Arnold Barnett: How an “Obnoxious” Quote Made Me an Aviation Safety Expert

---

Listen:

Check out all episodes on the My Favorite Mistake main page.

---

My guest for Episode #169 of the My Favorite Mistake podcast is Arnold (Arnie) Barnett. He is the George Eastman Professor of Management Science and a Professor of Statistics at the MIT Sloan School of Management.

Barnett holds a BA in mathematics from Columbia College and a PhD in mathematics from MIT. His research specialty is applied mathematical modeling with a focus on problems of health and safety.

Cited as “the nation’s leading expert on aviation safety,” Barnett was recognized with the 2002 President’s Citation from the Flight Safety Foundation for “truly outstanding contributions on behalf of safety.” MIT Sloan students have honored him on 14 occasions for outstanding teaching.

In this episode, Arnie shares his “favorite mistake” story about blurting out something to a New York Times reporter who called to get his comment about a US Airways crash that had occurred earlier in the day. Even though he regrets saying what he did, it gave him a reputation for being “willing to talk straight” which led a torrent of requests to speak and to be interviewed in venues ranging from radio programs to NBC's Today Show.

We also talk about questions and topics including:

• 1994 US Air had a number of crashes – a “temporary spasm of bad luck”?
• The NY Times article he was quoted in — the “mistake”
• The word “amazing”: “causing great surprise or wonder; astonishing”
• Lesson about talking with the press?
• Are we bad at estimating probabilities in general?
• Bad at estimating the risk of driving vs. flying?
• You wrote an opinion piece in late March 2022 titled “Don’t end the mask mandate for US airlines”
• “ending the requirement now would be a serious mistake.” —> why did you say that then and do you still say that now?
• I saw you give a talk about this — is the Electoral College a mistake? Is it a mistake that can be fixed?
• A simple fix for gerrymandering?
• “MIT now has a reputation of being very much woke”
• Tell us about the Leaders for Global Operations program… you are a popular internship and thesis advisor. Why do you like working with LGO students?

Scroll down to find:

• Video of the episode
• Quotes
• How to subscribe
• Full transcript

---

Watch the Episode:

---

Quotes:

---

Subscribe, Follow, Support, Rate, and Review!

Please follow, rate, and review via Apple Podcasts or Podchaser or your favorite app — that helps others find this content and you'll be sure to get future episodes as they are released weekly. You can also become a financial supporter of the show through Anchor.fm.

You can now sign up to get new episodes via email, to make sure you don't miss an episode.

This podcast is part of the Lean Communicators network.

---

Automated Transcript (Likely Contains Mistakes):

Mark Graban:
I’m Mark Graban, and this is My Favorite Mistake. In this podcast, you’ll hear business leaders and other interesting people talking about their favorite mistakes—because we all make mistakes, but what matters is learning from them instead of repeating them over and over. This is a place for honest reflection and conversation, personal growth, and professional success. Visit MyFavoriteMistakePodcast.com to learn more.

To learn more about Professor Barnett and get links to his articles and research, look for links in the show notes or go to MarkGraban.com/mistake169.

As always, thanks for listening. Hi, everybody—welcome to My Favorite Mistake. My favorite professor from my days at MIT is Arnold Barnett. And before I tell you more about him, I imagine “Arnold” is what you’re called when you’re in trouble, because most everyone calls you “Arnie,” right?

Arnold Barnett:
Well, either is fine. My mother always called me Arnold.

Mark Graban:
So, Arnie, thank you for being here. Arnie is the George Eastman Professor of Management Science and a Professor of Statistics at the MIT Sloan School of Management. He has a BA in Mathematics from Columbia and a PhD in Mathematics from MIT. His research specialty is applied mathematical modeling with a focus on problems of health and safety.

He’s been cited as the nation’s leading expert on aviation safety and was recognized with the 2002 President’s Citation from the Flight Safety Foundation for truly outstanding contributions on behalf of safety. MIT Sloan students have honored him on 14 occasions for outstanding teaching.

That’s not just me—my wife, who was also a student, is in that camp. We know other students who agree; we’re not a statistical anomaly.

One other memory from the summer of 1997: in our first summer of the Leaders for Manufacturing Program class, you told us our next class was canceled because you’d been invited to a meeting with the Vice President. We all thought that was a perfectly fine reason to be away.

Arnold Barnett:
I probably was lying, but it must have sounded like a good excuse.

