- iceberg methapor: system structure signal for action, no longer term - patterns of behavior (accumulation) e.g. johns hopkins dashboard - five steps of: problem articulation, dynamic hypothesis, testing, - think mechanistically: S and I (contact between infectious and susceptible): disentable when ambiguous: price and unit solde and revenue - to prove your model: physical meaning for each parameter, predictive check and SBC, hospitalized and testing, risk perception (remember for a long time), not only one but many time series (e.g. cases, other sources of data) is people's behavior more affected by stock or flow: accumulate or per unit ## Decision Is a 3x spike in cases in a few days a looming problem, or a blip? Should a sports arena be converted to a field hospital? Is it worth it to buy enough tests for 10,000/day? Is contact tracing worth it? Do masks work? What happens when you reopen the economy? actionable question two types of question formulation: is boston where do we expect covid to be prevelant? ## 0907 - policy resistence - if you push the system to one direction, it moves further away