, routinely speak of structures of human organization as a kind of technology. However, since most people think of technology as being a physical artifact, I use the phrase “packaged intervention” as its generalization.
20
.
In case you believe these are rare or short-term effects, consider the different trajectories of post-Soviet countries. Lithuania, Latvia, Estonia, and possibly Ukraine have made a lasting transition to democracy, but the rest have backslid into dictatorships, both real and virtual, despite initial elections. Even in democracies with peaceful, elected turnovers of power, long-held attitudes can impede governance. When I was in India, I found its claim to be the world’s largest democracy to be a bit of an exaggeration. What I saw was more like a feudal system with term limits. Politicians act like dukes and barons when in power, and most citizens happily oblige, bowing low to officeholders and paying
baksheesh
for routine government services. Everyone knows that government employees have an informal income stream, but by many it’s accepted as a privilege of power – even one to aspire to. And everyone knows that everyone knows. In 2010, when B. S. Yeddyurappa, then the chief minister of Karnataka, was taking heat for excessive government corruption, he reprimanded his own administration with a wink
on the public record
: “Let all of us stop making money for ourselves. All of us should now work for Karnataka” (IBNLive 2010).
21
.
Seymour Lipset (1959, 1960) was among the first to argue that a range of socioeconomic properties appear to encourage democracy, an argument to which I will return in Chapter 9 in a modified form. Political scientist Robert Dahl (1971) focuses on eight institutional requirements for democracy, among which are political parties, the right to run for office, a free press, associational autonomy, the rule of law, and an efficient bureaucracy.
22
.
See, for example, Achebe’s (1977) takedown of Joseph Conrad’s
Heart of Darkness
.
23
.
Achebe (2011).
24
.
Atlantic
(2012).
25
.
Porter (2013) reports that women of prime working age earn only about 80 percent of what their male peers earn.
26
.
The laptop-as-vaccine statement was made by Negroponte (2008) at a TED talk about One Laptop Per Child. He repeated the same claim when he and I were on a panel at MIT (
Boston Review
2010). He must have felt that the analogy resonated.
27
.
From the Global Polio Eradication Annual Report (World Health Organization 2011). It’s understandable that polio eradication efforts go poorly in areas with open conflict, such as Afghanistan or Nigeria. But even where violence is less frequent, as in western Chad, the report notes, “operational issues are the main reason children are still being missed by vaccination campaigns, although social and communication problems are also important, particularly in key high-risk areas.”
28
.
It is generally agreed that smallpox is easier to eradicate than polio because smallpox always results in visible symptoms. With polio, for every person who shows symptoms, hundreds of others carry and spread it without symptoms. Thus, the only sure way to eradicate polio in a given region is to vaccinate everyone in it. How feasible that is depends not on technology, but on the reach and quality of government administration and health-care institutions.
29
.
Yunus (2011).
30
.
Bloomberg Businessweek
(2007).
31
.
Yunus (1999), p. 205.
32
.
Ibid., p. 140.
33
.
Consultative Group to Assist the Poor (2008).
34
.
Banerjee et al. (2010).
35
.
Tripathi (2006).
36
.
Fears of sterilization occur regularly in the developing world, probably based on a past history of forced vasectomies in some countries (Population Research Institute 1998). Similar fears resurfaced recently in Pakistan (e.g., Khan 2013) and in Kenya (e.g., Gander 2014). As for vaccine fears in the developed world, see, for example, Mnookin (2011).
37
.
It can sometimes be helpful to project what would happen with technologies of the future. Put aside your conscience, and imagine a future in which a technology called “the Avatar” becomes available. It is a special chip that can be surgically implanted into a person’s spinal cord just below the cerebellum, enabling the total hijacking of a person’s voluntary muscle system via wireless command. (The spine’s information bandwidth is probably around 16 megabits per second, only slightly more than is promised by current 3G networks.) Health-care workers could be
turned into remote-control puppets who visit assigned households faithfully, execute engagement scripts flawlessly, and never give in to laziness or corruption. In other words, this is a technology that would overcome all the pesky problems of messy human behavior. Yet, as powerful a technology as the Avatar might be, even
it
would require a good system for delivery and implementation. Not only would the technology have to be surgically installed on a person-by-person basis, but someone would have to maintain the technology, restock broken units, monitor performance, and deal with the predictable problem of subversive hosts seeking to rid themselves of the devices. And each of these activities would require solid implementation. In short, there would have to be ongoing institutional support beyond the packaged intervention itself, and at a much higher price than the cost of hardware. Of course, this dark potential future is only satire, but the point is that even an absolute technology requires strong human implementation to work. Similar lessons naturally hold for far less powerful packaged interventions.
38
.
American Sociological Association (2006).
39
.
Rossi (1987).
40
.
There is an implicit fourth problem that Rossi mentions but resigns himself to. It corresponds to what’s required of beneficiaries of packaged interventions. Rossi recognized that no program works without motivation and capacity among those whom it is meant to help, but he felt it was beyond intentional policy to address this deficiency, saying, “It is likely that large scale personality changes are beyond the reach of social policy institutions in a democratic society.” This is a critical point that is addressed in Part 2 – I believe he gave up too easily.
