DeepSeek Upends AI Competition, With Adam Segal

LINDSAY:
Welcome to The President's Inbox. I'm Jim Lindsay, the Mary and David Boies distinguished senior fellow in U.S. foreign policy at the Council on Foreign Relations. This week's topic is DeepSeek Upends AI Competition. With me to discuss how the release last month of a new artificial intelligence program by the Chinese firm DeepSeek has disrupted the convention of wisdom about AI, as well as to discuss the state of the U.S.-China technology race, is Adam Segal.
Adam is the Ira A. Lipman chair in emerging technologies and national security and director of the Digital and Cyberspace Policy program here at the Council. From 2023 to 2024, he was a senior advisor in the State Department's Bureau of Cyberspace and Digital Policy. An expert on national security, technology development, and Chinese foreign policy, Adam is the author of three books, including The Hacked World Order: How Nations Fight, Trade, Maneuver, and Manipulate in the Digital Age. Late last year, he co-wrote a piece for CFR.org on Artificial Intelligence Priorities for the Next Administration. Adam, thank you for coming back on The President's Inbox.

SEGAL:
Glad to be here.

LINDSAY:
Okay. Let's begin with the big question, Adam. Help me understand why DeepSeek's release of a free chat box app made such a big splash in the tech world.

SEGAL:
So I think we can break it into at least three categories where DeepSeek shook a number of assumptions: how we think AI is developing, the U.S.-China tech competition and where the two sides are, and effective tools that the U.S. can use to try to slow China down. And so, this announcement that DeepSeek has a reasoning model that's competitive with the U.S. kind of shook all of those assumptions.

LINDSAY:
So let's sort of, I guess, begin with the first one about AI development, because it seems to me that as 2024 closed, most of the articles I read about AI touted America's advantage. And supposedly, the size, the complexity of these models, the need for compute power, electrical support, datasets, and the like had created this broad moat around America that it was winning the AI race. Thirty days passed, and now all of a sudden, we seem to be worried that we're not winning the AI race. So just help me sort of parse out fact from fiction here. How big a challenge is AI or is the DeepSeek version of AI?

SEGAL:
Yeah. That's exactly right. So as you said, the assumption was the U.S. was far ahead, especially on large language models and machine learning, based on compute, the ability in particular of U.S. chips compared to China, how much energy is needed, which is a huge amount, and how much you had to spend to develop these models, which was in the hundreds of billions of dollars.

LINDSAY:
We're talking real money.

SEGAL:
Real money, and also questions about proprietary and open software, and DeepSeek seems to say, "Well, we did it actually much, much more cheaply." The number that was flying around last week and the week before was six million dollars, although we've now discovered that it's probably not true.

LINDSAY:
So the original reporting on six million was wrong. DeepSeek encouraged a view of its investment that really understated how much it had spent.

SEGAL:
Yeah. Well, the first report that was put out in the paper said, "Oh, six million dollars for the training of the model." But since then, a number of analysis has said, "Well, what they were talking about was the last stage of the training." And in fact, if you look at everything that they've invested in this program, we're looking at three hundred times more, 1.3 something billion that they're saying. So that number doesn't really seem to be-

LINDSAY:
So it wasn't as cheap as it seemed to be.

SEGAL:
No. It doesn't fundamentally say that if you have ten million dollars, you're going to be able to do this competitive thing.

LINDSAY:
What about arguments that DeepSeek actually borrowed, used technology from other companies?

SEGAL:
Yeah. So OpenAI said that they think that DeepSeek did what's called distillation, which is essentially getting data from OpenAI's model by asking it questions, and then using that to help train its own system.

LINDSAY:
Is that a legal thing to do?

SEGAL:
OpenAI says it's against the terms of use, but it seems to be fairly widely used in the field, that people do this quite often, and you use other people's models to help refine your model. So the data side also seems to be, how much did they borrow from and rely on others? And then on the chips side, again, some people have done analysis that says, "Well, in fact, they probably had fifty thousand Nvidia GPU, graphic processing units." So using many more chips than we had thought they had done for this system.

