On March 30th, 2023, around 1 AM, a train carrying ethanol and corn syrup derailed outside the small town of Raymond, Minnesota. The mix proved to be just as flammable as it sounds. The town, passing motorists, and local environment were all endangered, out in the middle of nowhere in the middle of the night.
But the local firefighters, who arrived on the scene first, had trained for this. They knew who to contact, and how, and what plans were in place. A rapid, coordinated response among fifteen agencies and volunteer groups blocked off roads, isolated the fire, and fully evacuated the town by the simple mechanism of having someone visit every single house, wake the residents up, and make sure they knew what to do before heading to the next one.
Ardell Tensen, the mayor of Raymond and assistant fire chief, told an interviewer that his team had spent a day training for these kinds of disasters just two weeks ago, and were frequently updated as the process was tweaked. In particular, he’d learned how to quickly determine what potentially hazardous materials were in each car, using the train’s logbook, and what the appropriate responses were. They dampened the fire, then monitored it while it burned out. The evacuees were cleared to return by 11 AM that day. Nobody was hurt, and there was no noticeable environmental impact. The overall cost of the incident was about two million dollars.
No big deal, in other words. Not really national news.
When a similar train crashed in Ohio, though, two months earlier? That was national news.
The first responders, also firefighters, in East Palestine, Ohio, had a harder task than the ones in Minnesota. The chemicals spilled here were toxic, and less familiar. And they hadn’t been given the same level of support and planning. They hadn’t really been trained for this situation, according to a National Transport Safety Board investigation. They didn’t have a protocol for coordinating with other response teams—everybody was literally on a different radio frequency. They lost time trying to read the labels on the train car placards to find out what was inside, labels that were often scorched into illegibility.
They were not successful in containing the leak. All nearby water was poisoned, killing the fish. The air was likely poisoned, too. Nearby small animals, including pets, died over the next few days. Most first responders were unaware of the dangers. A police officer told The Guardian: “We were never told about the cargo on the train and we were never told to wear protective clothing, although it did not matter because our personal protective equipment dates back to 2010.”
The cleanup was soon taken over by the train company, Norfolk Southern, who had an incentive to move quickly as opposed to safely. Working off an outdated pamphlet with incorrect information, they mistakenly decided that they needed to deliberately breach and burn several of the cars, to prevent an explosion that was not actually coming. The burn released toxins into the atmosphere, toxins that are currently detectable in sixteen U.S. states. If you live within 530 miles of the crash, it is, as we speak, giving you cancer. (I’m exaggerating. I hope.)
Ohio, at first, waved off offers of federal assistance. On February 14th, eleven days later, Governor DeWine told reporters “Look, the president called me and said, ‘Anything you need.’ I have not called him back after that conversation. We will not hesitate to do that if we’re seeing a problem or anything, but I’m not seeing it.”
He probably should’ve called President Biden back. The cost of the disaster is over one billion dollars, and still rising.
Cool Guys Don’t Look At Explosions
Mitigated disasters, like the one in Minnesota, don’t make for good photo ops. Catastrophes do. Future running mates Donald Trump, whose administration had repealed first responder training mandates, and JD Vance, Ohio’s junior senator, visited the area on February 22nd. They made speeches blaming Biden, and handed out bottled water to the residents. Vance posted a video of himself poking a polluted creek with a long stick. Both considered it a political win.
Vance’s opponent in the 2024 election, Minnesota Governor Tim Walz, was photographed at the site of the Raymond crash, in his case the same day it happened.
It’s a terrible photo. He’s facing towards the disaster, so you can’t see his face, against Hollywood protocol. There aren’t any victims to be handing out water to, because they’re mostly in a church one town over, playing with puppies and having their basic needs met by a local CERT team. The hand gestures imply that he’s not even the one speaking—he’s listening. What a nerd.
This is typical of Walz. In his decade in the House of Representatives, his committee activity was focused on two things: veterans’ issues and risk management. Whatever the committee he was on, his subcommittee assignments were in the “but what if it goes wrong?” genre. And he was very engaged, as we can see from the transcripts of hearings held by the Subcommittee on Railroads, Pipelines, and Hazardous Material. Not making fiery speeches, not assigning blame. Just, affably, asking questions and making suggestions.
In one hearing, he discusses how best to ensure that firefighters are prepared to respond to spills in populated areas.
