Traditional approaches to making and assessing the skill of tropical cyclone forecasts have focused on the track and intensity of the storm. This approach suffers two important drawbacks:  First, the intensity depends on the track -- sometimes very sensitively -- so the two metrics are hardly independent. Second, neither quantity directly addresses the risk that most people are concerned about... the probability of TC-related hazards at their particular location. For this reason, we advocate moving away from the track-intensity axis to making and judging the skill of pointwise probabilistic forecasts of tropical cyclone hazards. To accomplish this it will be necessary to run very large ensembles. I will review a technique developed by Jonathan Lin and myself to bootstrap the GFS and/or ECMWF TC ensemble forecasts to 1,000 members and to use these large ensembles to make pointwise probabilistic wind forecasts. I will also present verification statistics, including statistics on predictions of rapid intensification, and present plans for enhancing the method to include surge and freshwater flooding.