A randomized controlled trial (RCT), if properly executed, ensures ex-ante equality of opportunities to all subjects involved in the experiment. However, it generates ex-post inequality in outcomes to the extent that an intervention program brings positive and significant impacts, and this inequality can be socially costly. Such inequality has been often overlooked or intentionally left unattended to observe longer-term impacts. To examine whether our intervention can create an intended positive impact on the society and whether “researcher-made” inequality disappears over time, we implemented a unique field experiment in Cote d’Ivoire where farmers who received training on improved agronomic practices were requested not to transmit information with control peers in the first year, but were encouraged to do so in the second year after the intervention. We found that treated farmers performed better immediately after the intervention. This corresponds to the experimental phase when the stable unit treatment value assumption (SUTVA) was imposed, suggesting that the training was indeed effective. Interestingly, control farmers successfully caught up with treated farmers after information exchanges were allowed in the second year. Our detailed social network analysis confirmed that information flow from treated to control farmers was less active initially, but became active after encouraged to do so. Our findings suggest that “researcher-made” inequality that was generated to precisely estimate the treatment effect, but possibly socially costly, can disappear by providing subjects with the opportunities of social learning, although it is generally recused in the RCT execution because it violates the SUTVA. We expect that this research design is more acceptable to potential beneficiaries and practitioners.