He gets into the weeds, suggesting more cost-effective solutions, talking to local officials and asking whether proposed interventions, such as mandatory spill response plans, would actually help. He asks witnesses for numbers on how often their inspections actually identify issues, and how quickly those issues are fixed. He also pays attention to culture—is there an expectation that we have a plan A, B, and C at every level? Do low-level people feel comfortable and empowered to raise safety concerns as they spot them?
It’s not very telegenic. Not like Vance’s speeches in committee, such as the one after the crash in his home state where he started off by suggesting that the root cause of the botched response was that the locals were being discriminated against for being White. (For the record, Minnesota’s demographics are whiter than Ohio’s.)
The difficulty of bragging about something that didn’t happen
Before serving in the House, Walz spent 24 years in the Army National Guard, starting when he was 17. For most of that time, he was assigned to disaster response postings. In 2001, he had earned the right to retire, and did…but only for a few months. He reenlisted after September 11th, and commanded a unit deployed to improve security at the Caserma Ederle military base in Italy.
U.S. military bases in Italy were, at the time, a popular target for domestic terrorist groups, most notably the Nuclei Territoriali Anti-Imperialisti. During Italy’s involvement in the 1999 Kosovo bombings, protesters very briefly breached the perimeter of one base and started a small fire. So this wasn’t an insignificant task. It was executed successfully, though—the NTA tried, without success, to infiltrate the bases, and eventually were captured in one such attempt. The best they were able to accomplish was to set fire to empty cars belonging to people deployed at the bases, parked outside the secure perimeter.
Walz, wisely, doesn’t crow about this (that I’ve seen). There were multiple redundant lines of defense. He can’t prove that anything he personally did mattered. He wasn’t in significant personal danger, unless you count the risk to his car.
In fact, predictably, he’s being attacked for his service. In twenty-four years, he only ever did the boring stuff, which smacks of cowardice and stolen valor. Any time he talks about his military service without mentioning that he never saw combat, his opponents say that he’s misleading the public about his record. And it’s tricky to push back. He didn’t kill anybody, or get shot. There’s only indirect, circumstantial evidence that because of him, people didn’t die.
That’s also been a theme of his current role, as Minnesota’s governor. Between 2018 and 2022, Minnesota’s age-adjusted all-cause mortality rate was the fifth-lowest in the country, 13% below the national average. So that’s, very roughly, 6,500 lives saved each year by Minnesota’s above-average competence. Or, if you compared it to Ohio’s mortality rate, one of the highest in the country, it’s more like 14,700 lives saved.
But here, too, we don’t know whose lives were saved. Walz’s COVID response, for example, is probably why Minnesota’s death rate from COVID was about half that of Ohio’s. But we don’t know who specifically would’ve caught it if Walz hadn’t been aggressive in his stay-at-home orders and school closures, and we can’t know for sure how much of a role his efforts actually played.
I’m inclined, though, to give him the benefit of the doubt on this one. Even if we decided to credit all of Minnesota's successes to their culture and local initiatives, rather than state government, Walz has spent decades trying to promote and support a culture and infrastructure of resiliency, so he deserves a share of the praise. Like pollution, his effects are too diffuse to point to, but hard to deny in aggregate.
Incidents and Accidents
Was Minnesota just luckier than Ohio in the severity of the spill? After all, corn syrup is a food and technically ethanol kind of is too, and they burn fairly cleanly—hardly the worst case scenario for hazardous materials. And their spill happened soon after the highly-publicized Ohio disaster, which probably helped.
But if you google things like “toxic spills in minnesota” you can get some quick evidence that Minnesota’s not lucky; they’re good. The year before either spill, Minnesota’s legally mandated 2022 annual report on readiness to respond to hazardous spills noted, among other things, 34 spills that year resulting in “the release or threatened release of anhydrous ammonia.” None of which had dramatic consequences. In 2023, 150 gallons of hydrochloric acid were spilled in downtown Minneapolis, which is closer to the worst case scenario. First responders showed up with the necessary protective gear, neutralizing the acid within an hour of the incident. Nobody was burned, or was at serious risk of long-term effects.
Walz in committee, like OSHA in its blog, talked about preferring the word “incident” to “accident.” “Accident” nudges you toward thinking of what happened as unavoidable, focusing entirely on response. Accidents feel like something you can’t possibly plan for. You can plan for incidents, and analyze why they happened.
It’s part of a mindset that treats all problems as institutional problems, all failures as institutional failures. Sure, you can sometimes find a single person who messed up, or was malicious. But that’s expecting too little of civilization. We should always be trying to build systems that avoid single points of failure, including human ones.