Mark Graban:
Well, the lie was very specific, which makes it believable. This was back when they were looking at the question of bag-match for passengers and checked baggage.

Arnold Barnett:
Yes, that’s true. But I probably postponed the class—I don’t think I would have canceled it—because you all pay sizable tuition to be at MIT, and I don’t think professors have the right to announce they’re too busy for you.

Mark Graban:
Well, it was for the greater good, and education is a weird product where sometimes people are happier to get less of it for the same money.

Arnold Barnett:
I hope my class wasn’t an example of that.

Mark Graban:
No, we got a lot of value and lessons from your teaching. I have questions to ask later about statistics and mistakes people, businesses, or governments make—but we always start here: Arnie, thinking back to your career, what would you say is your favorite mistake?

Arnold Barnett:
It’s a little obnoxious to describe it this way, but I’m trying to answer honestly: it is my favorite mistake. I’ve written papers on aviation safety using statistical principles to make sense of safety records. One point I’ve stressed is that bad crashes of airplanes, particularly in the U.S., are extremely rare events. When events are rare, observed statistics can be misleading.

Let me give an example. Suppose I have a fair coin and toss it twice. There are four equally likely outcomes: HH, HT, TH, TT. So, there’s a 50% chance of getting two heads or two tails. If someone sees two heads in a row and says, “The coin can’t be fair,” that’s misleading—because it’s not rare even if the coin is fair.

That’s what I stressed when people said, “This airline had two crashes, this one had none—so fly the one without crashes.” Even if they’re equally safe, one airline having two crashes is like getting two heads—it happens.

Back in 1994, US Air had a number of crashes in quick succession. I’d been saying this could just be bad luck—like multiple heads in a row. But in September 1994, there was a US Air crash in Pittsburgh that killed about 150 people. I didn’t know about it until I got a call from a New York Times reporter I’d spoken with before.

He said, “There’s just been a bad crash on US Air.” I blurted out, “US Air again? That’s amazing!” Of course, I tried to walk it back, saying, “With rare events, things can happen,” but he had his quote. On page one of the Times, it said, “US Air again? That’s amazing.”

Even though I didn’t mean to say it that way, I started getting comments like, “Finally, someone willing to talk straight.” People are used to academics saying, “On the one hand… on the other hand…” But here I said something direct, and it struck them as a breath of fresh air. I got a torrent of requests to speak—from newspapers, radio, even NBC’s Today Show.

That attention helped solidify my reputation as a go-to person on aviation safety statistics. I was invited to projects on mid-air collisions, runway collisions, aviation security, and experiments like bag-match for passengers.

So, my career benefited from that mistake—but it was a mistake. If it contributed to the perception US Air was less safe, maybe it did harm. Ironically, in the next 20 years, US Airways had a perfect safety record and three instances of heroic piloting, including the Miracle on the Hudson.

I later wrote that when US Air was absorbed by American Airlines, it “left in triumph.” But yes—had I heard about the crash earlier and prepared, I wouldn’t have said that. Still, in this strange world, the comment helped my career.

Mark Graban:
It’s an interesting reflection on how a moment can open unexpected doors. I’d argue there was societal benefit to your being asked for statistical analysis—it wasn’t just about publicity.

Arnold Barnett:
Well, you know, they say—Michael Kinsley once said—a gaffe is when a politician accidentally tells the truth. I think there was a sense I was telling the truth. If you have five crashes in five years, the natural reaction is, “What’s going on here?” Saying, “Maybe it’s just bad luck” might be true, but maybe there’s a real problem. In that sense, my reaction was like many others’. Perhaps it was good to legitimize asking the question—because if experts always tell us things we sense aren’t true, we pay less attention to them.

Mark Graban:
Did you ever hear—directly or indirectly—from anyone at US Airways who was mad at you?

Arnold Barnett:
Yes. In particular, a pilot from a Southern state wrote saying, among other things, “You have dishonored me in front of my family.” I was glad I wasn’t nearby and glad Massachusetts has gun control. His wife, who I believe was an alcoholic, started calling me—and even called the Dean of the Sloan School, saying, “Your faculty members are making idiots of themselves.”

To be fair, that was partly based on something I said to Katie Couric, but that interview wouldn’t have happened without the “that’s amazing” comment. I was concerned, because when someone says, “You’ve dishonored me in front of my family,” you think the next thing might be a duel.

