Introduction

Imagine being a transportation planner in the early months of 2020. Overnight, the COVID-19 pandemic upends many of your core assumptions about what drives demand for transportation. Transit ridership plummets, fare revenues collapse, and commuting patterns shift unpredictably. Whether you are working on long-term planning at a metropolitan planning organization (MPO), capital programming at a state department of transportation (DOT), or bus and rail service schedules at a transit agency, you are suddenly confronted with urgent questions with no clear answers: Will remote work become permanent? Will riders return to crowded buses and trains? What if new outbreaks occur? These immediate shocks compound longer term forces that were already reshaping transportation before the pandemic: emerging mobility services like ride-hailing and autonomous vehicles, decentralization of jobs and housing, expanding ecommerce logistics hubs, and increasing climate driven threats such as floods, hurricanes, and wildfires. In such an environment, planners must navigate deep uncertainty while balancing resilience (the ability to absorb and recover from disruptive events) with robustness (the capacity to perform well across a wide range of possible futures).

Uncertainty and change are not new to transportation planning. What is different about modern transportation planning and the growing uncertainty is the pace of change. Technology shifts, demographic changes, volatile funding streams, and disruptive shocks have the potential to reshape where people live, work, and shop; how they travel; and the resulting demand for transportation. Relationships between these factors and travel demand are evolving and so are interactions among them. Agencies have traditionally managed this complexity through land use and travel demand models that forecast a single “most likely” future. For example, the number of vehicles on a corridor by 2035 or the ridership on a rail extension 20 years from now. However, such methods rely on stable assumptions and linear trends, and experience has shown they are fragile in the face of rapid or unforeseen change. The pandemic dramatically illustrated how quickly travel behavior and demand can shift.

While scenario planning has helped transportation agencies move beyond a single “most likely” forecast, it typically explores only a small handful of qualitative stories about the future. This approach gives breadth but does not capture the full range of possible conditions. Exploratory modeling, central to the family of methods known as Decision Making under Deep Uncertainty (DMDU), takes the same idea further. Instead of creating just three or four crafted scenarios, it uses computer models to run hundreds or thousands of variations, each with different assumptions about and interactions between uncertain factors. This broader sweep lets planners “stress test” strategies, see under which conditions they work well or poorly, and identify both robust (work in many futures) and contingent (work in specific futures) options.

Within the DMDU umbrella, methods such as Robust Decision Making (RDM), Dynamic Adaptive Policy Pathways (DAPP), and Exploratory Modeling and Analysis (EMA) build on the scenario planning mindset but add a more systematic, data-driven process: structured sampling of uncertainties, performance metrics, and tools for spotting key patterns. In short, DMDU approaches are an evolution of exploratory scenario planning, keeping its recognition that the future can take many forms but adding the rigor and reach of large-scale quantitative experimentation.

The potential to improve our ability to plan effectively in the face of uncertainty is especially relevant given the scale of transportation investment at stake: U.S. agencies allocate more than $90 billion federally, plus $200 billion at state and local levels, annually to transportation projects.1 Directing DMDU-informed decision-making toward even 5 percent of these investments would influence $15 billion per year, helping agencies avoid stranded assets and build infrastructure that is both resilient and adaptable. The potential benefits are greatest when applied to high-cost, long-life projects and long-range strategic planning.

DMDU has been the subject of some federal and local interest. The Federal Highway Administration’s (FHWA’s) Transportation Modeling Improvement Program (TMIP)2 has promoted uncertainty-aware modeling since 2015. A variety of regional planning agencies have conducted early pilots, including the Sacramento Area Council of Governments (SACOG), TransLink in Metro Vancouver, the Houston-Galveston Area Council, Oregon DOT, and the Boston Region MPO. However, overall adoption remains limited and, in most agencies, DMDU has yet to become a routine part of project selection or strategic planning.

This raises a core question: Why have pilots not scaled into standard practice even as the sources and magnitude of uncertainty increase? This paper explores answers through three perspectives:

• Agency perspectives: How uncertainty is perceived and addressed, who champions DMDU internally, and how leadership, boards, regulators, consultants, and the absence of mandates influence adoption.
• Practical performance: Lessons from agencies that have applied DMDU, including how results compare with conventional methods and what conditions enable benefits.
• Institutional dynamics: Why claimed early successes have not become the norm, what differences across agency types play a role in its adoption and use, and the organizational enablers and constraints shaping mainstreaming.

By examining these questions through agency interviews, case studies, and implementation experiences, this research aims to identify pathways for translating uncertainty-aware planning from a promising exception to routine practice.