What Amazon.com and Netflix can teach us about fighting poverty

An intriguing article from Foreign Policy about fighting global poverty. Here's an excerpt:
"Unfortunately, the word "poor" has lots of opposites, and not all of them have to do directly with money. "Healthy," "well-educated," "having access of clean water," and "nourished" are among the many opposites of "poor," and when we think about the relative merits of antipoverty programs, we have to weigh each of these things -- and more -- against each other. But how do we compare the importance of, say, health versus education versus housing? And how do we make tradeoffs between them? One approach is to apply our own values and priorities, but this ignores the preferences of the very people for whose benefit these programs are designed. This happens often in the world of development aid; a donor focusing on education, for example, might care more about classroom quality than hospital beds. But wouldn't it be better if we could instead ask the people receiving our help what they want?
This isn't just about trying to please. Development aid lore is rife with stories of well-intentioned outsiders missing the mark, offering people goods and services they don't really want. Recipients sometimes manage to extract some value from unwanted items by trading them for things they actually do want, or by jury-rigging them to serve other purposes (often with limited success). A mosquito net may get swapped for a machete, for example, or a kitchen set might be sold in order to just buy food. If we want to avoid these outcomes, we must answer the question: How can we best understand people's priorities and tastes?
Outside the poverty field, there are a growing number of ways of ascertaining and predicting what people like, and all of them are imperfect -- but they're getting better. Think about the ubiquitous taste-based suggestions on Amazon.com and Netflix, for instance: "Customers who liked this also liked _____." This "taste-matching" approach looks for other people whose preferences are similar to yours, then recommends things those people like that you haven't tried yet. It stands to reason that taste-matching methods improve over time; they look at thousands of consumers' feedback about thousands of products, and see what patterns emerge. As more wide-ranging data is amassed from more consumers to inform these suggestions, they become increasingly accurate..."