Mark Graban:
Or at least being slapped across the face with a glove.

Arnold Barnett:
Yes. But I did write to the family and said, “I know you’re unhappy.” I pointed out that later that day, talking to Newsweek, I’d said it was likely just a spasm of bad luck. I told them no one was rooting for US Air more than I was—though, of course, that was also true for the people at US Airways.

Ironically, the crash wasn’t US Air’s fault—it was due to a rudder problem on the Boeing 737. Any airline flying that plane could have been the first to lose one because of it. Southwest actually had more 737s than US Air, so statistically, it could have happened to them. United had a previous 737 crash due to the same rudder problem, but it hadn’t been recognized yet.

US Airways employees helped find the cause. So it really was an airline that had suffered bad luck and then had a perfect record thereafter.

Mark Graban:
There’s also a lesson here about talking with the press. You can’t say something and then ask for it to be off the record after the fact—it doesn’t work that way.

Arnold Barnett:
I’ve generally had positive interactions with the press and have rarely been misquoted. Sometimes I’m impressed at how carefully they quote. Once, I realized after an interview with The Washington Post that if I were identified as saying something, it could interfere with a security project I was involved in. I called back, apologized, and asked if they could treat it as off the record—even though I hadn’t said that initially. They did.

When you’re dealing with the New York Times, Washington Post, Wall Street Journal, or the major networks, they tend to be serious and professional. Sometimes the media gets a bad rap simply for printing things people would rather not have said.

Mark Graban:
Another thing I remember from your class was the comparison of driving risk versus flying risk—especially around holiday weekends. Has that changed?

Arnold Barnett:
Detroit to Chicago is about 200 miles. Now you have to consider COVID as a risk when flying. Planes are full, masking is optional, and maybe 15% of passengers wear masks. None of the crew do. I wear one, and I think people should based on CDC guidance, but they don’t.

In your own car, there’s no incremental COVID risk. Driving that distance on the interstate is very safe—Massachusetts Turnpike, for example, can go an entire year without a single fatality. Flying is extremely safe too, crashes are almost extinct—but for short distances like that, there’s no safety advantage to flying.

If you do fly, wear an N95 mask, don’t talk to the person next to you, and don’t eat or drink during the flight—save it for after you land.

Mark Graban:
Does accident risk when driving increase linearly with time on the road?

Arnold Barnett:
Fatigue can increase risk, but I haven’t seen data that breaks it down by hours driven. In healthcare, fatigue clearly increases error rates in long nursing shifts, so it’s reasonable to think the same could apply to driving. But there are complexities—like stopping for coffee could make you more alert later than earlier in a trip.

Mark Graban:
And, of course, laws for truck drivers limit how many hours they can drive per day.

Arnold Barnett:
Yes—partly to prevent unsafe conditions in a deregulated industry where drivers might be pressured to meet unrealistic schedules.

Mark Graban:
I also remember from class that the main risks in flying are during takeoff and landing.

Arnold Barnett:
Correct—most accidents happen during climb or descent. In-flight cruise accidents are rare. So a 2000-mile flight isn’t 10 times as risky as a 200-mile flight. But there are other factors—terrorists have tended to target longer flights at higher altitude, where decompression is not survivable.

Mark Graban:
You wrote an opinion piece in March 2022 titled Don’t End the Mask Mandate for U.S. Airlines, saying that ending the requirement then would be a serious mistake. What was the statistical basis for that view? And it sounds like you still agree—what would have to change in the data for you to say the mandate is no longer needed?

Arnold Barnett:
Part of what I was doing was responding to claims from the airlines. U.S. airlines do a fabulous job of preventing crashes, but on other safety risks, they don’t always distinguish themselves. United Airlines said the risk of getting COVID on their flights was “nearly non-existent,” and Southwest said it was “virtually non-existent.” That’s simply not true.

I cited multiple overseas flights documented in peer-reviewed medical literature where an infected passenger transmitted COVID to several others on the plane. Not everyone gets infected—you have to be seated relatively close—but the risk is real. Air filtration systems help, so someone in 8A is unlikely to catch it from someone in 20B, but if you’re in 20A or 20C, it’s another matter.

One example: on a Vietnam Airlines flight from London to Vietnam, an infected passenger in business class infected 11 of the 12 people within two rows. Other business-class passengers farther away had far lower risk. Airlines often claim people caught COVID in the airport lounge, but unless you think passengers arranged themselves in the lounge exactly as on the plane, that doesn’t make sense.