LINDSAY:
Let's step back a second on just that point, Adam, because that seems to be a big part of why the appearance of DeepSeek caused so much consternation in the tech world, and I should also say in the markets. Nvidia briefly lost something like six hundred billion in market capitalization when this news came out, and that's bigger than the market capitalization of most companies listed on the American stock exchanges.
But on this particular point, when we look at DeepSeek, the argument was that DeepSeek succeeded in creating this very robust chatbot without relying on cutting-edge Nvidia chips, which, since the Biden administration in 2023, we've limited the ability of Chinese firms to get. So sort of walk me through how it is that DeepSeek claims it built its model, because I've heard a lot about necessity being the mother of invention, and they did all kinds of clever optimization things, and I'm not sure, A, whether it's true or exactly how innovative it was.

SEGAL:
So it seems to be true, but the reporting last week and the week before overplayed the impact on the importance of the chips. So they did clearly make an important breakthrough on the algorithm and scaling on the architecture, was what the breakthrough happened on, which does reduce the demand for total chips.

LINDSAY:
Is that a good thing?

SEGAL:
It is a good thing for innovation broadly and AI innovation broadly. And I think when we talk about the U.S.-China tech competition more broadly and in AI, one of the things the Chinese argue is that they are going to make these technologies more accessible, especially to developing countries, and bringing the price down will make it more accessible to developing countries.

LINDSAY:
But not necessarily good news for Nvidia.

SEGAL:
Not good news for Nvidia, not good news if you think that AI is going to be used for nefarious purposes. So there are clearly trade-offs. But yes, for Nvidia, not good news, because the narrative had been, "You need extremely powerful chips, and these chips are really only produced by Nvidia." And then once that seems to not be completely true, then that certainly affects Nvidia's stock, and it affects the U.S. policy.

LINDSAY:
Right. And this was part of the supposed moat that existed around American AI, American companies, whether Amazon or Microsoft, what have you, had access to those chips and the Chinese didn't, or at least not to the cutting-edge chips.

SEGAL:
Yeah. And that was the moat that the Biden administration and, theoretically, the Trump administration will want to keep, and the way that they did that is to prevent the sale of those chips to China.

LINDSAY:
So I have to ask you, Adam, because I've heard this phrase used a lot in the last week, certainly more than I can recall hearing it for quite some time, and that is we are witnessing a Sputnik moment, sort of taking us back to the late 1950s when the Soviets were the first to put a satellite in space. That galvanized the American foreign policy establishment. We invested heavily in science and in military affairs. Is DeepSeek really a Sputnik moment, or is this sort of a bit of exaggeration?

SEGAL:
I think it's a bit of an exaggeration. It comes at a time when the rhetoric around this race, if you want to call it that, which I don't think is a great phrase, but is extremely high. Right? So we also had President Trump announcing, with OpenAI and SoftBank from Japan, this five hundred billion dollar initiative.

LINDSAY:
And Oracle, I believe, as well.

SEGAL:
And Oracle. This five hundred billion dollar initiative called Stargate, which is essentially AI infrastructure, so compute and energy.

LINDSAY:
This is initiative that Elon Musk then criticized on X saying they didn't actually have the money.

SEGAL:
They didn't have the money, and there's a long-running battle between Sam Altman of OpenAI and Musk about the structure of OpenAI and how dangerous AI is and other things. So they have this announcement. And so, some people have suggested that DeepSeek made this announcement soon after to deflate that, to deflate Nvidia and others. I think it's still too early to tell about how much of an impact the innovations that DeepSeek made is going to have on the larger trajectory of AI development.
I think even what we've been saying about the fact that it did take less money, it does use less energy, it did rely on fewer chips, the broad outline still seem to be pretty clear that you do require all of those things if you want to move forward. And you can see that from the head of DeepSeek himself. He said, "The biggest constraint we faced were the chips, and we're going to want more chips as we continue to scale and move forward."

LINDSAY:
How was the reveal from DeepSeek of this model treated in China? And I ask both from the terms of, say, national pride, but also in terms of the tech competition within China, because I would have to imagine that DeepSeek is not the only Chinese entity seeking to master or win the AI race.

SEGAL:
Yeah. So from the international competition perspective, the press and other tech entrepreneurs have really embraced nationalist pride. Right? "Here is this massive breakthrough. We did it under these conditions where you are trying to prevent us from getting chips, and we've shown all this innovation and breakthrough."
People that work at DeepSeek were trained in China. They're not people that were lured back from the States or lured from Taiwanese companies. So there really is a huge nationalist pride on that. DeepSeek is not a tech company. Right? So inside of China, it's a hedge fund company. It's a hedge fund company that created this data analytics AI project to support its hedge fund and its trading, and its culture is very, very different from the Chinese tech companies. Right? The Chinese-

LINDSAY:
So it was essentially like a financial services firm that decided to dip its toe in the tech world.

