I have a confession to make #1…in a series. From time to time, while writing The Leech and researching speech topics on various contemporary issues, I am presented the opportunity to challenge my presumptions and replace them with a more rigorous understanding. Although these essays may not be life altering or change the world, I share them for two possible benefits. First, to convey what I learned on a topic, how it deflated my original hypothesis, and replaced it with a better one. Second, to illustrate a hidden benefit and reality of the painful writing process: you often land in a very different place than where you targeted. That’s growth.
As an engineer, I have a life-long passion for STEM. I also have an inherent skepticism whenever non-STEM entities meddle in scientific matters. That often is a situation where science is demoted, contorted, or appropriated to fit a convenient narrative the journalist, artist, or politician is wishing to propagate. In 2020 we saw this troubling phenomenon of non-scientists playing fast and loose with science in heavy matters ranging from the pandemic to election polling.
When this occurs, the sanctity of scientific integrity is jettisoned, as long as the storyline is bolstered. This abuse of science cuts across party lines and political affiliations. As it becomes more common, tolerance for the ill-advised practice grows. Today, I fear much of our society has become desensitized to journalists and politicians treating the scientific method as nothing more than a convenient, disposable means to an end.
Weather is not immune to this crisis. Contemporary journalism has no problem manipulating science and data to create a desired premise and then broadcasting that premise across the wires. An important example is in the morbid, but potentially life-saving, pursuit of accurate hurricane fatality modeling. Hurricane Maria, which slammed into Puerto Rico in September 2017, serves as a case in point.
Instead of focusing on repairing vital infrastructure and delivering needed medical care, time and resources were being expended on both sides of the political spectrum on debating which model and fatality estimate were accurate. President Trump fired off tweets. Reporters upped the rhetoric. Both sides of non-scientists labeled the other as an attacker of science, with one side looking to push down the fatality count and the other looking to boost it. Both sides became distracted from the task at hand: getting the island functioning again and delivering aid to Puerto Ricans in need.
When Hurricane Maria hit Puerto Rico, the island was heavily damaged, and the electric grid was wiped out. The original death toll from the category-4 storm was reported by the Puerto Rican government to be 64 people using traditional historic measurement methods. Recognizing that even a single fatality is a human tragedy, the initial estimate of 64 deaths from Maria did not register the hurricane as being one of the top-ten deadliest U.S. hurricanes. The extent of the storm’s physical destruction quickly shifted attention from where the storm ranked all-time to the national response on how to restore power and repair the island’s infrastructure.
But that’s also when the commandeering of science began. The media and certain politicians desired a higher fatality estimate. A higher number would help bolster the case for catastrophic climate change and a host of other issues. So, the media, working closely with allies, modified the methodology used to estimate the storm’s fatalities. Under the new approach, deaths that occurred after the storm passed and that could be indirectly attributed to lost power, inability to get medical care, or other infrastructure failings were added to the total. The new estimate rose to somewhere around 3,000 people. That placed Maria as the third deadliest hurricane in U.S. history.
How did the death toll estimate increase by 4,500%? By using new, black-box statistical models and subjective judgments based off imperfect data. Before you knew it, other models and estimates were being constructed and published, each making its case as the best and most accurate. Certain estimates employed statistical models that included island deaths that occurred months after the storm. Other estimates were based on polls of funeral homes on the island. The range of these estimates was quite wide; all were significantly higher than initial estimate of 64 deaths.
After the post-Maria debacle, I wanted to know: What gives when it comes to estimating a storm’s deadliness? How’s it properly done under a rigorous scientific method?
I had the pleasure of spending a few weeks talking storm fatality modeling with an expert in the field. The individual, who is on the faculty at a renowned academic institution and wishes to remain anonymous, provided invaluable insight into this technical profession and craft.
Scientists and engineers in this arena are steeped in statistics, computer modeling, and data analysis. Expertise in meteorology or the health sciences is also common. The complex computer models built to estimate storm fatalities are constantly refined and improved. The history and data set of storms continue to grow, making the models more accurate as time goes on. The endeavor is a never-ending quest to build the better mousetrap to accurately estimate the impact of storms on human life.
My contact/expert had assessed Hurricane Maria and continued to recalibrate the assessment as refinements and improvements were adopted to the model and to the data inputs over the past few years. The expectation was to continue the effort for years to come.
So, where did this model land with its measure of Maria’s fatality number? Well, I am not at liberty to say. But suffice to say it was much higher than 64 and much lower than 3,000. Maria, according to this model and engineer scientist, ranked as one of the ten deadliest hurricanes in U.S. history.
But the estimated number is only a superficial output of the process. More importantly, when one screened the model results to search for correlations, some obvious and crucial conclusions quickly manifested. During the storm, fatalities don’t correlate to age or gender and highly correlate to location: hurricane force winds and flooding rain will indiscriminately kill different demographics, but location makes a huge difference. When the storm is raging, it doesn’t matter much who you are, but it matters terribly where you are. Yet in the days and weeks after the storm, a strong correlation links fatality rate to the youngest and oldest of the island, primarily due to lack of health care. Conclusions like these are crucial to drive informed policy and future investments for the island.
The vital need for access to reliable electricity also stood out in the modeling results: if the island’s grid was buttressed and dependable, storm fatalities would be drastically reduced, particularly for deaths occurring in the weeks immediately after storms. Such an improvement in Puerto Rico’s grid would also bring the long-term benefits of alleviating poverty and extending life expectancy, regardless of hurricane frequency. Truly life-altering and impactful.
The scientists and engineers working in this noble field deserve society’s support. Future lives are at stake. Society should demand the rest of us, including media and politicians, leave our technical wizards unencumbered and follow their findings with policy and action that aids the human condition.
My friend took my original hypothesis, deflated it, and built a superior view. I went in believing one estimate over another, based on where each estimate emanated from. I exited understanding a truth in between. More specifically, a scientific truth not tainted with political science but forged in logic, reason, and math. That kind of truth can save lives.
I thanked my friend and asked him what I could do in return. The answer: leave the science and engineering to us; please leave us alone. Perfect.