Research shows that remote rural communities need both internal and external links for their sustainable development. The question is what can be done about that? Can outsiders reinforce local social networks in a controlled manner? We invited randomly selected members of 16 Sumatran agrarian communities to three-day training and networking events outside of their villages. In one of the events, the participants from different villages were jointly transported from Sumatra to Java, where they learnt about sustainable agricultural practices and socialized with each other. Eighteen months later, we surveyed the information-sharing networks in these communities and analysed these 16 networks by Exponential Random Graph Models. The network structures vary widely among the communities, which demonstrates the limitations of cases studies of single networks and highlights the need for more comparative collective-level studies of multiple networks. However, some general patterns emerge. It was found that in general, local inhabitants who have more extra-communal links, typically seek less advice in their own communities but are sought more for advice by other community members. (Although, these effects are statistically significant only in two and three communities respectively.) The randomly selected participants in the external networking events have overall become more popular as informal information-providers in their communities. In particular, the participants of the three-day agricultural training program in Java were one and half year later still significantly more sought for agricultural advice in six of these agrarian communities. Small external interventions can make strong lasting impacts on local social networks and these can be now rigorously measured by new tools of network science.