SEGAL:
Exactly. Exactly. And so, the tech culture in China has become not as attractive to young people. Right? The days of kind of saying, "Oh, you work for Alibaba or Tencent or those companies," and that being considered the coolest of the cool. Lots of people have complained about the culture. It's called 996. You work from nine to nine, six days a week, and you're expected to really just sleep at work and do all these things. And this culture, at least the reporting on it, is that people were allowed to be creative. They were allowed to pursue these interesting side projects. And so, that's what helped them innovate.

LINDSAY:
So in essence, in China, you have the disruptors disrupting other disruptors.

SEGAL:
Exactly. So this is a firm that... It's not Huawei, which is the Chinese have been talking about, "Well, they're the ones that are going to help us escape the constraints from U.S. export controls, and they're the national champion," and things like that. It's this company that nobody would have expected to play this role.

LINDSAY:
So as we look at this from a national security, foreign policy point of view in Washington, what are the lessons we should take away from this, Adam? Because one of the things I have read, and certainly have seen people talking about, is that this success by DeepSeek is an indictment of the U.S. policy of trying to deny China cutting-edge chips.
And I've seen two variants of that argument. Variant number one is that we had policies to deny the Chinese chips, but we didn't really enforce the sanctions, and the conclusion from that is, is that what you should do is actually enforce the sanctions. The other argument I have read is that this shows the futility of the sanctions approach, because in essence, what you're doing is requiring the Chinese to develop their own native technology, and they're actually pretty good at that, that necessity is the mother of invention. So in essence, we're encouraging the Chinese to get better at the area that we want to preserve as our superiority. How do you assess those arguments?

SEGAL:
So I think on the chips control side of things, it's a little too early to tell, because there clearly were loopholes that the Biden administration wanted to address. And we saw October 2022, they issued controls on chips, and then they revised them a year later in October 2023. And then on the way out, they issued another set of executive orders that were trying to control the diffusion of technology and, in particular, were worried about access to the cloud. So companies could train data on clouds without even having to worry about buying the chips.
So that makes sense, right? That's what happens in policymaking. You often try to figure out the technical details. You don't get them exactly right. You adjust, and it takes a while for hardware to be controlled. I think the fact that, as we mentioned earlier, the CEO of DeepSeek said, "Chips are a problem. And even now we've made this breakthrough, we still need chips for diffusion." And so, I think the chips controls are having an impact, and we're not focused just on narrow breakthroughs in specific technologies. We're also focused on the larger ecosystem. And in particular-

LINDSAY:
What does that mean, a larger ecosystem?

SEGAL:
Well, so there's an argument that Jeff Ding, who's a professor at George Washington, makes that we're overly focused on the breakthroughs in AI, and we should be more focused on the diffusion of AI, how companies use it, how it's spread in the economy. Right? So if you think about general purpose technologies, which AI probably will be used widely, the equivalent is electricity.
So it wasn't like electric ray guns that changed how war was fought. It was how it was used in all of these other things. And then, so chips are going to matter for that, how it's applied in other areas and other technologies. So we're as concerned about that as we are just the narrow breakthroughs, and slowing China down on chips will have an impact on that.
I think the second argument has a grain of truth in it in the sense of there is no doubt that the export controls are catalyzing changes in the Chinese system that are going to make it more innovative, that we don't really want to happen. But that is kind of a natural outcome, because, as you pointed it out, this is a national security and economic security priority for China. They are smart. They have a lot of resources. And so, what we've seen is changes in how China thinks about innovation.
And in particular, universities in China used to do very little work with the private sector. Research in Chinese universities was very, very theoretical. Chinese professors thought it was... They looked down on working with the private sector. That has changed because of the export controls. There's a huge amount of pressure for them to support what companies are doing. Same thing with Chinese state labs. They're connecting. So the innovation system in China is changing in ways that are going to make them more innovative in the long term.

