Op-iccj130030 1..37

Decision-making about innovative change in high-risk networks is exceptionally difficult because system failure may result in catastrophe. We adopt a historical method to compare the US and Soviet choices in their nuclear attack submarine programs between 1970 and 1996 and to surface their complex political, technological, and operational relations. One program achieved high reliability in the face of innovation while the other did not. Actor network theory (ANT) helps illuminate the interactions and resulting innovation paths and dependencies. We study how (i) open communication and power dispersion across high-risk networks influence system reliability, (ii) individuals spanning multiple groups within the network generate dominant coalitions, and (iii) strong safety advocates impact the network.