More than four years since the start of the Flint water crisis, thousands of residents still have lead pipes. Flint is in the process of finding and replacing those pipes, but it is challenging. City records are spotty, making locating at-risk homes somewhat of a shot in the dark. However, simply digging up all the lines in the city would be costly and time-consuming.
This is where the computer scientists come in — specifically Eric Schwartz of the University of Michigan and Jacob Abernethy of Georgia Tech (formerly of the University). The two assistant professors designed a machine learning model to predict where lead pipes are most likely to be in Flint using digitized city records as well as property data such as age and location. The Flint pipe replacement team was able to achieve above an 80-percent success rate for finding lead pipes in 2017 with this model.
The problem is Flint stopped using it. Due to misconceptions about predictive models, the strategy of prioritizing at-risk homes was abandoned. Residents questioned why their neighbor’s pipes were being excavated while theirs received no attention. City council members asked the same thing about their wards. Facing mounting political pressure, Flint adopted the approach of excavating the pipes of every active water account (without prioritization).
The misconceptions don’t seem to be limited to those without a technical background. The international engineering firm hired to execute the pipe-removal project, AECOM, explained they aren’t using the model to prioritize homes because it is “94 percent accurate” — an especially puzzling statement considering the alternative is to dig and hope.
This change to a non-prioritized approach has resulted in an accuracy rate of 15 percent for finding lead pipes. That is a pretty damning statistic when you consider that, while using the predictive model, the replacement team was achieving an accuracy rate above 80 percent. I agree that every Flint resident deserves the peace of mind of knowing that their home doesn’t contain lead pipes. The situation in Flint has been far too dire for anything else. However, the city failing to prioritize those most at-risk is not acceptable — especially when they have a proven option which can help them do just that.
Poor decisions like the one described above will become commonplace if tech-illiteracy among decision makers remains the status quo. Members of Congress have demonstrated their incompetence on tech issues repeatedly this year, with hearings intended to hold tech companies accountable devolving into basic explanations of how Facebook and Google make money and other questions that could be answered by using your favorite search engine (Google perhaps?).
In order to best serve their constituents, policymakers should be prioritizing getting informed about the basics of technology and the tech industry at large. Facebook, Google, Amazon and the like are no longer scrappy upstarts carving a place for themselves in the economy. They have their hands in a wide range of sectors such as entertainment, health care and defense and are the most valuable companies in the world. Until they are treated as the giants of industry that they are, we risk allowing unchecked externalities to run their course.
Furthermore, aside from the industry giants’ reach, technology is everywhere. Our digital age ensures that whatever issue a policymaker cares about most, whether it’s the financial industry or the justice system, has pressing questions involving the use of technology. Understanding the basics will help all legislators better tackle the problems they care about most.
I don’t expect politicians to become subject matter experts on technology. That’s not productive or feasible. However, a baseline understanding of the most prevalent use cases is a worthwhile goal. Such an understanding would help policymakers more effectively address constituent concerns and confusion, like what city officials faced in Flint. Technology that truly improves the lives of citizens could be identified and prioritized. Decision makers could ask the right questions about companies and their technology. Our representatives are expected to understand the basics of health care because it affects every citizen. Shouldn’t they understand the basics of technology for the very same reason?
The narrative that tech companies are motivated by altruistic sensibilities has been proven to be clearly and demonstrably false by now. They are driven by profit just like any other company. The tech industry’s promise of self-governance and a wave of public benefits has proven to be unrealistic, and it is now the responsibility of our legislators to represent the best interests of citizens when it comes to technology.
Unfortunately for Flint residents, their leaders’ lack of tech knowledge has led to yet another misstep in the city’s long running saga. We can only hope other policymakers take note before they make similar mistakes.