It's important to understand the limitations. The first being that the scenario falls outside of the range of data used to create the model. This is necessarily

It's important to understand the limitations. The first being that the scenario falls outside of the range of data used to create the model. This is necessarily an extrapolation and like all extrapolations, subject to more critique than an interpolative prediction. Next, from a methods perspective, while a single-value estimate is convenient, a distribution, in this scenario, would be much more useful to understand the range of possible deaths and their associated probabilities. As for the rest of the author's own critiques, I'll lay them out here. One of the most important one is that the population's behavior and government responses are not taken into account. There is, of course, debate as to how much this affects the prediction. > Our estimates do not account for individuals’ behavioral adaptations or the actions of first responders during such a severe heat crisis. In reality, if a blackout event was regionally localized, many individuals would be able to leave their homes and travel to cooler locations. We also do not account for potential governmental responses to a concurrent heat wave and blackout, such as evacuation of the most vulnerable to heat illness or the deployment of mobile power generators for cooling centers. Other important sources of uncertainty in our analysis include (1) the degree to which measures of relative risk based on outdoor temperatures are reflective of indoor heat exposures, (2) uncertainties in the precision of the ambient and indoor temperature models used for the study, (3) uncertainties in behaviors and locations of individuals throughout the day and hence their daily IETs, and (4) uncertainties in the exposure-response functions in the applied epidemiology studies (including the changing nature of population susceptibility with a period of prolonged heat exposure).