December 30, 2021 is a day that will live in infamy for Coloradans. On that day the Marshall Fire ripped through 6,000 acres and close to 1,000 structures with lightning speed. The improbable circumstances of this tragedy offer bitter lessons about risk and resilience.

The proximate cause of the fire increasingly appears to be human stupidity, an infinite resource in a populous society. But the less common variables that converted a misbegotten trash fire into record-setting property damage and suffering are worth examining.

Nassim Nicholas Taleb, the investor and philosopher who famously predicted the crash of 2008, has made an exhaustive study of our inability to envision and plan for highly improbable events. In his book The Black Swan, Taleb argues that both individuals and societies are routinely shaped by “low predictability, large impact” events that share certain dynamics.

For Taleb, a black swan has three attributes. It is a statistical outlier, defying expectations, because nothing in the past points to its possibility. Second, it carries extreme impact. Third, human nature demands that we concoct explanations for it after the fact, inventing causal correlations that were not previously evident.

The Marshall Fire seems to meet these criteria. Several cognitive biases prevented us from anticipating it. First, it started in the winter. Although Colorado is acclimating to an extended fire season, winter fires are rare. Second, a suburban conflagration that behaves like an extreme forest fire is also a novelty. Without extreme wind conditions, this fire would likely have remained a localized phenomenon.

The severe impact of this event is self evident, with damage estimates topping $1 billion and thousands of people struggling to find shelter as they grapple with rebuilding shattered lives. And various media are already proposing correlations that claim to explain the event with the blinding clarity of 20/20 hindsight (some of which have merit while others are raw sensationalism).

Taleb argues that many improbable events can be foreseen, at least in general terms. For those of us who live in the “wildland urban interface” (WUI) rather than the suburbs, that foresight could be a lifesaver.

One of the most stunning aspects of the Marshall Fire was how it jumped across open rangelands to consume building after building, often without the benefit of trees and shrubs to provide ladder fuel.

The intensity and movement of a wildfire ultimately depends on three factors: weather, fuel, and topography. Wind obviously facilitated rapid spreading in the Marshall case, but the uncharacteristically dry weather in November and December also played a significant role.

A fuel’s composition, including moisture level, chemical makeup and even density, influences how quickly a fire will spread and at what temperature. The moisture content of a fuel determines how much of an area will burn. So if vegetation has a low moisture level and is very dry, a fire will burn faster and more intensely because the heat doesn’t have to eliminate water.

High winds explain the rapid spotting behavior of the Marshall Fire, and the lack of ambient moisture also helps explain the extent of the spreading, but why did so many structures burn?

A 2015 study by Michael Gollner and scientists at the Fire Protection Research Foundation offers some insights. It found the single greatest factor that determines structure flammability is the composition of the roof: “Flammable roofing on WUI-exposed structures, especially wooden shingles, has been found to be the most susceptible building component to firebrand attack and ultimately the single most effective predictor of a home burning down.”

Fire ratings for roofs are classified as either Class A, Class B, Class C, or are unrated if a roof covering cannot meet the requirements for any of the three classifications.

Common Class A roof coverings include clay tiles, slate, asphalt glass fiber composition shingles, and concrete tiles. Assembly-rated Class A roof coverings are those that meet Class A standards when combined with other elements. For example, shake roofing with a fire-retardant treatment rates Class B on its own, but achieves a Class A rating when combined with specified underlying materials such as Type 72 roofing felt material.

When Gollner’s team examined data from multiple California fires, it found that all houses with wood shake roofs were destroyed while only 33% of structures with a Class A roof type were destroyed or damaged.

The Front Range communities of Colorado are statistically prone to wildfires, and changes in climate are further increasing fire frequency. We know this. Our communities also experience high wind conditions on a regular basis, and extremely low humidity in the fall and winter.

Other than aggressive mitigation of the fuel load on our respective properties, the biggest variable determining how much wildfire risk we each accept is the roof over our heads. How much risk do you live under?