41
.
Yunus (1999), p. 140.
42
.
Ibid., p. 205.
43
.
Wikipedia (n.d.), “FINCA International,”
http://en.wikipedia.org/wiki/FINCA_International
.
45
.
Opportunity International (n.d.).
46
.
Yunus (1999), pp. 135–137.
47
.
Based on data available at MixMarket (2014). The estimate is low because it includes only organizations registered with the exchange at the time and excludes microcredit activities in the developed world.
48
.
Heeks (2009).
49
.
Organisation for Economic Co-operation and Development (OECD) (2014a). The figure cited includes all bilateral aid from OECD countries and multilateral aid from organizations like UNICEF.
50
.
See Fraser (2012) and Plumer (2012) for critical reporting on cookstoves.
51
.
The performance of charter schools is an ongoing debate, but it seems fair to say that the results are mixed. A good summary, with references to primary
research, is offered by Ravitch (2011), pp. 138–143. Ravitch’s book, incidentally, offers a brilliant argument against packaged interventions in American education, based on her life’s work as an educator and education researcher.
Chapter 5: Technocratic Orthodoxy
The Pervasive Biases of Modern Do-Gooding
1
.
George Packer (2014) cites an estimate of $5.25 billion in book sales for Amazon in 2013, and Milliot (2014) estimates total book sales at $15 billion. Packer also notes that in 2010, Amazon captured 90 percent of e-book sales.
2
.
The reference to Orwell was pointed out by Streitfeld (2014). The quotation is from Orwell (1936).
3
.
Barnes & Noble Booksellers (n.d.).
4
.
Thompson (2010) offers a rich history and analysis of the book industry in the four decades since about 1970.
5
.
Thompson (2010), pp. 389–392, describes this trend in detail and calls it a “winner-takes-more market.”
6
.
In respective, eponymous books, Chris Anderson (2008) describes the long tail, and Robert Frank and Philip Cook (1996), the winner-take-all society.
7
.
Duflo et al. (2012).
8
.
In another paper with different colleagues, Duflo herself writes, “[The] recent evidence suggests that many interventions which increase school participation do not improve test scores for the average student. Students often seem not to learn anything in the additional days that they spend at school” (Banerjee et al. 2007). As the next few paragraphs explain, if you accept the generalized conclusions implied by the two papers, there is a contradiction that isn’t addressed by either one.
9
.
The authors note in a footnote that tamper-proofing only meant placing “heavy tape” over the controls.
10
.
Duflo et al. (2012). “Nonformal education” is what Seva Mandir calls its program, in contrast with the formal government education system for which the children are being prepared. Despite the name, the pedagogy is formal and modeled on good classroom teaching.
11
.
Abdul Latif Jameel Poverty Action Lab (n.d.).
12
.
In addition, many of the best RCTs are overseen by high-caliber researchers like Duflo – yet another unusual circumstance that doesn’t come with the packaged intervention itself.
13
.
For excellent tips on good classroom interaction, see Doug Lemov’s (2010)
Teach Like a Champion
. I have found his book to be invaluable for teaching K-12.
14
.
When an RCT carried out in partnership with Organization X results in data suggesting that Program Y has an impact, what it proves is not that Y has
impact on its own, but that Y has impact if carried out by an organization like X. All too often, X and Y are
both
necessary for impact, just as Seva Mandir’s efforts and the camera monitoring were both needed for impact. In their conclusion, Duflo et al. (2012) consider whether their results would generalize to other schools, and they optimistically suggest that it would. They write, “Our results suggest that providing incentives for attendance in nonformal schools can increase learning levels.” There is just one nod to external vailidity in the paper. They mention, “The question arises, however, as to whether incentive programs can be instituted for government teachers, who tend to be politically powerful. It may prove difficult to institute a system in which they would be monitored daily using a camera or similar device. Our findings suggest, however, that the barriers currently preventing teachers from attending school regularly (e.g., distance, other activities) are not insurmountable. Given political will, it is possible that solutions to the absence problem could be found in government schools as well.” The wording is careful not to overreach, yet in the way of technological optimists, they hide any reservations behind the word “can,” as noted by Toyama (2013a).
I exchanged emails with Duflo on June 24–25, 2014, to seek her response. In the hyper-confident tone characteristic of many economists, she wrote, “I absolutely stand by the conclusion of the paper, and I don’t really see why the fact that Seva Mandir was trying to improve (all) their schools in other ways as well makes it less externally valid than if they were not.” If Seva Mandir schools differ from other schools, obviously, it has an impact on external validity. Our differences, then, come down to whether Seva Mandir’s teaching and management are atypical. I thought they were exceptional in comparison to many rural government schools. Duflo dismisses the possibility. She says Seva Mandir’s teaching is “not so spectacular.”
My larger critique with RCTs is not with the methodology itself, but with the tendency of experimenters to generalize too confidently and to be too cavalier with external validity.
15
.
Banerjee and Duflo (2011), many places, but, for example, p. 272.