We don’t have comparable U.S. data because our contact-tracing systems are poor. Even if someone believes they caught COVID on a flight, it’s possible they were infected before or after. But absence of precise data doesn’t mean the risk is zero. When airlines say it’s “non-existent,” they’re misleading the public.

Mark Graban:
Even if you can’t calculate it precisely, recognizing it’s non-zero is worth acknowledging.

Arnold Barnett:
Exactly—especially when people are being told it’s zero. Delta Airlines, which had been doing well early in the pandemic, later said COVID was like seasonal flu and didn’t need to be treated as unusual. That got a lot of pushback, rightly so.

Mark Graban:
Switching topics—you’ve done work on elections and politics, particularly the Electoral College. Is the Electoral College a mistake, and can it be fixed?

Arnold Barnett:
I haven’t taken a position on whether it’s a “mistake”—that’s a political argument. What I’ve tried to do is propose a compromise that all sides could live with. Keep the current number of electoral votes per state, which overrepresents small states, but instead of winner-take-all, divide each state’s electoral votes proportionally to its popular vote.

For example, a state with four electoral votes that’s split 50/50 would award two to each candidate, rather than four to one side. I ran the numbers, and this approach gets you much closer to the national popular vote while preserving the small-state advantage—ironically, even increasing it slightly. Small states are split politically, so the advantage doesn’t consistently benefit one party.

It’s a mathematical solution that’s politically neutral. But whether it will be adopted soon? I doubt it.

Mark Graban:
And do people make assumptions about your political leanings just because you’re at MIT?

Arnold Barnett:
Yes. It used to be MIT was seen as technical and apolitical, unlike Harvard, which Joe McCarthy once called “the Kremlin on the Charles.” Now MIT is seen as more left-leaning, even “woke.” I think that’s unfortunate, but perceptions are hard to change.

Mark Graban:
There’s also the perception of MIT as “the university that World War II built,” with a history of defense work—people can pick and choose their narrative.

Arnold Barnett:
True. And MIT recently restored its SAT requirement, which got praise from some, like The Wall Street Journal, for acknowledging that objective measures of talent can have value. MIT isn’t viewed the same as Harvard, but it’s no longer seen as apolitical.

Mark Graban:
I remember overhearing older professors on campus in 1997 complaining about “well-rounded students” and saying “nerds make for good students”—a pro-nerd statement.

Arnold Barnett:
Yes. Now admissions ask about extracurriculars, which may not be relevant if the goal is to admit those who’d benefit most from an MIT education. Playing trumpet may not matter as much as being good at math.

Mark Graban:
Before we wrap up, I want to ask about the Leaders for Global Operations (LGO) program. You’ve advised many students—what do you enjoy about that?

Arnold Barnett:
My probability course for LGOs is shorter now to make room for machine learning, which I think was the right call. But advising internships has been a highlight for me—it’s interesting, it’s made me a better teacher, and I’ve learned things from students’ projects.

One example: a student at a paper mill in Wisconsin faced a problem where random tears in paper would weaken pages if two tears were close together. The question was, “Why does this happen so often?” The solution was related to the “birthday problem” in probability: even with few random defects, the chance of proximity is higher than you’d expect.

I’ve had wonderful experiences with LGO students—they’re a delight to teach. The program is strong and has great industry support because it’s good. I even remember being at an LGO alumni conference in Seattle on 9/11—the title of my talk was “Air Safety: End of the Golden Age?” That morning, the question mark seemed less necessary.

Mark Graban:
The program is great in part because of professors like you who are deeply involved. Thank you for that, and for joining me on the podcast.

Arnold Barnett:
It’s been a pleasure. I wish your listeners and viewers safe flights and happy landings.

Mark Graban:
Thanks, Arnie. For links to Arnie’s work and even the New York Times article from his “favorite mistake” story, see the show notes or go to MarkGraban.com/mistake169.

As always, thank you for listening. I hope this inspires you to reflect on your own mistakes, how you can learn from them, and maybe turn them into positives. I’ve had listeners tell me they’ve become more open about mistakes in their work and are trying to create workplace cultures where it’s safe to speak up about problems—because that leads to improvement and better results.

If you have feedback or a story to share, email MyFavoriteMistakePodcast@gmail.com, and visit MyFavoriteMistakePodcast.com.

---