How data vs perception shapes safety narratives and policies

This is such a powerful example of how perception and data often tell very different stories, and how dominant narratives can shape public opinion (and policy) in misleading ways. It reminds me of research I did for a ride-hailing company, where many passengers assumed drivers were the biggest safety risk. But the data showed the opposite: drivers were actually the ones most frequently attacked, not the passengers. One shocking story can dominate headlines and distort how we see an entire group. That bias doesn’t just influence public opinion. It also shapes how safety features are designed and how policies are enforced. We saw things like surveillance tools focused on drivers instead of protections for them, simply because the assumption was that they were the threat. And it worked. Passengers felt safer. Passengers were the paying users, so the focus stayed there. Ironically, further research showed that one of the top reasons passengers chose our service was the quality and trustworthiness of the drivers. They felt safe because of them. But we never got leadership to invest meaningfully in driver retention or support. The thinking was that drivers were employees and should follow the rules because that's their job. End of story. This is exactly why good design and policy must be informed by data, not assumptions. The gap between perception and reality is not just inconvenient, it can create real harm.

View profile for David Riedman

🤖PhD in AI (measuring variance in LLMs), 🎓Professor, 📊Founder of K-12 School Shooting Database, 🥋BJJ Coach, 🎙️Weekly Podcast Host

From 2012-2014, I mapped the locations of homeless people in Washington, DC. This was an ancillary part of my job as the Homeland Security Advisor to the Downtown DC Business Improvement District. Homelessness was the most common complaint from property managers, and they always said, “the homeless problem is getting worse”. But was it really getting worse? Using iPads (big deal in 2012) and ArcGIS, we walked 144 blocks between the Capitol (east), White House (west), Convention Center (north), and National Mall (south) to find every homeless person on the street. We dropped a pin for the location and took a picture of each one. For almost 2 years, we did a monthly daytime and nighttime count to track the number and locations. Turned out the homeless population was pretty steady from month to month. The hot spots were four parks—Farragut Square, Franklin Park, Lafayette Square, and Pershing Park. These parks were all near the Department of Veterans Affairs headquarters which was a point of gravity for drawing homeless veterans. We estimated that ~20-30% of the downtown homeless population were veterans and their highest concentration was in parks near the VA Headquarters…which is also next to the White House. Anytime the President leaves the White House, he passes one of the parks where homeless veterans gather. Since we had an ArcGIS map with all the homeless locations, we created maps that overlayed homelessness with violent crime, property crime, and vandalism. There wasn’t any correlation. For another project, we created a map of the nighttime streetlight brightness (using ArcGIS and lumen meters) and we found a strong correlation between homelessness and well-lit places. While many homeless would congregate in the parks during the day, they would move at night to a building or alley that had bright outdoor lights. We usually did our nighttime homeless counts between midnight and 3am. I never felt unsafe walking around downtown DC on a random weeknight carrying an iPad (and crime was higher in 2012 than it is right now). What this all means is the DC takeover to address crime and homelessness is completely misguided. We’ve had great data for more than a decade on the root causes and patterns of these issues.

  • No alternative text description for this image

To view or add a comment, sign in

Explore content categories