Can human-powered devices solve the electricity gap for the millions of rural Africans adopting mobile phones? Findings from our long-term evaluation of two personal crank-based charging systems in Kenya reveal that small hand and leg-powered devices do have potential to meet the needs of rural mobile phone users. Unfortunately, device breakage, theft and incompatibility with handsets, coupled with lack of consumer credit and poorly functioning markets for these goods mean these represent only a partial solution to the mobile phone charging problem. Drawing from our fieldwork, we motivate a HCI4D/ICTD design and evaluation agenda that better accounts for unique individuals' geographic, financial, and economic circumstances or their human computer ecosystem. Key strategies for implementing this agenda are engaging with diverse users on their own terms and conducting long-term qualitative evaluations to reveal how acceptance and usability change over time.