If you’re so smart, why are you rich?
Here’s one more possible explanation for Minnesota’s competence and safety: the median income in Minnesota was $77,720 in 2021, while the median income in Ohio was $62,262. So maybe the difference is more that Minnesota has more money to spend on hazmat suits.
Of course, we could then have the same kind of argument about the causes of that. But I don’t think we need to, here, because Ohio’s state government actually spends significantly more than Minnesota’s, both in absolute terms and per capita. Its budget is the third-highest in the country, while Minnesota’s is right in the middle.
So Ohio has the money to be Minnesota, but its leaders aren’t spending it well.
(Ed: On further investigation, I don’t think the figures I was using were annualized consistently, and therefore they weren’t directly comparable. I’ve updated the linked Wikipedia table. See footnote.1)
How can we better incentivize our leaders to be boring?
We don’t reward our leaders for preventing, or mitigating, disasters. You get famous for being photographed digging people out of the rubble, not for hiring the person who trains the person who stops the explosion. The difficult and complex problems in government require competence that’s difficult and complex to judge. Whereas it’s easy to see that JD Vance looks good in a white button-down shirt. (It appears to be both his “poking toxic waste with a stick” shirt and his “Senate committee hearing” shirt. I hope he at least washes it.) And he sure seems angry at the spill, so that’s good, right?
How can we make disaster-prevention work more politically viable? There are some tempting Hansonian possibilities—subsidize a market that has an incentive to accurately forecast long-term outcomes, and then judge governors on which direction their stock moves in. But as the 2008 collapse famously showed, our financial system doesn’t do a great job of forecasting risk, even when billions are at stake.
The best strategy I can think of is to write articles like this one, and do TV news spots like it, and so on, except not just when someone like Walz is running in a major election. Disasters are being averted all the time, and we should report on it all the time. If I were running a national news outlet, one simple idea I’d consider is picking a random local news story to highlight every day. Terrible things happening will bubble up to national news, but success stories, even decidedly nuanced ones like “we spilled acid but then cleaned it up right away,” should have a chance of bubbling up too, for balance. If we celebrate those little successes, we’ll end up putting boring competence in the spotlight as a side effect.
In fact, let me put my money where my mouth is. Searching for “spill” in recent Google News articles just now, the first clearly good response I see to a clearly dangerous situation is an incident in Vicksburg, Mississippi. An 18-wheeler parked on hot asphalt flipped over, near the Mississippi River, and started leaking chlorobenzene, one of the contaminants from the East Palestine spill. Local fire chief Derrick Stamps called in a containment team from the Mississippi Department of Environmental Quality, which worked overnight to successfully contain the spill. Chris Wells, the director of that department, is an environmental lawyer who is also certified as a professional engineer. There were a lot of ways, as we’ve seen, that this could’ve gone badly. But thanks to the swift and coordinated efforts of multiple agencies, nobody was hurt and there was, we think, no environmental impact or other serious damage.
So here’s a picture each of a couple of the heroes who did the boring work that possibly, ambiguously, prevented the poisoning of the Mississippi. Smiling into the camera, while behind them, nothing is in any way on fire.
After my corrections, Ohio’s budget drops to the 5th-highest, and its per-capita spending falls below Minnesota’s, even when taking into account that Ohio receives much more federal aid. I’m not sure whether it’s best to compare absolute or per capita numbers—presumably, disaster prevention gets more expensive with more people, but sub-linearly, e.g. you don’t need twice as many reusable hazmat suits when you have twice as many people, just 40% more, because the odds of needing to dispatch two separate teams at the same time only increases quadratically.
"If I were running a national news outlet, one simple idea I’d consider is picking a random local news story to highlight every day. Terrible things happening will bubble up to national news, but success stories, even decidedly nuanced ones like “we spilled acid but then cleaned it up right away,” should have a chance of bubbling up too, for balance. If we celebrate those little successes, we’ll end up putting boring competence in the spotlight as a side effect." PLEASE DO
Preventing stuff by actively looking for potential sources of instability in systems is so much easier and convenient than dealing with the consequences of negligence it's ridiculous. As to the problem of rewarding people for stuff that didn't happen, aside from wonderful articles like this one, I can only think of having computers (or Bay Area Bayesians) trying to constantly calculate publicly available probabilities of incidents such as forests fires or traffic crashes on certain roads so politicians have a number they can brag about having diminished. Goodhart's law considerations apply though.