LINDSAY:
Let me draw you in on that, Adam, because you're one of the world's leading experts on the Chinese technology ecosystem, and I hear sort of two competing stories. One story I hear a lot about is just the immense amount of talent in China, the immense amount of support from the central government to support technology. We had the whole Made in China 2025, Xi Jinping driving this process in which he wants China to dominate the cutting-edge technologies and industries of the 21st century, and that's, from an American point of view, intimidating, scary, concerned about where things are going.
The alternative point of view I hear, and you sort of implied it a little bit earlier, is that the Chinese system is sort of overworking people, that it may be very good at copying things, less good at innovating things. I keep hearing the term lying flat phrase for it in Mandarin essence, younger people becoming disenchanted with the system. How do you assess those two competing stories of what is happening in China?

SEGAL:
I mean, they're both true. Right? I mean, China is big enough that, in some ways, they're both true. We see, as you said, the state, and Xi Jinping in particular, focused on moving China up the value chain. And I think we can see that in the market between electric vehicles and new energy and batteries and all these other places where China has made incremental innovations over time, that have a real market effect, and aren't what we consider new-to-the-world, science-based innovation, but are important innovations in manufacturing and other spaces there.
But the system doesn't operate completely in the ways that China wants it to do. They looked at Silicon Valley for years and tried to create a similar kind of environment, like Route 128 in Boston. How do you get that environment where people work in a lab and then have this brilliant idea, and then get the venture capital and scale it to be a world-class company?
And that has been much harder for them, partly because the state still plays a large role, partly, as you said, competition is intense, and the companies, for a long time, competed with their partners by trying to drive them out of business by driving the price down and working everybody to death. So I think they're both true. But the idea that China can't innovate, that's no longer the case. In a number of critical technologies, Chinese scientists are innovating. Chinese entrepreneurs are innovating. It's just not happening at the scale that the Chinese state wants it to do.

LINDSAY:
We've been focusing on technology in the United States, innovation in the United States, innovation technology in China. Are there any other countries out there that can compete or are trying to compete in the artificial intelligence race, or is this just really two countries that are going to set the standard for the rest of the world?

SEGAL:
Yeah. There's a French company, Mistral, that is competitive, but the Europeans have tried to argue that they're going to compete on regulation. Right? So we have the EU AI Act, which is the first effort to regulate AI. But, of course, the European entrepreneurs are already complaining about it, saying-

LINDSAY:
Well, this seems to play to the stereotype of Europe. I mean, I often hear it described as, "In America, we innovate. In Europe, they regulate."

SEGAL:
Exactly. And so, on the positive side, Europeans will say, "We have what's called the Brussels effect. We set the regulations. Companies have to ascribe to them to enter the European market, and then they get globalized, and European standards play a role." But I think there's a sense that regulation is burdensome, and that's why Europe doesn't have these technology companies.

LINDSAY:
Well, certainly, the United States' approach right now, even more so under the Trump administration, is not to regulate these technologies. Is that a fair assessment?

SEGAL:
Yes. So the Trump administration came in. They immediately overturned the Biden executive order on AI, which talks a lot about safe, secure, reliable, trustworthy AI, and the Trump administration's argument was that, "These were overburdensome regulations, and we need to go as fast as we can." And so, I think the EU, in particular, is of two minds. I think the bureaucrats, of course, say, "This is our influence." The entrepreneurs are saying, "It's getting in the way."
The United Kingdom for a while was trying to bridge that gap. So they were saying, "Where are we going to build these safety institutes, but still be fairly open to the growth there?" There are some small players, but right now, I would say it really is a bipolar world in AI.

LINDSAY:
I'll just note you did not mention Russia, even though historically, Russia is very good at producing mathematicians, physicists, and the like. I take it they just don't have the, let's call it, ecosystem that's conducive to this kind of work.

SEGAL:
Yeah. They don't have the chips. Right? So they're relying on China right now for chips because of the sanctions with the-

LINDSAY:
Apparently, a lot of refrigerators are going from China to Russia so that the Russians can-

SEGAL:
Exactly.

LINDSAY:
... scavenge them for chips, but those are mostly for military purposes.

SEGAL:
Right. So those are much lower-level chips that can be used for drones and things, but they don't have the high-computing chips that they need. I think one of the things the Chinese do get with the close relationship with the Russians is access to Russian math skills and the kind of historical sense of the Soviets did have some real centers of technological and scientific might. But no, Russia is really not a player in this space.

