I thought this was a very interesting and readable paper about aid effectiveness by François Bourguignon and Mark Sundberg. They describe an emerging consensus for the aid architecture, based on what they call implicit aid contracts:
The use of implicit aid contracts, based on monitorable evidence of improvement in final results and the observable quality of policies, is the direction in which donors are beginning to move. Aid allocation is increasingly done on the basis of country performance that combines governance, general policy environment and some intermediate or final results. The use of the CPIA index, although imprecise in its coverage, is a move in that direction. The selectivity of aid allocation based on the quality of governance and general policies, has increased significantly since the mid-1990s, particularly among the multilateral institutions but also among bilateral agencies, possibly pointing to the emergence of a new model. Consistent with this trend is the donor commitment to make aid more predictable, and deliver more aid as budget support rather than tying it to specific imports, projects, or policies.
I agree with that, though I think we may go further than this implies towards linking aid to ex post outputs, and rather less to measures of governance and policy environment.