The article is based on two qualitative research projects held in Rio's favelas. The empirical material is drawn from one of its activities - conducting and recording approximately 50 hours of a process that has been denominated as "trust collectives". This expression intends to stress the way participants of 15 focal groups (involving 150 residents of more than 40 favelas) were recruited - acquaintance and trust built through previous relationships with at least one of the researchers. This criterion was planned as a way of minimizing the expected effects of the "law of silence" imposed by agents of violence in these localities. The text analyzes enunciations collected during these debates, which were clearly concentrated on the different ways in which criminal gangs and police forces make their presence and behavior felt within Rio de Janeiro´s favelas. The authors sustain that, less than questioning criminal and police violence as a whole, favela residents are intensely worried about some of its manifestations which prevent them from carrying out their daily activities. Without being directly taken up as a theme for dialogue, this is the horizon that calls out their attention and that organizes all their critical comments to "people outside", as well as their attitudes and behavior towards their fellow favela inhabitants. The authors go further by arguing that, in their descriptions and denouncements, residents do not homogenize the agents they consider responsible for destabilizing their routines. On the one hand, there are institutional police practices (the famous "operations" - recurrent raids on favelas in order to fight drug dealers) and the arbitrariness of agents' regular behavior, both of which are felt to be almost completely unpredictable. On the other hand, when dealing with the violent practices of drug dealers, with whom residents are obliged to share their territory, there were many accounts of attempts (successful or not) of reducing the unpredictable quality in the flow of daily life. This occurs by adjusting behavior to a calculation - unworkable and/or ineffective in the case of police action - of the risks involved in such a situation of forcibly sharing the same space. This calculation, making a small part of local violence easier to "administrate", permits re-absorbing it as an "extraordinary" part of "ordinary" routines.
violence; crime; police; routine; risk; sociability