In geohazard practice, we sometimes forget that nearly all inputs to decisions are just informed opinions. We are often reluctant to put numbers to these opinions before collecting and analyzing large amounts of information. And then, when the data are collected and analyses complete, we assign numbers and believe them to be true. In this talk, Michael argues that there is substantial value in putting numbers to informed opinion from the outset of a geohazard assessment, while clearly acknowledging uncertainty and the potential for revision as new information emerges. He begins by contrasting frequentist and degree-of-belief interpretations of probability and argues that much of geohazard analysis is inherently subjective. He then shows how early quantification of probability and consequence enables estimation of the expected value of new information, supports more efficient investigation and monitoring programs, and provides a transparent framework for explaining technical judgment to stakeholders and decision-makers. Building on this foundation, he examines how economic risk analysis can complement safety-based approaches to support clearer, more defensible decisions about risk reduction, and briefly touches on utility theory and the role of stakeholder financial resilience in shaping risk tolerance. Finally, Michael argues that effective geohazard decision-making requires institutional space for learning, allowing opinions, models, and decisions to evolve as new information becomes available. This adaptive approach often sits in tension with existing land-use policies and regulatory frameworks, but is essential for managing low-probability, high-consequence geohazard risk in a changing world.