LINDSAY:
Okay. So we really have sort of a two-actor, bipolar world when it comes to AI. Can we actually talk in any meaningful sense about winning the AI race? And I ask that because I see a lot written, Adam, which seems to be focused on this notion that there is a goal of winning, that if you win the AI race, you now dominate the 21st century, and I'm not sure where that comes from or what it actually means.

SEGAL:
Yeah. So I think when people say Sputnik or moonshot, they think there is an end point. Right? You end up on the moon before the other side does. I don't think that's going to be the case with AI. I think we still don't really know where AI is going to end up. Right? So there's those who worry that we're going to end up with artificial general intelligence, which is going to be human-level types of intelligence.

LINDSAY:
Just unpack that for me, because I hear a lot about AGI, and it's not clear to me what we're talking about here.

SEGAL:
So essentially, we're beginning to see some sense of that with some of the models we see where you can give it all types of different problems, because most of the artificial intelligence systems we saw before were trained on a specific set of data, and it could answer lots of questions, say, about how to place an ad, but then if you asked it, "Well, fold this protein," it couldn't do that. It was a very specific system.
But now we're building systems increasingly that you can put into all different types of situations, and seem to do some type of reasoning like human minds do, and will be faster and smarter than humans, can do it at a scale... All those things. So that is still an open debate. Right? You hear from some of the boosters of AI that's coming in two years, five years. They've been saying that for a decade now, but that's what we keep on hearing. It's going to be-

LINDSAY:
That presumably will put us out of business.

SEGAL:
What it's going to do to all of us? That's also the open question, is how expansive is it going to be, and what is it going to mean? Right now, all of us who've probably interacted with it can see the strengths and the weaknesses of it. Right? The systems tend to hallucinate. Sometimes they are pretty lowest common denominator, don't really show a lot of creative thinking.
But all of that, when you talk to people that work in the field and they talk about the scale and speed of the changes, that's where people start saying, "We don't really know where it's going to go," or it could be where AI in the end turns out to be much more narrowly used. It is great on helping fold proteins and very specific problems, but is a tool that humans can really relate to. And then there's also a sense that if you get there first, that somehow you can lock in those-

LINDSAY:
First-mover advantage.

SEGAL:
Yeah. You can somehow control the chips, and that the scale is going to be so large that nobody will ever be able to-

LINDSAY:
You'll have a big moat around your technology.

SEGAL:
And that, I think, is, DeepSeek kind of reminds us that that probably is not going to be the case. I think if you-

LINDSAY:
So disruption will not end?

SEGAL:
Disruption will not end, and diffusion will happen. Right? I think if we talk to our colleague Mike Horowitz, who wrote a lot about technology diffusion and its impact on military power, I think he thinks right now, the models are very likely to diffuse, and whatever advantage, whoever gets there first is probably going to be short-lived.

LINDSAY:
Well, let me ask you about that, because that seems to go to a core principle behind U.S. policy toward China on an economic front, military security front, and that's this notion of decoupling or de-risking. I understand that you can limit the exchange of chips, and that'll work as long as the Chinese don't catch up to that technology. But it seems to me that we're in an environment which if you have open-source models and you can have this, I think what you call, distillation, I would call it piggybacking, that is going to be very hard to sort of shut China out of the innovation ecosystem.

SEGAL:
I mean, I think from the beginning, people who have been... Even the promoters of export control and decoupling or de-risking said that alone is not enough. Right? And so, you have to continue running faster, and that means Stargate or other types of investment in U.S. capacities. Now, the question has always been, as you said, the U.S. and Chinese systems are tightly interlinked, and you can look at flows particularly of people training for AI.
There's just a lot of back-and-forth between China and the U.S. There was a lot of joint publication and joint research. Open source, as you said, everybody puts their stuff out there. So that is a harder line to kind of draw. How do you make sure that the U.S. system remains open enough to take advantage of all the things that are happening outside of China, but close enough that you can try to slow China down? And that's not easy policymaking.

LINDSAY:
Well, how do you do it? I mean, I guess the question facing the Trump administration is, how do you respond to the reality of this technological competition? Do you simply say, "Let it rip," and just tell the people in Silicon Valley to do what you can and just pedal to the metal, or is that an unwise strategy, either because you may end up producing something that's reckless? And this gets back to the debate you alluded to earlier within the AI community as to whether it is a great boon to humanity or a great threat to humanity.

