**Summary:** This paper presents a method for cost-effective outbreak detection in networks, applicable to both water distribution systems and blog networks. The goal is to select nodes (sensor locations or blogs) to detect the spread of information or contaminants as quickly and efficiently as possible. The key insight is that many outbreak detection objectives, such as detection likelihood and population affected, exhibit submodularity, a property that allows for efficient approximation algorithms. The authors propose an algorithm called CELF (Cost-Effective Lazy Forward selection) that leverages submodularity to find near-optimal sensor placements. CELF is significantly faster than a simple greedy algorithm, achieving up to 700 times faster performance. The algorithm also provides online bounds on the quality of the placements, ensuring that the solutions are close to optimal. The method is evaluated on real-world data, including a water distribution network model from the EPA and real blog data. The results show that the algorithm achieves near-optimal placements, providing a constant fraction of the optimal solution. The approach also scales well, achieving significant speedups and storage savings. The paper also discusses the importance of considering different cost models for nodes (sensor locations or blogs), and shows how the approach can be adapted to various criteria, including detection time, population affected, and cost-sensitivity. The algorithm is shown to provide deeper insights into the applications, helping to answer multicriteria trade-off and generalization questions. The study demonstrates that the proposed method is effective in both blog and water network scenarios, and that the submodularity property is crucial for achieving efficient and near-optimal solutions. The results highlight the importance of considering both the structure of the network and the cost of node selection in outbreak detection problems.**Summary:** This paper presents a method for cost-effective outbreak detection in networks, applicable to both water distribution systems and blog networks. The goal is to select nodes (sensor locations or blogs) to detect the spread of information or contaminants as quickly and efficiently as possible. The key insight is that many outbreak detection objectives, such as detection likelihood and population affected, exhibit submodularity, a property that allows for efficient approximation algorithms. The authors propose an algorithm called CELF (Cost-Effective Lazy Forward selection) that leverages submodularity to find near-optimal sensor placements. CELF is significantly faster than a simple greedy algorithm, achieving up to 700 times faster performance. The algorithm also provides online bounds on the quality of the placements, ensuring that the solutions are close to optimal. The method is evaluated on real-world data, including a water distribution network model from the EPA and real blog data. The results show that the algorithm achieves near-optimal placements, providing a constant fraction of the optimal solution. The approach also scales well, achieving significant speedups and storage savings. The paper also discusses the importance of considering different cost models for nodes (sensor locations or blogs), and shows how the approach can be adapted to various criteria, including detection time, population affected, and cost-sensitivity. The algorithm is shown to provide deeper insights into the applications, helping to answer multicriteria trade-off and generalization questions. The study demonstrates that the proposed method is effective in both blog and water network scenarios, and that the submodularity property is crucial for achieving efficient and near-optimal solutions. The results highlight the importance of considering both the structure of the network and the cost of node selection in outbreak detection problems.