Words Through a Screen: Ten Years of Rescue, One Unbuilt Bridge

A : T E N O F , O N E T | Artificial Intelligence, Cajun Navy, Community Caretaking, Innovation

Words Through a Screen: What Ten Years of Citizen Rescue Taught Me About the Bridge Emergency Management Refuses to Build By Rob Gaudet, founder of Cajun Navy Ground Force In August 2016, a storm with no name dropped 7.1 trillion gallons of water on south Louisiana — roughly three times the water Katrina left behind. It killed 60 people, damaged 146,000 homes across 56 of our 64 parishes, and generated over eleven thousand requests for rescue. The cell towers in the hardest-hit areas went down. The 911 system buckled. And something happened that had never happened before. Ordinary people — thousands of them — coordinated their own rescue operation through the phones in their pockets. Facebook groups became the intake desk. A walkie-talkie app called Zello became the dispatch radio; our channel peaked at over 800 members. Glympse shared live GPS locations. Volunteer developers built crowdsourced pin maps overnight. Boats launched, found strangers on rooftops, and moved them to dry ground — all without a single agency in the loop. I was in the middle of it, and my background in technology is the reason I could see what it was: not just another volunteer surge, but the birth of a new coordination layer. My method was simple — engage volunteers through social media, and agree to work alongside local authorities. That discipline separated us from frustrated boaters getting turned away at checkpoints. Ten years later, I'm still convinced that what we built in those flooded parishes is the most important undeveloped capability in American disaster response. I'm also convinced almost nobody in emergency management understands it yet. The instinct is ancient. The tool is new. Let me be clear about something first, because I've said it for years: the Cajun Navy did not invent citizen rescue. Nothing about the impulse is new. Go back 10,000 years to any flood where there were canoes or rafts, and people got in them and helped each other. It's innate to humanity — one of the rarest and best things about being human. The research backs this up overwhelmingly. In the 1985 Mexico City earthquake, untrained citizens saved hundreds before professional teams arrived. After the 1995 Kobe earthquake, most people pulled from collapsed homes were saved by neighbors, not firefighters. In one study of 38 landslides around the world, 77% of survivors were found by their neighbors. After the 2010 Haiti earthquake, 71% of surveyed survivors were

pulled from the rubble by a friend, family member, or neighbor — less than 1% by professional rescue forces. Former FEMA Administrator Craig Fugate said it plainly: the real first responders are neighbors, bystanders, and people willing to act. The original Cajun Navy proved it in 2005 — mustered by nothing more than a radio and TV plea to meet at the Acadiana Mall. Nearly 400 boats showed up at 4 a.m. and pulled more than 10,000 people out of the water. That was before the iPhone existed. Broadcast in, convoy out, and every boat on its own once it hit the water. What changed in 2016 was the coordination. For the first time, the rescuers themselves — not remote volunteers, not agencies — had real-time voice, live location, and a shared map in their hands while they worked. The ancient instinct got a nervous system. A year later, Hurricane Harvey scaled it to Houston: Zello usage jumped twenty-fold, Crowdsource Rescue registered 60,000 volunteers, and civilian rescues ran into the tens of thousands alongside the Guard and Coast Guard totals. That's a force multiplier. And it's been sitting in plain sight for a decade. Words through a screen do not discriminate Here's the part almost nobody wants to talk about, and it's the part I think matters most. Words through a screen do not discriminate who is capable and who is not. On a feed, a boat captain with thirty years on the water and a panicking poster who has never touched a throttle look identical. The screen makes all voices equal when they are absolutely not equal. I learned this in the water. The research learned it in the data. University of Washington professor Kate Starbird went into crisis research believing in the "self-correcting crowd" — the idea that online communities fix their own errors the way Wikipedia does. Her own data destroyed that belief. Rumors spread much further than corrections ever did, and the correction, when it came, reached far fewer people than the falsehood. The operational costs are real. During Harvey, roughly half of social-media rescue requests lacked the information needed to act on them, and a large share came from areas that weren't even flooding. The Houston Police Department was begging people to stop sending rescue requests through social media. Self-dispatched volunteers become victims themselves — over a hundred rescuers died in Mexico City in 1985. And by Hurricane Helene in 2024, misinformation had grown from a nuisance into a primary operational threat: FEMA's administrator called it the worst she had ever seen, and the agency had to stand up a dedicated rumor-response page to fight claims that it was seizing land and blocking aid. The technology that lets a certain strata of citizen rise up and coordinate is the same technology that amplifies the loudest incompetent voice in the room. Both things are true.

