The many successes in volcano forecasting over the past several decades owe mainly to pattern recognition, both in monitoring data and the geologic record. During the early stages of the 2018 Kīlauea eruption, the conceptual model of Stearns (1925, https://doi.org/10.1007/BF02719505), based on the explosive 1924 Kīlauea eruption, was highly influential. This model postulates that explosions are triggered by liquid water inflow into a recently vacated magma conduit. Modern quantitative modeling approaches, supplemented by hydrogeologic data unavailable in 1925, yield a more nuanced view. Results demonstrate that liquid water inflow would likely be delayed by months to years, owing to the inability of liquid water to transit a zone of very hot rock surrounding the conduit. The exercise demonstrates the use of physically based modeling to supplement traditional volcano forecasting approaches during an ongoing event.