This is a woefully inadequate way of displaying the content of additional notes and annotations. I have canceled plans I had to move a great deal of content to Medium. I had thought Medium would be a great tool for bridging between academic research and general readers but the lack of the ability clearly annotate completely undermines that use.
Many of the interviews cite frustrations with observational data and instrumental variables as a major factor in the rise of RCTs. See in particular the interviews with Michael Kremer, Lant Prichett and Jonathan Morduch.
For Kremer and Miguel’s response, with others, to the Cochrane Review which Deaton alludes to, see: http://www.poverty-action.org/blog/cochrane%E2%80%99s-incomplete-and-misleading-summary-evidence-deworming
Note that the Cochrane authors respond in the comments.