Anyone who tells you only one of them is true is selling something. The answer isn't a better crowd. It's a trusted network. If the screen erases competence signals, then the fix isn't more openness — it's trust established before the water rises. This is the thesis I built stability.org around: smart force multipliers through trusted networks that already exist in communities. Churches. Rotary clubs. Neighborhood coalitions. Volunteer groups with track records. The people who already know who can drive a boat, who owns a chainsaw, who can cook for two hundred. The network is the verification layer. The relationships are the credentials. And here's what surprised me when I finally dug into the academic literature: the field's own founders agree with me more than the field does. Russell Dynes — one of the giants of disaster sociology — argued back in 1994 that emergency planning built on military command-and-control analogies is fundamentally broken, because it treats disasters as chaos to be commanded rather than communities to be coordinated. He warned that legislative and technological "improvements" often make planning more rigid and more inadequate. His alternative model? Continuity, coordination, and cooperation — building on what already exists in communities. I've said the very idea of a plan is a barrier to effective response. Dynes taught me to sharpen that: the command-and-control plan is the barrier. A plan that assumes the public is helpless, that treats every spontaneous volunteer as a liability to be contained at a reception center, that routes everything through a hierarchy designed for enemy attack scenarios — that plan will lose to a Facebook group every single time. It has, in every flood since 2016. Meanwhile, the field's own literature calls trusted digital volunteers "force multipliers" — my exact words — in describing Virtual Operations Support Teams. The concept has existed since 2011. The adoption is patchy at best. Emergency management is a young field, and after ten years of doing this work I'll say it straight: it's also a stagnant one, bound by liability fears, procurement systems built for hardware vendors, and a doctrine that has no category for the neighbor with a boat other than "problem." And now the federal government is stepping back The timing makes all of this urgent. The FEMA Review Council's 2026 report keeps the agency but shrinks it, raises disaster-declaration thresholds by more than half, and shifts responsibility to states and communities under the banner of "locally executed, state

managed, federally supported." Nearly a third of the disasters declared over the past decade wouldn't qualify under the new threshold. You can argue about whether that's the right policy. What you can't argue about is the math: more disasters are about to land entirely on communities. The federal backstop is receding exactly as the citizen coordination layer has proven itself — and exactly as it remains unintegrated, uncredentialed, and unfunded. That gap is the biggest story in disaster relief, and almost no one is covering it. The honest ending I'll be honest about the hard parts, because a thesis you can't stress-test isn't a thesis. The model dies when the network dies — Puerto Rico after Maria and Lahaina after the fires proved that a coordination layer built on cell towers collapses with them, which is why satellite and mesh comms have to be first-class requirements, not afterthoughts. And platforms like stability.org face a brutal problem I know intimately: communities won't invest in a tool they need once every few years, and emergency managers won't trust data from a network they didn't credential. I stepped back from stability.org for exactly these reasons. The work required was enormous, and the disaster community wasn't ready to wrap its head around it. But the instinct that put 400 boats in a mall parking lot in 2005, and 800 voices on a Zello channel in 2016, isn't going anywhere. It's older than any agency and more reliable than any plan. The only question — the question the next ten years will answer — is whether emergency management finally builds the bridge to it, or whether communities, abandoned by a shrinking federal system, simply build it themselves. My money, as always, is on the neighbors. Rob Gaudet founded Cajun Navy Ground Force after the 2016 Louisiana floods, where citizen volunteers first used social and mobile technology to coordinate rescue at scale. He is a 13th-generation descendant of Jean Gaudet, one of the Acadian patriarchs counted in the 1671 census of Acadie.