From the perspective of sub-Saharan Africa, I agree with the analysis of Porter, Isser and Berg that the standard approach to Justice and Security (J&S) in ‘fragile and conflict-affected states’ has been ineffective. An initial drive by peacekeepers to curtail the violence and restore law and order, followed by an intensive training and equipment programme for the state security forces so that they can take over, has proved both over-ambitious and under-ambitious: too ambitious to expect to reform in a short period, poorly managed and trained J&S institutions that have a history of corruption and abuse; under-ambitious in not looking beyond the state providers nor looking at the structural causes of injustice and insecurity. As regards the structural factors, I concur that the programmes have largely addressed ‘symptoms without engaging in the political, social and economic dynamics and processes that produce insecurity and injustice stresses, and that give rise to the institutions that either successfully manage or exacerbate these stresses.’ Yet I am not so sure that the ‘lack of attention to politics’ has been from donor unawareness, as the authors imply. There is certainly talk about the importance of politics (in its narrowest sense) amongst in-country donor agency personnel, but their ability to translate it into practice often fails for very prosaic reasons. Staff struggle to get access to over-busy national Ministers; they are directed by external design teams made up of practitioners who are more conversant with their own speciality than politics; the background of the J&S team officers within the civil service ensures a preference for clearly defined sectors and for J&S issues to sit in familiar J&S institutions; retention of donor institutional memory and even of ‘good ministerial contacts’ is affected by rapid donor staff turn-over and poorly maintained central recording systems; and implementation of J&S is weakened by outsourcing to external implementers who are locked into contracts where politics gets over-looked in the specified methodology and J&S outputs.