SEGAL:
Yeah. I think you can thread the needle. I don't think we want to race to the bottom, so to speak, or race as fast as we can. And I think some types of regulation, in fact, could help innovation, because you give a certain stability. We know the companies are worried about fifty different types of regulation if each state starts saying, "We're going to control AI." So you want a national federal law.

LINDSAY:
But couldn't AI just solve that? You tell AI?

SEGAL:
Generate fifty different policies? It could do that. So I think that was a positive thing. I think we don't know about the risks yet. And so, certain things about red-teaming models and making sure that they're secure and all of those things, I think the companies have accepted on voluntary standards, but we need some kind of national standards there, I think, as well.
The other major tension that we saw in the administration already is the degree of openness, and we saw that around immigration and H-1B visas. So people like Elon Musk and Vivek Ramaswamy said, "We need to attract the best and the brightest to the United States. We don't want to keep them out." And parts of the Trump coalition want to restrict all types of immigration. So that, I think, is going to be a tension moving forward.

LINDSAY:
Let me ask you specifically about that, Adam, because one of the concerns about that process is that people come to the United States, work in companies or go to universities. They get the skills. They get information, but then they go back home. And there's been a particular concern about Chinese nationals coming here and then going back home, and you're ending up basically preparing your rivals to challenge you.

SEGAL:
For the longest of times, they wanted to stay. Right? There was all these studies done in the '90s and the early aughts of startups in Silicon Valley. And I don't remember the exact numbers, but they were in the sixty to eighty percent range that startups were started by either Indian or Chinese PhD students in the States. And when you asked them if they wanted to stay, they would say, "Yeah. We pretty much want to stay fifteen, twenty, thirty years."
Now, there were definitely concerns about Chinese scientists, often with ties to military universities, coming to the States studying specific things, and tightening those controls, I think, really makes sense. But the immigration and the argument that they may one day go back, I think we really benefited more than China did.
In fact, we can see the growth of all those talent plans in China, an attempt to try to lure them back, was a signal of that, is how badly the Chinese wanted to get them back and how hard it was, because they wanted to stay, and especially as we talked about how tech culture in China right now and kind of the growing sense of malaise among a lot of urban middle class in China right now should be an opportunity for us, although right now, we're not that welcoming.

LINDSAY:
Let me close by just asking you about how optimistic you are that the United States will remain at the forefront in the AI race, and I ask that against the backdrop, I'm old enough to remember when the American automobile industry dominated the world automobile industry, and that no longer is the case. Are the signs optimistic, positive for continued U.S. tech dominance, or are we likely to see it eclipsed?

SEGAL:
I mean, AI, I think we will continue to lead, be at the cutting edge, certainly for the near- to midterm, let's say, five to ten years. I think there are lots of issues in the U.S. innovation system that we've identified for fifteen years. We haven't done a lot to address. So funding in basic R&D has been flat. We know that the movement away of manufacturing from where innovation happens is really bad, even though that's what we thought was going to help us remain competitive. We want people to actually be able to manufacture these things.
We've had a long debate about people not wanting to study STEM subjects, and how do you increase the enrollment there? So we know kind of what the policy answers are to those questions, but getting the domestic support for them has never been easy. We see bursts right under the Biden administration. The CHIPS Act addresses some of the manufacturing, addressed some of the basic R&D around those things, but sustaining that over time and doing it at the scale is harder, given the domestic polarization, and not at the scale of what we face from the challenge from China.

LINDSAY:
On that note, I'll close up this episode of The President's Inbox. My guest has been Adam Segal, the Ira A. Lipman chair in emerging technologies and national security and director of the Digital and Cyberspace Policy program here at the Council on Foreign Relations. Adam, as always, it was a delight to chat.

SEGAL:
Thanks for having me on again, Jim.

LINDSAY:
Please subscribe to The President's Inbox on Apple Podcasts, YouTube, Spotify, or wherever you listen, and leave us a review. We love the feedback. The publications mentioned in this episode and a transcript of our conversation are available on the podcast page for The President's Inbox on CFR.org. As always, opinions expressed on The President's Inbox are solely those of the host or our guests, not of CFR, which takes no institutional positions on matters of policy. Today's episode was produced by Justin Schuster, with recording engineer Robert Micheels, and director of podcasting, Gabrielle Sierra. This is Jim Lindsay. Thanks for listening.