Harsh. The costs of data-driven development

OK, here it is at last, the fruits of 1000s of hours of research, and argument, suggesting that money spent on HIV/AIDS is not the most effective use of donor funds: 

On the grounds of Uganda’s biggest AIDS clinic, Dinavance Kamukama sits under a tree and weeps.

Her disease is probably quite advanced: her kidneys are failing and she is so weak she can barely walk. Leaving her young daughter with family, she rode a bus four hours to the hospital where her cousin Allen Bamurekye, born infected, both works and gets the drugs that keep her alive.

But there are no drugs for Ms. Kamukama. As is happening in other clinics in Kampala, all new patients go on a waiting list. A slot opens when a patient dies.

The cause of course is the drop in donor funding for anti-retrovirals and for treatment programs in developing countries. Everyone from BMGF to the US government is reducing, or limiting increases, in their funding for HIV/AIDS programs. Meanwhile, in Uganda....

 500,000 need treatment, 200,000 are getting it, but each year, an additional 110,000 are infected.

“You cannot mop the floor when the tap is still running on it,” said Dr. David Kihumuro Apuuli, director-general of the Uganda AIDS Commission.

Believe me, I understand the benefits of adjusting policy and priorities, especially when each AIDS patient treated with anti-retrovirals costs US $11,000. There are a lot of simpler, cheaper and more effective healthcare efforts that can be funded with some of that money. 

And I don't want to harsh on the people who are pushing for more informed decision making, and particular for decision making informed by randomized field trials. However, I do think that _at this point_ groundbreakers such as Esther Duflo could present more balances and realistic pictures of the cost and benefits of development decisions driven by data: