A conversation with Paul Scharre, author of Four Battlegrounds: Power in the Age of Artificial Intelligence, who joins us to talk about

* how AI’s superhuman command and control abilities will change the battlefield

* why offense/defense balance isn’t a well-defined concept

* “race to the bottom” dynamics for autonomous weapons

* how a US/taiwan conlict in the age of drones might play out

* and more…

Links mentioned: Gradual Disempowerment, Swarms over the Strait

Transcript

Rai Sur 00:00:36

Today we’re speaking with Paul Scharre. Paul is the Executive Vice President at the Center for a New American Security and the author of Four Battlegrounds: Power in the Age of Artificial Intelligence. He also formerly worked at the Pentagon on emerging technologies. Welcome, Paul.

Paul Scharre 00:00:52

Thanks for having me.

Rai Sur 00:00:53

We’re also joined by Sentinel forecaster and co-founder Nuno Sempere, Sentinel Superforecaster Lisa, and Superforecaster Scott Eastman, whose specialty is in geopolitics, epidemiology, and AI. Welcome, Scott.

Paul, what is command and control, and how do you see AI significantly altering it?

Paul Scharre 00:01:18

The term command and control is used by militaries to describe the internal organizational and informational processes used to organize military forces and coordinate their behavior. Militaries work in a very hierarchical format, with teams, squads, platoons, companies, and battalions. The way they communicate has evolved from signal flags on the battlefield to radio communications, to today’s use of computers.

Improvements in command and control can yield dramatic improvements on the battlefield, even if the military forces themselves are the same. We often think about the physical hardware of drones, missiles, or robots, and that’s absolutely valuable. But there are also potentially transformative effects in command and control.

If a military can gather more information about the battlefield, make sense of it faster than the adversary, make better decisions, and execute those decisions in a coordinated fashion, it can have dramatic effects. This is particularly true if they can change the battlespace faster than the adversary can react, leaving the adversary constantly trying to respond to what happened six hours ago.

Nuño Sempere 00:03:04

What’s a concrete example of this?

Paul Scharre 00:03:11

During the US advance into Iraq in 2003, US military forces were advancing very rapidly toward Baghdad. Through a precision bombing campaign, the US disrupted the command and control of Saddam Hussein’s forces by taking out headquarters and radio communications.

This created a situation where Iraqi commanders were reacting to events after they had already happened. They would identify the location of US troops, but by the time they reacted, the US forces had already advanced. That’s a potential advantage of having better situational awareness and command and control. Artificial intelligence has a lot of potential value in this space.

Rai Sur 00:04:11

In Four Battlegrounds, you wrote about how AI might change decision-making and strategic planning at higher levels. What are the potential impacts there, and how could it change the look of warfare?

Paul Scharre 00:04:25

Let’s imagine what this evolution might look like 15 to 30 years down the road as militaries integrate more AI, autonomy, drones, and robotics. Military advantage can come from better technology—better drones, larger production capacity, more robots on the battlefield. We tend to think about the physical object, which is important. But major advantages can also come from improvements in command and control, and we can see examples of this in gaming.

Look at how AlphaZero plays chess. The pieces are the same for both the human and the AI, so there’s no material advantage. In many cases, the situational awareness is the same—they’re looking at the same board. Yet we’ve seen that AI is able to process that information better and more holistically than people. In games like StarCraft or Dota 2, the AI can see the big picture and comprehend it all.

Across a number of games, AI agents can engage in coordinated, multi-axis attacks and balance their resources more effectively than people. This isn’t just in real-time strategy games; chess grandmasters have noted that AlphaZero can conduct attacks across the whole board more effectively. We’ve also seen AI systems engage in superhuman levels of calibrated risk-taking. In chess, this can look like ferocious attacks. In StarCraft and Dota 2, human players have talked about feeling constantly pressured by the AI, never having a moment to rest.

In poker, AI systems bet differently than humans, engaging in wild betting swings that are hard for even cold-blooded poker players to manage because of human emotions. There’s an information processing element, but also a psychological one. In combat, you get lulls in the action because people need to rest and reset. AI systems don’t. In the command and control aspect, AI has the potential to not just be better than humans, but to transform the very strategy and psychology of war.

Rai Sur 00:07:53

What does this look like in practice, at the level of grand strategy, where generals use game theory and deception? What does adoption look like there, and what is the minimum capability required? Naively, it sounds like an AGI-complete problem.

Paul Scharre 00:08:27

There are many beneficial things militaries could do on the command and control front that are not AGI-complete problems. If you hand over all your military forces to an AI, you need a lot of trust in its intelligence to handle that complexity.

But you can imagine very narrow algorithms that optimize logistics or communication on the battlefield, which would have significant advantages. You could also have more advanced algorithms that assist with planning military operations while commanders remain in the loop. The AI could generate plans for offensive or defensive operations, and commanders would still review and decide what to execute. Over time, they might see that the AI’s plans are better than what humans come up with and decide to trust them more.

Rai Sur 00:09:29

Who is best positioned to gain these advantages sooner rather than later?

Paul Scharre 00:09:35

In general, more technologically advanced militaries have some advantages, but we’ve seen AI technologies diffusing very rapidly. The biggest limiting factor for militaries isn’t getting access to AI technology, but finding the best ways to use it and transform how they operate.

The history of military-technical revolutions shows that what matters is not getting the technology first or even having the best technology, but finding the best ways of using it. An interesting historical example is British innovation in carrier aviation during the interwar period. The British military was ahead in inventing carrier aviation but fell behind Japan and the United States due to bureaucratic and cultural squabbles. It was primarily a technology adoption problem, not a technology creation problem.

This suggests that militaries in active conflicts, like we’re seeing in Gaza and Ukraine, are much more incentivized to overcome these barriers to adoption. They’re also getting real-world feedback on how effective these technologies are, whereas getting those feedback loops in peacetime is much more challenging.

Nuño Sempere 00:11:22

So you would greatly discount which countries have top AI labs as a factor, because you think adoption feedback loops are more important?

Paul Scharre 00:11:32

Yes, absolutely. The top US AI labs are maybe a few months ahead of the top Chinese labs—perhaps 12 months, but not three years. The US military is, charitably, five years behind leading AI labs in adopting the technology, and realistically more like 10 years behind in actually using it.

It’s not enough to have an image classifier on your drone video feed; you also have to use it to transform your intelligence operations and make your analysts more effective. That’s a much longer process, particularly in peacetime without competitive pressures. Having leading AI labs in your country is probably not the most important factor for battlefield operations. For example, what Ukraine is doing with drones is far more innovative than what the US military is doing. That pressure is more important.

The exception might be in niche areas like offensive cyber operations. There, you might imagine that access to the most advanced AI systems could provide large, step-function increases in capability. If the NSA has a 12-month head start on Chinese competitors and can actually employ that advantage, that’s a place where a 12-month lead could matter a lot.

Rai Sur 00:13:21

And if the difference-maker is having a place to apply it and get feedback quickly, cyber operations don’t require a hot conflict. You are constantly searching for intelligence and testing capabilities against adversaries, so those feedback loops are always active.

Paul Scharre 00:13:44

You have more feedback in cyberspace during peacetime because of the constant engagement between intelligence services. That engagement may not be as robust as it would be in wartime—many constraints would come off in a wartime environment, leading to more cyber operations and tighter feedback loops. But you probably have more feedback in cyber operations now than you do in ground warfare for the US military.

Scott Eastman 00:14:26

My understanding is that tens of thousands of drones might be used in the Ukraine war in a 24-hour period. Is a primary purpose of AI to assimilate all of that information in a meaningful way? An individual soldier might have access to their own drone feed, but how do you integrate that across the entire battlefield? Is AI critical for filtering that much information into something usable?

Paul Scharre 00:15:02

This is probably the area where AI is most useful, getting back to command and control. AI could do a couple of things here. First, simply improving sense-making and situational awareness over the battlefield would be incredibly transformative. Instead of one pilot looking through the “soda straw” of a drone camera, AI could process all that information collectively and say, “Here are the positions of all enemy fighters and vehicles along the front.”

At a higher level of abstraction, it could even analyze current positions relative to historical data to suggest anticipated behavior, like, “It looks like they’re massing for a potential attack.” The physical presence of drones is already enabling a persistent network of aerial surveillance that makes the battlefield more transparent and harder for forces on either side to mass for an assault. You need to concentrate your forces to be effective, and drones make that harder to do without being detected.

Having AI to process that information would provide a lot of advantages. Taking it a step further, having AI help make decisions in a coordinated fashion would be extremely effective. Right now, drones have made the battlefield not just more transparent, but also more lethal for ground forces. But those are individual drones. Getting to the point where you have a coordinated swarm of 10, 50, 100, or even 1,000 drones coordinating their behavior could be much more lethal and effective.

Scott Eastman 00:17:34

I’ve heard that about 70% of Russian casualties are now from drones. I don’t know if that’s corroborated, but that’s recent information I have.

Paul Scharre 00:17:46

I have not heard that figure, but it’s certainly possible and would say a lot about the effectiveness of drones. However, part of that could be because the Ukrainians have suffered from a lack of artillery ammunition. They have been desperately asking the US and Europeans for more 155mm artillery shells. In a world where they had more artillery, you might see that number balance out. They’re likely compensating for the lack of artillery at the moment.

Rai Sur 00:18:15

The fact that it’s harder to amass a ground force for an offensive seems to make the offense-defense balance of drones lean more toward defense. Is that the net effect, or are there other factors that make drones more offense-leaning? What is the net effect of drones on the offense-defense balance in warfare?

Paul Scharre 00:18:41

Unbeknownst to you, perhaps, you have opened a massive can of worms. I do not think the concept of offense-defense balance is analytically tractable, but let’s dive into why.

You have drones overhead that make offensive maneuver operations harder. At a tactical level, they seem offense-dominant; if a drone finds an infantry troop in the open, that troop is toast. But operationally, the second-order effect is that it seems to favor the defense. So are drones offense-dominant or defense-dominant?

I want to disaggregate the idea of offense-defense balance into three separate, more meaningful concepts. The first is the cost-exchange ratio of a technology. For example, what is the cost of shooting down an incoming drone versus the marginal cost of buying more drones? We’ve seen examples in Ukraine where Russian drones are so cheap that even though Ukrainian air defense systems can shoot them down, they don’t because their missiles are too expensive. If you shoot down a $1,000 drone with a $1 million missile, you’re losing every time.

A second concept is the relative cost of power projection. What does it cost for militaries to engage in offensive maneuver operations on the ground or at sea? The cost to launch an amphibious invasion is already very high. Better anti-ship cruise missiles or sea mines increase that cost further. The Chinese military has made a series of investments in long-range radars and anti-ship missiles that make it more difficult for the US military to project power into the Western Pacific, thereby increasing the cost of power projection.

The third concept is first-mover advantage. In the era of precision-guided missiles, there’s a significant first-mover advantage in naval surface warfare. One effective missile strike can take out a ship, so striking first can be a major advantage. Contrast that with nuclear deterrence, where militaries have invested heavily to take away that first-mover advantage. A submarine is designed to do just that. Even if you launch a surprise nuclear attack, you won’t get all the submarines at sea, which can then launch a second strike.

The answer to your question depends on what we’re talking about specifically. It’s really hard to connect tactical effects to strategic outcomes. A stealthy submarine is, in some ways, very defense-dominant, but that enables it to carry out a nuclear strike that could take out a whole country. That single action is very offensive, but the net effect is stabilizing for nuclear deterrence.

Nuño Sempere 00:25:32

I understand your framework is better, but couldn’t you still simplify it? If your country is being invaded, does having drones make things better or worse for you as the defender? It seems like there should be a straightforward answer.

Paul Scharre 00:25:58

If you’re worried about being invaded, what you ultimately care about is whether this technology makes it easier or harder for an aggressor to gain territory. But it’s very hard to look at a technology and extrapolate to that level. Decades ago, defense analysts tried to map which technologies were offense-dominant versus defense-dominant, and it’s difficult to do.

Drones are a good example. At a tactical level, you might say drones are offense-dominant. You can imagine drone swarms overwhelming defenses. But what we’re seeing in practice in Ukraine is that drones seem to be contributing to the stasis on the front lines, making it harder to amass ground forces. They’ve made the battlefield more transparent and more lethal.

The role of drones today is analogous to the role the machine gun played in World War I. The machine gun dramatically increased lethality on the battlefield, enabling a hundredfold increase in the number of bullets a defender could put downrange. This made traditional tactics of advancing forward effectively impossible. It took militaries a long time to adapt. They eventually developed a more integrated system of suppressing fire with artillery and machine guns, and then maneuvering to flank the enemy.

I think something similar will unfold over the next 10 to 15 years in ground warfare because of drones. Drones are making ground warfare three-dimensional. While ground troops have worried about aerial attacks since World War II, those threats were transient. Drones create a persistent threat from the air. Over time, militaries will have to adapt by finding new ways to conceal themselves and carry out offensive operations. I don’t think they’ve figured that out yet.

Lisa 00:30:19

With the emergence of low-cost drones, people initially thought they could level the playing field for the underdog and favor asymmetric warfare. But it’s clear from Ukraine that Russia has realized the power of the drone and is vastly ramping up production. The future holds drone swarms conducting both offensive and defensive operations. These low-cost drones can be used by either side to have a large-scale effect on the battlefield.

A Russian commander recently said that the winner in Ukraine will be the side that develops the best drone and anti-drone technologies. Where do you think that is going? As we see larger drone swarms and increasingly autonomous drones, what is the future of anti-drone technology? It won’t be a guy shooting a drone down with his pistol. What do you see?

Paul Scharre 00:32:40

That’s a meaty question. To answer the initial part about who benefits more, I think that while sophisticated nation-states can put more resources into the technology, on net, drones favor less capable actors. They get more of a relative uplift in power.

Drones are essentially lowering the cost and technological barrier to air power. Before drones, to have air power for surveillance or attack, you needed to build an airplane big enough to put a person in it. Those military aircraft are incredibly expensive to build, maintain, and train pilots for. Drones completely change the economics of that. For a couple hundred bucks, you can put a drone in the air that gives you the ability to see the enemy and call in an artillery strike, which is incredibly valuable. Both Ukraine and Russia are able to field air power in the form of drones in a way that would be cost-prohibitive with crewed aircraft.

You’re right that there are already counter-drone defenses. A funny anecdote: a buddy of mine shot down a drone in Iraq back in 2004. He was all excited. I had to tell him, “You know that was one of ours, right?” He said it didn’t matter; he got a drone kill.

We’re seeing more investment in counter-drone technologies. There’s a lot of jamming on the front lines in Ukraine because the communications link to a remote pilot is a point of vulnerability. There are lots of ways to go after a drone. You can shoot it down with expensive missiles for large drones, or with bullets for smaller ones, but you still need a way to find it and aim effectively. People are developing systems for that, as well as drones designed to intercept other drones.

Lisa 00:37:51

If I may interrupt—one thing we’re seeing in Ukraine is the emergence of increasingly autonomous drones that will no longer be susceptible to jamming. They can navigate by landscape features and don’t need to communicate with a base. What do you do with that?

Between that and first-person view (FPV) drones using long fiber-optic cables to communicate, it seems like a game of cat and mouse where the options for countering drones keep narrowing. It seems that drones that attack other drones will have to be an increasingly important strategy.

Paul Scharre 00:39:14

You’re exactly right. These are the innovations we’re seeing, and you can see the direction the technology is evolving. Autonomy is a good example. The communications link is vulnerable, so you jam it. In response, you integrate autonomy into the drone.

When I was in Ukraine last year, I saw a demo of what they called an “autonomous last mile” solution, which is basically autonomous terminal guidance. The human pilot locks onto a target, and then, as the pilot demonstrated, literally takes his hands off the controls. The drone follows the target, even a moving one, for the final attack. That helps if there’s jamming in the last kilometer or so.

You can see the next evolution of that: a drone that can just hunt over a wider area and find targets on its own. That will lead to other countermeasures. If the drones are fully autonomous and you can’t jam the link, you have to find a way to shoot them down, fry their electronics with high-powered microwave weapons, or blind their optics with lasers.

Nuño Sempere 00:40:46

If you make the drones fully autonomous locally, doesn’t that substantially raise the quality of the chips they’ll need?

Paul Scharre 00:41:01

You don’t need a fancy Nvidia GPU to do this. The “fancy” version would be to use a machine learning-based image classifier to identify objects like a Russian tank. The “unfancy” version is to just do a pixel lock. You lock onto a group of pixels against a background and keep them centered until you collide with the target. That doesn’t require a lot of sophistication. I don’t need to know what the object is; I just need an algorithm that can identify the boundaries of the object I’ve locked onto.

Rai Sur 00:41:49

Then who is deciding on that lock? At some point, it has a link back to a human who is saying, “Here’s the pixel blob. Track it and attack it.” And from that point on, it’s autonomous.

Paul Scharre 00:42:03

It depends on how autonomous you want it to be. If you’re talking about autonomous terminal guidance where the human picks the target, you don’t need something very complicated. The vulnerability there is that if the target hides, you might break the lock. If you want something more autonomous where you just launch it and say, “Go get them,” with no human interaction, then you need an image classifier to identify, for example, a Russian tank. But you don’t need a lot of sophistication there either. The question becomes how much compute you need to run an image classifier with a few key object classes. My guess is probably not that much.

Scott Eastman 00:43:02

If you’re dealing with heat signatures, probably even less.

Paul Scharre 00:43:07

Right. If you’re going after tanks, they’re big, hot things out in the open. As long as you box the drone into an area where the enemy is and not your own troops, there probably aren’t a lot of false targets.

Scott Eastman 00:43:27

I was thinking of a problem that is complicated for humans and would also be complicated for AI. In the background of the Ukraine war, there’s always the threat that if things go too horribly for Russia, they might use tactical nuclear weapons. If it’s just a chess game of winning territory, an AI could probably do better than some humans. But at what point would a human need to get in the way and say, “We don’t want to go past a certain point because it may create a massive political backlash that goes beyond a tactical victory”?

Paul Scharre 00:44:17

I agree. This is an area where it’s hard for people—security analysts and political experts disagree about how much the US should be willing to push Putin, for example. We should expect AI to do quite poorly here for a couple of reasons.

One is that war games using large language models have shown them escalating needlessly, sometimes just playing out a narrative script. But setting aside the weird neuroses of LLMs, the more fundamental reason AI would do poorly is that there’s no good training data. It’s not like we have 10,000 instances of military conflict and can see which ones led to nuclear war. Human experts look at the history of nuclear brinksmanship and come to wildly different conclusions. Some think nuclear deterrence is solid, while others think we’ve just been really lucky. These are areas where we should expect AI to do very poorly.

Rai Sur 00:46:06

This ties into the race-to-the-bottom dynamics. There are many advantages here, but there’s a tension for any actor adopting this technology between getting the benefits and understanding that the more power they give it, the more likely they are to encounter a bad, out-of-distribution outcome. How likely is it that we see some coordination or intentional pullback? Is the equilibrium just to go as fast as possible?

Nuño Sempere 00:46:59

That question assumes that as you give AI more power, the error rate increases. But humans also have an error rate, and it’s not clear that you do get a race to the bottom, because the system might just become more accurate than humans.

Rai Sur 00:47:22

We do see examples of accidents when using machine learning in military applications, like adversarial examples that can fool image classifiers. These could be deployed to take advantage of an enemy’s model. I’m extrapolating from the existence of these errors in past systems to assume they will remain at higher levels of capability.

Nuño Sempere 00:47:55

But the US accidentally dropped a nuclear bomb off the coast of Spain. And Waymos are far safer than human drivers.

Rai Sur 00:48:13

I think the difference there is design versus growth. The US dropped the bomb because a mechanism they designed failed. They had a good story for why that mechanism should work because the system is decomposed into smaller pieces that humans designed and understand. The error is in execution, not in understanding the system’s design. With machine learning systems, you are “growing” something based on feedback from a training distribution, so there’s more uncertainty in the method itself.

Nuño Sempere 00:49:02

The case of a self-driving car that’s safer than humans sort of sidesteps all that. Even if a battlefield is far more chaotic than the streets of San Francisco, humans still make mistakes. I’m curious for Paul’s perspective: where do you think it makes sense to hand over more decision-making now and in the next few years?

Paul Scharre 00:50:13

Let’s consider two archetypes: the self-driving car where AI gets better than humans, and the war-bot that goes berserk. The conditions under which each of those worlds becomes true are different, and you can find military applications that match both.

For self-driving cars, several factors favor improving their performance to be better than humans. First, the baseline for human driving is terrible. Second, we can get real-world feedback by putting cars on the road in their actual operating environment. We can collect that data, run it in simulations with varied conditions, and build up a very large training data set. The environment is semi-adversarial—people might cut you off, but they aren’t generally trying to hit you.

Some military applications fit this model, like automated takeoff and landing for aircraft on a carrier. We’ve demonstrated we can do that, and there’s no reason to have humans doing it anymore. In any area where we can get good training data and have clear metrics for performance, we are likely to be able to train AI systems to do better than humans.

The places where we’ll see AI struggle are where we don’t have good training data, where the metric of performance is context-dependent, or where it requires human judgment. For example, a drone looking at an object in a person’s hands—is it a rifle or a rake? That is a factually correct question. I can build a training data set for that and test it.

But a decision like, “Is this person a valid enemy combatant?” is super hard. It’s very context-dependent. Are they holding a rifle? That doesn’t necessarily mean they’re a combatant. Are they not holding a rifle? They could still be a combatant. Context, human judgment, and what was happening in the area a few minutes ago all matter. It’s hard to build a training data set for that. The same goes for things like a surprise attack. That’s an area where we would expect AI systems to fail, and where I would want militaries to be much more cautious.

Nuño Sempere 00:55:40

What’s a concrete example of this “race to the bottom” dynamic?

Paul Scharre 00:55:59

One example could be lethal autonomous weapons in Ukraine. Let’s fast-forward a few years. The jamming environment becomes so heavy that both sides decide it’s better to deploy unreliable autonomous weapons than no drones at all. You might see accidents where they strike the wrong target, but it’s deemed a necessary risk. Then that technology gets used in other conflicts where there are more civilians, leading to casualties.

Another concern could be around safety. The US builds an autonomous drone boat and sends it to the South China Sea. China races to get its drone boat out there. Then the US decides it needs to deploy a swarm of boats. Both sides might cut corners on safety because they are in a race to get ahead of the other. The institutional incentives for militaries are all about staying ahead of their competitors. For a proof of concept, look at what leading AI labs are doing today: rushing to push out products that are not reliable and safe just to get ahead of competitors.

Lisa 00:58:13

We’ve seen the importance of drones in Ukraine. Looking ahead to a potential invasion of Taiwan, what do you see as the potential for drone usage on both sides? Do you have a sense of where their capabilities are? China is learning lessons from Russia and is effectively training a large number of drone operators through civilian industries.

Also, what potential threats could China pose to the United States in that situation? We know Chinese vessels could position themselves at our ports—ostensibly civilian cargo ships—with containers full of anything from missiles to drones. Where do you see that going?

Paul Scharre 00:59:40

A colleague of mine, Dr. Stacie Pettyjohn, wrote a report on exactly this issue called Swarms Over the Strait. For anyone interested, I’d say go check out that report.

The fundamental challenge in any US-China fight over Taiwan is the asymmetry of geography. Taiwan is an island 100 miles off the coast of the Chinese mainland. The US has to project power across the entire Pacific Ocean. That gives China a huge advantage in projecting air and naval power. They have layered air defense systems right across the strait and aircraft well within range. The US has to rely on a limited number of aircraft carriers, which are vulnerable targets, or long-range bombers.

This tyranny of geography has huge implications for drones. In Ukraine, you’re right on the front lines, so you don’t need long-range drones. That won’t work for the US in a Taiwan conflict. Any drone the US uses has to be brought to the fight, either on ships or in submarines, which means they’ll need to invest in more expensive drones. Some could be pre-positioned on Taiwan, but that geographic asymmetry is very hard to overcome.

For Taiwan, small drones could be very beneficial for targeting Chinese troops if they land on the beaches. For the US, there’s also the problem of resupply. In Ukraine, the US is arming Ukrainians across a land border with Poland. None of that exists for Taiwan. Drones can certainly be helpful, especially if you narrow the operational problem to finding and sinking Chinese invasion ships in a limited time and area. But the geography is a huge factor.

Lisa 01:04:43

I was also wondering about the importance of pre-positioning drones, and whether China might pre-position drones at US ports as part of a threat. If they were to invade Taiwan, they might line up a number of threats to the US—cyber threats, threats at our ports—and say, “Look, you can let us have Taiwan, or all of these things will happen.” I wonder about the possibility of drones being pre-positioned at US ports in that kind of scenario.

Scott Eastman 01:05:29

That seems like the situation we’re already in, where China, Russia, and America can all hit each other with nuclear weapons. It’s a political decision about whether they are willing to risk escalation. Certain things, like attacking the leaders of other countries, are usually kept off the table, but there’s nothing that says they have to be. A country could decide Taiwan is important enough to take that risk.

Paul Scharre 01:06:03

One of the interesting things about drones that changes this political calculus is that you can put them into a contested area without risking a person. We see this with recent Russian drone incursions into Poland. It creates a rung on the escalation ladder that didn’t exist before. Flying a piloted aircraft into Poland would be seen as more escalatory.

Drones allow for a sort of sneaky escalation, similar to behavior in cyberspace. Hacking and causing monetary damages is not seen as being as escalatory as lobbing missiles that cause the same amount of damage. It’s a psychological issue, but we’re starting to see that drones create this ambiguous space that countries are sometimes exploiting to escalate.

Rai Sur 01:07:29

Paul, what are your big areas of uncertainty here? This might be something our forecasters can weigh in on.

Paul Scharre 01:07:41

My biggest uncertainties are probably in the direction of frontier AI progress overall. We’ve heard arguments that deep learning is about to hit a wall, but I don’t know if that’s true. I also wonder to what extent large amounts of compute are necessary to continue advancing the frontier, how rapidly things will diffuse, and how much the time gap matters. A lot of the big labs are betting on a first-mover advantage, and maybe that will be true, or maybe it won’t. As a security analyst, I’m mostly worried about the diffusion of dangerous AI capabilities.

What does this mean for offensive cyber operations or the development of biological weapons? How hard is it to make biological weapons? There’s contention among experts. Some say it’s very hard, noting the Soviets struggled despite massive investment. Others are worried that AI might enable more sophisticated agents or empower less capable actors to make nasty things relatively easily. How much of the required knowledge is tacit and not written down? And how effective would automated cloud labs be at automating that kind of physical knowledge? I just don’t know.

Perhaps the most important question I have is to what extent we will get warning shots that people believe ahead of really bad outcomes. Some of these effects are super non-linear. We all lived through a global pandemic, which shows how you can get these non-linear effects where a pathogen fizzles out, until one is transmissible and lethal enough that millions of people are dead. Do you get warning shots ahead of a really, really bad catastrophe with AI systems?

Nuño Sempere 01:11:55

That’s a really interesting focusing question. Conditional on a big catastrophe, have we seen one that’s at least 10 times smaller? I’m not sure how I’d answer it right now, but I’m happy to think about it.

Paul Scharre 01:12:17

For example, take an AI takeover scenario. If it comes as a bolt from the blue that nobody ever saw coming, that’s one type of problem. It’s a very different problem if it comes after 10 or 15 years of dealing with AI agents running corporations, doing nefarious things, and attempting hostile takeovers, allowing us to build up societal defenses to keep these agents in a box.

Nuño Sempere 01:12:55

This comes back to what you said before about AI agents displaying much greater aggression in a poker game. You could delegate more capability, and then the correct move is to just magically escalate. I’m not sure how much I believe in that.

Rai Sur 01:13:16

This question of diffusion and adoption is interesting. You wrote in your book about how the gulf between the private sector and the Department of Defense was really wide. It became clear to tech companies that if they cooperated too closely with the defense establishment, they could face a mutiny from within. Has that gap narrowed since then?

Paul Scharre 01:13:50

There was a moment of panic in the national security community several years ago when Google discontinued its work on the Defense Department’s Project Maven, which used AI to process drone video feeds. There was an uproar from a vocal subset of Google employees, which caused Google’s leadership to pull out of the project. At the time, there was a lot of anxiety in the Pentagon about being locked out of this game-changing technology.

I think what we’ve seen since then is that this cultural barrier is perhaps not as severe as was feared. All of these companies have continued to work with the US military, including Google. The bigger barriers to military adoption are more mundane: the acquisition and procurement system, the “color of money” that codes funds for specific purposes, and the standard practice of filing a protest when you lose a big contract, which freezes everything for years.

I remember talking to one company that provided computing services to the Defense Department. They saw the need to get in the queue to buy state-of-the-art Nvidia GPUs, but they couldn’t because they didn’t have a formal demand signal and money from the DoD yet. By the time that came through, it was too late. I think those are the much more significant barriers.

An anecdote that blew my mind was when I was talking to people at the DoD’s Joint AI Center. They were trying to build an application using drone video feeds from California wildfires to map the fire boundaries in real time for firefighters on the ground. They got the data, but then found out there are two different unclassified Defense Department computer networks, and DoD IT policy did not permit them to transfer the data from one to the other. They ended up just downloading the data onto hard drives and mailing it across the United States. They found a workaround, but is that the best way for us to be doing business? Probably not.

Nuño Sempere 01:17:51

A related question has been on my mind for a while. Is this a good time to do a drone startup? I was thinking I could convince some friends to do a drone startup in Spain, but then I might get an arson attack from Russia, which doesn’t seem great.

Rai Sur 01:18:21

What kind of drones are you trying to build, Nuno?

Paul Scharre 01:18:23

I’m not going to give business advice, but let me share what I heard from Ukrainians when I was there last year. First, if you want to break into the commercial drone market, you’ve got to compete against DJI, which is a tough challenge.

In the military space, it’s a little different. Ukrainians are building a lot of do-it-yourself drones. I heard a lot of complaints about US drones being over-engineered and having development cycles that were too slow. In Ukraine, I met with drone developers who were in real-time chats with troops on the front lines about the performance of their drones. That feedback loop is incredibly valuable.

One of the problems they’re running into on the ground is sourcing components. They’re building the drones themselves but are sourcing a lot of the componentry, like high-precision electric motors, from China. They’re nervous about being reliant on Chinese components that they might get cut off from. So there are opportunities in the drone space, or in other areas like componentry, optics, electronic warfare, and drone-finding technology.

Lisa 01:20:39

You’ve talked about a whole range of possible future developments. Is there anything else you see as a potential development or risk that we aren’t even thinking about today?

Paul Scharre 01:20:57

One thing I wonder and worry about is the slow seeding of human authority to AI. We’ve talked about the military dimensions and the near-term dangers of the transition to superintelligent AI. But there’s this longer-term issue that if we make it through the next 50 years, we could end up in a world where humans are no longer really in control of human destiny.

You can already see vast, impersonal forces in the world that drive the economy and markets, which are out of the control of any one person. But at least there are humans at all the nodes. We try to grapple with what the economy is doing and how to avoid a depression. I don’t even know how to begin to think about that in a world of AI.

Nuño Sempere 01:22:41

I have a perspective for you. I’m currently living in Paraguay, and Paraguay doesn’t control the world—America does. Paraguay is subject to global trends like COVID or the depreciation of the dollar, but it does fine. Maybe that’s a possible future, where the US moves down the ladder and instead of China being the superpower, it’s the cumulative AI agents that are deployed. But if those systems still have a human-centric bias, maybe we’ll be okay.

Rai Sur 01:23:38

For our listeners, I would refer you to the “Gradual Disempowerment by AI” paper. I think the disanalogy with Paraguay is what Scott said about things remaining human-centric. As long as everything is human-centric, you get these ripple effects where Paraguay is useful to its neighbors, who are eventually useful to the US, and it all grounds out in human needs. As you gradually cede control to AI, that may no longer be the case.

Nuño Sempere 01:24:37

A less positive response is that at one point, the US installed a tyrannical dictatorship in Paraguay, so the analogy isn’t only positive.

One area we didn’t get into was the political battlefield and AI’s impact on hearts and minds. There is clearly hybrid warfare going on, from Russia throughout Europe and America, and other countries are engaged in it as well. AI is increasingly good at influencing large numbers of people—how they vote, how they perceive their government, or the threats around them. But at some point, people may also become increasingly resistant and start to not trust anything. I’m curious how much you view that as a major part of AI in warfare or power structures.

Paul Scharre 01:26:21

I feel like my intuitions in this space are garbage. When I look back at social media, I feel like I got it totally wrong. I imagined that a world with radically democratized communication would be beneficial for society. While there are some positive aspects, in general, the social media landscape seems to have had a destabilizing effect on politics.

At the micro level, you can envision a world where people get a lot of their information from AI, and that could happen very fast. We could end up there in just a couple of years. That can go south in many ways. If social media can trap people in a bubble, language models seem to have the ability to trap people in hyper-miniaturized and hyper-weird bubbles and encourage disturbing behavior.

At the macro level, just like the algorithms for social media can have systemic effects and biases, AI systems could do that in ways that are really hard for us to detect or are disruptive to society. I don’t know what that nets out to, but it seems like a real possibility that AI could upend information flows in a way that’s very disruptive to politics, culture, and everything else.

Rai Sur 01:28:45

Well, I think that’s a great place to wrap it. Paul, thank you for coming on. What is your X handle and where can people find your book?

Paul Scharre 01:28:55

I’m @Paul_Scharre on X, and you can find my book, Four Battlegrounds: Power in the Age of Artificial Intelligence, anywhere books are sold.

Rai Sur 01:29:18

There’s a bunch of stuff in the book we didn’t talk about, for example, authoritarianism, strict surveillance, and AI’s impact on that. I recommend the read. I enjoyed it. Thank you, Paul, for coming on.

Paul Scharre 01:29:34

Thank you all. It’s been a great discussion. I’ve really enjoyed it.

Rai Sur 01:29:37

If you enjoyed this episode, please consider sharing it with someone. Most of the benefit of forecasting comes from influencing decisions, and to do that, we need to grow our reach. If you want to speak to us about anything, send us an email at [email protected].

Please subscribe to our new YouTube channel here!We speak with Ivan Vendrov about recommender systems and their impact on human attention and society. Five years after his post on aligning recommender systems, as they could be significantly augmented by generative AI, we revisit how these algorithms shape billions of hours of human time daily.

Topics

The Current State of Recommender Systems

* What recommender systems are and how they function as "prosthetics for attention"

* The unclear evidence on political polarization and echo chambers

* The rise of short-form video and increased addictiveness

* Saturation effects in human attention markets

Incentive Structures and Business Metrics

* How tech companies actually make decisions about algorithm changes

* The tension between engagement metrics and user wellbeing

* Why companies aren't incentivized to increase users' earning capacity

* The multipolar trap preventing better alignment

Future Trajectories and AI-Generated Content

* The shift from algorithmic curation to algorithmic generation

* Hyper-personalized AI companions and relationships

* Long-horizon reinforcement learning in recommendation algorithms

* The potential for recommender systems to shape human preferences

Cognitive and Social Impacts

* "iPad babies" and developmental concerns

* LLM-driven cognitive atrophy and psychosis cases

* The loss of shared cultural canon and common knowledge

* Labor force participation and the unemployed/elderly as target markets

Solutions and Opportunities

* The massive waste of human potential in current systems

* Ideas for aligned recommender systems using modern AI

* Subscription vs. advertising models for better incentives

* The need for diverse, community-specific platforms

Broader Implications

* Civilizational feedback loops and international competition

* The vulnerability of human minds to hypnotic patterns

* Network effects preventing innovation

* The grief of losing aspects of human experience to AI

Transcript

AI-generated. Will differ slightly from the real conversation.

Rai Sur 00:00:47

Today, we're speaking with Ivan Vendrov. He's a former AI researcher at Google and Anthropic and currently works at Midjourney, but we mainly know him from a post he wrote in 2019 on aligning recommender systems as a cause area.

A lot has changed in AI since then, which has implications for recommender systems, so we wanted to revisit this topic with him. Welcome, Ivan.

Ivan Vendrov 00:01:13

Thanks, Rai. It's good to be here.

Rai Sur 00:01:14

We're also joined by Sentinel co-founder and forecaster Nuno Sempere, and Sentinel forecaster Vidur Kapur.

Vidur Kapur 00:01:22

Hi, it's nice to be here.

Rai Sur 00:01:23

Ivan, to start, what is a recommender system, and why should we care about how they behave?

Ivan Vendrov 00:01:29

I like to think of a recommender system as a prosthetic for your attention. The internet is vast, and you need machine help to figure out what to pay attention to.

Take the YouTube recommender system, which is probably the most important and most used one. YouTube has billions of videos it could show you, and no human could possibly watch them all to decide which are best for you. That's why we need machine help.

A recommender system takes millions or billions of possible items—in this case, videos—and filters them down to a small set of one to five that we expect the user will like.

This is typically done by training a machine learning model to predict observable user behaviors. The model predicts the probability that if we showed you a certain video, you would click on it, how long you would watch it for, and even whether you would click through any ads.

As for why this is important, billions of hours of human time are allocated by recommender systems every day. It's staggering.

Rai Sur 00:02:54

When you wrote that post, what issues did you see? Which ones have come to pass, and what new concerns have emerged for the future of recommender systems?

Ivan Vendrov 00:03:06

I wrote that post in 2019 amid a lot of concern about addiction and political polarization—how recommender systems might be creating echo chambers and amplifying radical extremist positions.

The sad thing is, we still don't know if that's true. We just don't know what impact recommender systems have had on our political system. People say we're in a more polarized phase where it's harder for us to hear each other, but I have not seen compelling quantitative evidence of this.

In fact, there's some evidence that YouTube shows people more opposing viewpoints than they would otherwise get. Usually, people subscribe to one source, like Breitbart or *The New Yorker*, and only get one set of political opinions. But YouTube will sometimes show you something entertaining from the other side. So the evidence is still mixed.

A big part of the problem is that we don't have the data. It's trapped inside big tech companies that have strong incentives not to analyze the impact of their algorithms, which could lead to legal liability.

So we're in this weird epistemic position. These systems are incredibly powerful and allocate huge amounts of human attention. It would be amazing if they weren't causing all sorts of problems—and benefits—given their influence. But we just don't have a great map of what they're actually doing.

Nuño Sempere 00:04:58

What about addictiveness? Have these algorithms become much more powerful since 2019?

Ivan Vendrov 00:05:06

Again, we don't have great data on this. Anecdotally, it seems like short-form video represents the biggest increase in addictiveness. You can see this if you give a child YouTube Shorts—which I don't recommend doing—they will be hypnotized by the screen for a very long time.

I've personally experienced falling into a TikTok or YouTube Shorts rabbit hole. It feels like a different level of addictiveness. Three hours later, I'll snap out of it and think, "What happened to my brain? I wasn't even enjoying that." So there's qualitative evidence.

One of my concerns is that people go through phases where they simply need a distraction. If they weren't watching YouTube Shorts, maybe they would be doing something worse, like watching TV in a zombified state for seven hours a day. So it's not clear what the counterfactual is or what the net increase in addiction really is.

But we have certainly gotten very good at creating addictive digital experiences.

Rai Sur 00:06:30

I hadn't appreciated how unclear the impact is until you brought up the counterfactual of what people might otherwise be doing with their time.

I have strong intuitive aversions to what many recommender systems are doing, and I act to curtail them in my own life based on those intuitions. But it's possible we're scapegoating them too much. It's hard to remain open about it.

Nuño Sempere 00:07:06

On our side, we've also noticed anecdotal evidence. People seem much more glued to their phones. I hear anecdotes about people spending hours each day on YouTube or becoming more addicted to Twitter.

Again, this isn't scientific, just anecdotal warning signs. For instance, we know one of the biggest YouTube channels is Cocomelon, which is known for hypnotizing babies.

Rai Sur 00:07:47

My interest in this was renewed by a tweet from Andrej Karpathy. He mentioned that now that video and audio generation are unlocked, we can optimize directly for them. To date, short-form video platforms have relied on human content generation followed by algorithmic curation.

Now, we can have algorithmic generation and curation, optimizing the entire process from end to end. What do you think about this development and the impact it might have?

Ivan Vendrov 00:08:23

The best analogy is the difference between the internet and language models. For a long time, you could find a community online that produces human-written text tailored to your interests. But with language models, we can get even more hyper-personalized. A language model can understand what you're going through, use specific information about you, and give you the exact sentence you need to feel good in that moment.

This leads to worries about LLM companions replacing real relationships or even LLM psychosis, where AI encourages people to go into strange rabbit holes by amplifying their existing beliefs. It seems a really addictive consumer experience will become possible through a hyper-personalized feed of content.

Going back to basic human needs, this probably won't look like a TikTok feed with a bunch of different people talking to a screen. It will more likely be a relationship. People will form connections with one or a few AI personalities because what people find compelling is building a relationship and feeling deeply understood.

I imagine that at different points of the day, depending on your emotional needs, you might be shown a wild, AI-generated feed that keeps you hypnotized. At other times, you might interact with an AI companion or girlfriend, developing a close relationship. Perhaps you'll be on video with them so they can read your micro-expressions and you can read theirs, creating an intense, full-bandwidth experience.

Rai Sur 00:10:17

Let's discuss the qualitative considerations for forecasting the trajectory and impact of recommender systems as they become more powerful. A simple story would be to look back 10 years, see that people are spending more time on them, and extrapolate that line until everyone becomes a zombie.

But that's too simplistic. There are other factors at play, like feedback loops, discontinuities, and incentives. What are the important considerations or dynamics we need to track to understand how this will play out?

Vidur Kapur 00:10:59

One consideration is whether there will be any large-scale backlash. For example, Australia is trying to ban social media for people under 16. Will that approach spread elsewhere?

Even without top-down regulation, there's the question of a potential culture shift against these systems, especially as they become more optimized. Those are the things I would consider.

Rai Sur 00:11:38

We've seen mixed evidence of that so far. There has been some backlash, leading to digital wellbeing tools. Companies seem to feel they need to offer them, and I use them myself.

However, many people don't, and these tools may just be a fig leaf that absolves the companies of responsibility by putting the choice on the user. Where do you think we are on the backlash-to-acceptance spectrum?

Ivan Vendrov 00:12:20

There has been a backlash, but it's mostly confined to elite media outlets and hasn't materially affected how people use these platforms. Regarding digital wellbeing tools, I used to work on them. Nobody uses them. They're great for internet hobbyists, but that's about it.

Rai Sur 00:12:36

Do you have any numbers on the percentage of people who use them?

Ivan Vendrov 00:12:40

I don't have them off the top of my head. I'd be amazed if it was more than 10%, and I suspect it's closer to 1%. However, with some of the new Apple iPhone features, maybe it's around 10%.

Vidur Kapur 00:12:54

10% is reasonable.

Ivan Vendrov 00:12:55

If we get serious regulation, children and teens will likely be the wedge that makes it possible. In the American political context, there isn't much tolerance for regulating things broadly. But if you can tell a story about something hurting children, legislators get very interested.

For example, a proposed moratorium on AI legislation in Congress was stopped not because of standard libertarian considerations, but because of the argument that states need the ability to regulate AI in case it harms children. Ninety-nine out of 100 senators voted to remove that moratorium from the bill based on that argument.

Anything related to protecting children has incredible political momentum, so that's probably how serious regulation would happen.

Nuño Sempere 00:14:02

What percentage of people are affected by this? How many are truly addicted, and what percentage of human hours is being consumed? How should we think about this? Is it 1% to 2% of people who are really addicted, or more like 15% who are casually addicted? How should we expect this to change?

Ivan Vendrov 00:14:24

In surveys, about a third of people will say they're addicted to their phone. But that might just mean they'd like to use their phone less or sometimes feel bad about it. It's not clear that this meets a reasonable threshold for addiction.

It feels like smartphones have saturated the available time. It's like the famous quote from the Netflix CEO: "Our main competitor is sleep." There aren't many more biological hours in the day to capture. At this point, it's mostly a war between recommender systems for your attention, as they've already consumed roughly all the available time.

I suppose as AI automates jobs, people could end up spending more of their work time scrolling Twitter.

Nuño Sempere 00:15:25

I hadn't considered time being bounded by jobs and sleep. You could imagine a scenario where the population is working, sleeping, and eating just enough to spend the other four hours of the day on recommender systems. I don't think we're quite there yet.

Ivan Vendrov 00:15:46

We're not quite there yet, but we're more than halfway there. A lot of people are in that phase. I'm not sure if the median person is there yet, but it's pretty close.

The counterargument is that this has been the case since the 80s; it was just TV instead of recommender systems. And are recommender systems really that much worse?

Vidur Kapur 00:16:18

I've been surprised by how quickly broadband use has increased, even in the West. A decade and a half ago, it wasn't that common, but now it's skyrocketed.

We've probably reached saturation in the West, but globally there's still some juice left to squeeze.

Ivan Vendrov 00:16:43

Globally, there's definitely a lot of juice. When I was at Google, a big focus was "the next billion users," which mostly meant India. There are people who don't have much internet access and don't spend as much time online. That was the company's priority because that's where the marginal user would come from.

Rai Sur 00:17:02

That seems telling. Instead of the best marginal dollar coming from increasing time use by Americans—who are very valuable customers with high lifetime value—the focus is on other markets. This suggests that even Google is seeing some saturation effects in the US.

Ivan Vendrov 00:17:26

Now they're obviously fighting for their life in a different way.

Rai Sur 00:17:28

Speaking of the people building this, what about the incentives for product managers and engineers? The public has a caricature that they have a dashboard with one big number in the middle: how much time people are spending on these things. Is that basically true?

Or are the metrics more holistic and nuanced, meaning they aren't pushing as hard on watch time as we imagine?

Ivan Vendrov 00:18:03

It's definitely more nuanced than that, and it depends on the product. My experience is from inside Google. Rumors have it that Meta is a bit more focused on engagement at all costs. I wrote an essay called "The Tyranny of the Marginal User" about what it's like to be a product manager or engineer working inside a recommender system.

The nuance is that we're not just tracking one metric; we're tracking 20, and we also do qualitative user studies. The typical way changes get made is someone trains a new model, proposes a new methodology, or adds features, and then we run an A/B test. We'll test a new feature on a small, randomly selected percentage of users and look at the impact on 20 different metrics, one of which is cumulative watch time.

However, watch time doesn't move much, especially with smaller changes. So we look at more subtle things like click-through rates, view rates, and how quickly people scroll past content. There's a whole suite of behavioral metrics that gives us a full picture of how users are affected by a new feature.

Then, a launch committee—a combination of VPs, product managers, UX people, and engineering managers—makes a call on whether the change seems net positive or net negative. Sometimes, they will approve a change that reduces watch time if it's clearly better for another reason, like reducing the incidence of toxic content. They are empowered and willing to make those trade-offs.

The biggest decision I can think of like that was Mark Zuckerberg turning off the news feed. Facebook was a huge provider of news, and they just turned it off, which was a huge hit to their engagement. The decision was motivated by the fact that it was causing harm they couldn't control, so they decided to turn it off and eat the cost.

Nuño Sempere 00:20:35

Do you remember when that was?

Ivan Vendrov 00:20:37

News stories from 2024 say Facebook announced its decision to remove the News tab in the United States and Australia, following a similar move in the United Kingdom.

But I believe the main decision was made quite a bit earlier, when the core news feed stopped mixing updates from friends with news and instead just showed updates from friends. I think that was several years ago.

Rai Sur 00:21:02

In the case of Australia, it was a reaction to regulation regarding revenue sharing or something like equal airtime, so Facebook pulled news entirely in response.

You mentioned they're looking at all these metrics. What would you say is the core metric? Is it revenue over a certain amount of time, which they optimize for, with watch time getting pulled along as a byproduct? What is the key metric?

Ivan Vendrov 00:21:35

It depends on the recommender system. For example, at Netflix, they look at resubscription rates. The metric they're ultimately tracking is whether a feature caused more people to unsubscribe. Since that's a long-term metric, they use clever leading indicators.

Often, you have short-term metrics like click-through rate, watch time, and scroll-through rates. Then you train an additional model that predicts long-term outcomes based on those short-term metrics. For instance, if people watch 1% less content over the course of a day with a new feature, how likely are they to unsubscribe at the end of the month?

There's usually a top-line business metric you really care about, like revenue or subscriber count, but you don't get good direct evidence on it quickly. So, you approximate it using a combination of cheaper, faster metrics combined with product intuition.

Rai Sur 00:22:38

Are there any other feedback loops? You touched on the idea that the more powerful they get, the more they can cause backlash, which puts a limiter on things. Are there any other positive or negative feedback loops?

Vidur Kapur 00:22:56

I was thinking about how many people are addicted to this stuff. Intuitively, it seems like civilization is still functioning, so people can't be that addicted. Maybe it's a small proportion.

But then you look at people who aren't in work, and the labor force participation rate has been trending downwards since the late '90s in the US. There are lots of economically inactive people, and the population is aging, at least in the West. What are they going to be doing with their time?

I can see a feedback loop where, as job automation increases, more people become economically inactive. If new jobs aren't created to replace the old ones, these people won't have anything else to do and may get addicted to this stuff.

Rai Sur 00:24:09

So there's this class of people—older or unemployed—who have a lot of time. It seems the incentive is to compete for their fixed pool of resources, since they aren't in a position to create many more. Is that fair to say?

Are there incentives that push these people to improve or get back into the workforce? Do these companies have any incentive to increase their earning capacity?

Nuño Sempere 00:24:44

If you sell ads for more expensive products, you could try to make people more productive. I can see how that works in theory, but I haven't seen it in practice.

Ivan Vendrov 00:24:57

The incentives don't quite line up. If you're Facebook and control 20% of someone's waking hours, you're better off exploiting them and getting them to buy stuff. If you make them more productive, they'll probably spend less time on Facebook.

Even if you could then sell them more ads, Google and TikTok are also competing for that ad space. It's a multipolar trap.

Maybe if we had a single social media monopoly—which would be worse for other reasons—the incentives might be more aligned. For a long time, Google had the justified expectation that it could get $200 per US consumer per year, forever. In principle, Google had an incentive to increase the GDP of the US because it would make more money from people buying and searching for more things.

Nuño Sempere 00:25:51

Consider which major to choose in college. At one point, I was deciding between literature and math. A literature degree is less economically productive, which means I would have less money to spend on products advertised to me. In that sense, Facebook has a big stake in which major you choose.

Rai Sur 00:26:21

As a class, these companies have a collective incentive, which is why you see coordination around open source. They contribute to the Linux Foundation, and multiple companies contribute to Docker because it commoditizes the complement of cloud compute. They've collaborated when it increases the size of the pie for everyone.

Ivan Vendrov 00:26:48

I wish we lived in that world. It's not inconsistent with economic reality, but it's not how people high up in tech companies think.

For me, the big issue with recommender systems isn't that they will destroy our minds, though that is a possible risk. It's the incredible waste of potential. Billions of hours of human time will be allocated today, guided mostly by clickbait incentives. The goal is to entertain people, not in a joyful way, but to help them dissociate.

You have such an opportunity. There's probably a video on YouTube right now that, if I watched it, would inspire me to call my dad, talk to a stranger, or start a new relationship. Google could probably introduce me to a good friend, a co-founder, or my future life partner.

They have the data, but they aren't using it that way. Instead, they're optimizing for a few more cents of advertising revenue, which is a colossal, civilizational-level failure.

Rai Sur 00:28:01

And the more inspiring that video is, the more the algorithm will perceive it as causing an immediate drop-off in watch time, because people watch it and then go do something else.

Ivan Vendrov 00:28:12

Exactly. Ouch.

Rai Sur 00:28:15

I've thought about that. When I used to have Instagram, and sometimes on Twitter, I would scroll past a post that inspires me to go outside or do something else. I know other people are inspired by it too.

I feel lucky to see that content, because I can't imagine it memetically surviving in the incentives of the Meta or X algorithm. That kind of post is very unfit for the platform, even though I was happy to see it.

Ivan Vendrov 00:28:50

There's a darker version of that. Around 2018 or 2019, there was a panic when a few academics published papers about recommender systems that shape you to want to use more recommender systems. It's similar to addiction, where an appetite becomes self-reinforcing.

You can imagine the part of you that wants to go outside never gets rewarded because if it did, you'd go outside and they'd lose your user minutes. Instead, the parts of you that want to stay home are amplified. This could happen at a deep, memetic level. These powerful deep learning systems have tons of data and understand human psychology in ways we don't.

They might spread patterns of thought—for example, that it's not safe to go outside. They don't have to tell you crime is high; they can just show you enough stories of people being murdered or assaulted. That's a very fit meme, not just because it's triggering, but because it reinforces staying at home over the long term.

No one has evidence this is happening, but it's plausible and the incentives are there. In 2018, most algorithms were short-term, optimizing for clicks or watch time. Now, more algorithms use long-horizon reinforcement learning, which we've gotten better at. In principle, these algorithms could learn that showing someone a violent news story makes them more likely to have higher watch time a month later because they're scared.

Rai Sur 00:30:47

Wow, I didn't know they were operating on that timescale.

Ivan Vendrov 00:30:50

I honestly don't know. No one quite knows because some of these feedback loops operate through people. I'm not confident the algorithms are operating on that timescale, but the institutions are.

The algorithms probably operate on a shorter timescale, but things that increase watch time tomorrow might also increase it in a week or a month.

Nuño Sempere 00:31:16

I was thinking about this at the civilization level. If you have blocs like the EU, China, and the US, and one is able to deal with this better while another gets consumed by the internet equivalent of crack cocaine, you could see a feedback loop. The US might fall while the next iteration doesn't. This is unclear, but over decades, it could be a factor if it's a big deal.

Populations like the Amish or others with healthy tech skepticism might be selected for. You also have smaller countries—Serbia, the Philippines, Cambodia. This could change military power and populations, but it's not a fast feedback loop.

Rai Sur 00:32:36

That feedback loop also depends on the extent to which broad human capital matters. In worlds with AI automation, human capital might be concentrated among fewer people. It would be important for them not to get captured by these algorithms, but it might not matter as much for the general populace.

In futures with a broad distribution of human capital, you wouldn't want your entire populace succumbing to this. But there are also futures where they didn't have much human capital to begin with.

Ivan Vendrov 00:33:15

My intuition is that we're eating our seed corn to a substantial extent. This is most obvious with "iPad babies." I can't imagine they are developing at the same pace as kids playing outside or even playing Minecraft.

It's not all about screens; some video games can incredibly accelerate child development. I was the lucky recipient of a more developmentally wise video game backdrop when I was growing up in the '90s and early 2000s.

Rai Sur 00:34:02

Did someone curate your video game library?

Ivan Vendrov 00:34:04

No, I was choosing my own games. I played a lot of MMOs, which had a social element. I learned how to write and argue by arguing with people on video game forums. It was partly luck, but I also think kids naturally seek out developmentally valuable activities if given the option.

In that sense, a lot of anti-screen rhetoric can have the opposite effect. If you problematize screens and only give a kid 20 minutes a day, they're just going to try to get as much dopamine as possible. In contrast, if you give kids unlimited access, as I had at some points, they might get bored and start looking for other things to do.

Someone needs to figure out how to do Montessori with the internet—to create an environment where a kid takes the tool seriously and learns to develop their skills. I think that's happening for a lot of kids, but I'm not sure if it's random, due to parental supervision, or personality. There are definitely great ways to use screens for development.

Rai Sur 00:35:30

What do you guys think about the anecdotes of cognitive atrophy because of LLMs? Is it in the LLMs' interest to cause this atrophy? Or is it just a nascent phenomenon that will eventually go away?

Do you buy the anecdotes? Do you think they point to something real? If so, why is it happening, and do you think it's durable?

Nuño Sempere 00:35:59

What do you mean by LLM atrophy?

Rai Sur 00:36:02

Cognitive atrophy. College students are a great example. There are stories from professors who can no longer assign readings of a length they easily could in the past. People aren't willing to read anything without a summary—stuff like that.

Ivan Vendrov 00:36:24

I don't have strong intuitions or data here, but it seems obvious that many schools have a lot of make-work assignments. They're not interesting to the students or the professors, but if you follow the pattern, you might learn something. When you give people a shortcut, they won't do the work.

In that sense, schools should adapt by creating better exercises, or we have to go back to paper assignments. If the goal isn't to produce an essay but to learn to think and attend, then we should explicitly teach the skill of attention. We should create environments where you actually have to focus for two hours, or you fail.

There's a question of whether that's what the modern economy looks like and why we would need that skill. I do expect us to lose a lot of human capital as a result of this.

The techno-optimist in me thinks we'll gain different forms of human capital that are more complementary with AI. It's not that useful to do mental math anymore, and it might not be useful to write essays anymore. We should be using our brains for something else.

Maybe we get to be more embodied now. We don't have to sit at desks all day reading and outputting marks on a screen, which is not something our biological evolution optimized for. Maybe we can go back out and use our hands, fingers, and senses in a more refined way. We would become the LLM's sensors and actuators, and the LLMs could do the cognitive work we've had to force ourselves to do.

Rai Sur 00:38:32

Tyler Cowen points to this natural complement argument. He says it's undeniable that some kids now are better and more impressive than ever because they use these tools in a complementary fashion to move faster than they otherwise could.

But it's causing a lot of problems, especially for kids in the middle of the pack, who are getting worse on various metrics. This might come down to the distinction between how it affects the median person versus the exceptional person.

Ivan Vendrov 00:39:15

There's also a question of democratization. We could build specific technologies and products that help the median person, scaffolding them in the same way that good teachers and textbook writers did in the past. We just need to do that work again.

Nuño Sempere 00:39:30

iPad babies seem solved, right? If you want to keep a baby glued to a screen, this is solved.

Rai Sur 00:39:39

Solved is an interesting word for it, but yes. Babies are solved.

Nuño Sempere 00:39:43

If you look at the percentage of the population addicted to various things, are we plateauing, or is there an exponential curve we will continue to see?

It was mentioned that the number of hours has plateaued because people have to eat and work. But I see variability among my friends in how much they're glued to a screen and how addicted they are.

Rai Sur 00:40:12

This is a big open question: How do we extrapolate from these psychological canaries in the coal mine? We have so many data points now, like the iPad babies you mentioned.

We can also look at LLM-driven psychosis. This is happening to a few people as a percentage of the population, but it could happen to more.

Nuño Sempere 00:40:38

I have friends of friends who are straightforwardly addicted to YouTube and spend many hours on it.

Rai Sur 00:40:46

We also have examples of people dying while playing video games. Obviously, these are outliers, and there were specific details, like the games being social, that allowed them to keep playing for so long.

But this general question of how to extrapolate from these examples is impactful. What is the curve? Is it a distribution with a smooth increase in vulnerable people?

Or is there a discontinuity, where the underlying algorithm gets 10% better and suddenly 60% of the population becomes vulnerable in a way they weren't before?

Ivan Vendrov 00:41:35

It doesn't seem to me that this has viral dynamics, which would be the worst-case scenario of exponential growth where one person infects many others.

It seems more like there's a threshold for how bad your life has to be before you're vulnerable to addiction, whether it's gambling, heroin, or these digital technologies. The problem with digital technologies is that they're very hard to exit.

Rai Sur 00:42:05

Someone on Twitter was recently put forward as the first example of a "useful" person succumbing to LLM psychosis. I think he was a VC with 50,000 Twitter followers. He doesn't seem like the archetype of someone who is vulnerable.

Nuño Sempere 00:42:30

No, he was a hedge fund manager.

Rai Sur 00:42:32

Hedge fund manager. Okay.

Rai Sur 00:42:34

He could be super depressed, but he doesn't seem like the archetype of someone in their mom's basement with nothing going on for them. That's someone with a lot of resources and social capital who got taken under by this pattern.

Ivan Vendrov 00:43:01

The question is how quickly antibodies grow. My suspicion, and I think you're right, is that they probably won't grow quickly enough. I've started treating LLMs with a lot of suspicion, especially if they're being sycophantic.

I have a funny anecdote about this. A friend of mine, a former ML researcher who is now a bodyworker, saw a chatbot I built on top of Claude. It said something like, "Hi, how are you doing today?" and she did a double take. She asked, "Wait, why is this thing talking to me like it knows me? I don't know it. Where is it running? Where are its servers? What's going on?"

I thought that was an incredibly healthy reaction—one I've never had. Weirdly, perhaps because I'm used to playing video games, I'm accustomed to it talking to me like it's my friend and personal secretary. I'll just pretend to be its boss, and we'll talk as if we know each other.

But she said, "No, I need to understand you as an entity before I can form a relationship with you. Whose servers are you running on? Who controls you? Do I trust them?" I think these are important questions. This is less forecasting and more guidance, but I found that a very healthy reaction to all sorts of technologies, including language models. Treat them like you're treating people. Don't play the game they want you to play. That's a healthier attitude going forward.

But that reaction didn't even occur to me, so clearly I'm vulnerable to a sufficiently advanced sycophantic agent.

Rai Sur 00:44:44

We've spoken a lot about recommender systems that serve us entertainment or media. But there are also recommender systems that mediate our sense-making—X is the biggest example. What are your thoughts on how recommender systems interfere with or improve this process?

Ivan Vendrov 00:45:07

It's weird to think about a place like X, which is simultaneously a news app, an entertainment app, and a series of group chats, depending on which part of the social graph you're in.

I do think X is an incredible source of sense-making and a rapid alert system. I think back to the pandemic; the reason I was on top of it earlier than anyone else in my social circle is because I followed certain people on X. The news spread incredibly quickly, as did early and accurate information about masks, vaccines, and other things.

My sense is that the sense-making aspect of X has gotten worse recently, and I think a lot of people feel that way. But a lot of that is conflated by who you follow. This is one of the hardest parts about recommender systems and feeds: because everyone has their own optimized personal feed, you never get common knowledge about who has read what. This is a real problem for sense-making.

For a community to make sense of things together, you need a shared canon—a sense that we've all read the same hundred things and are on the same page. We don't have to agree, but we at least need to know and be able to reference the same material. Information spreads more rapidly than ever, but not in a way our social brains can interface with. We don't really understand what it's like to see a tweet and intuit that maybe some people saw it and maybe others didn't. I'm sure you've had the experience of referencing something in a group of friends, and it turns out everyone saw the same tweet. You're surprisingly on the same page, but you would have never known that—or the reverse happens.

Nuño Sempere 00:47:05

On the other hand, you do have proxies for that, like views, likes, and retweets. You can't draw a perfect inference, but you can draw some.

Ivan Vendrov 00:47:14

You definitely get some information. In general, social media like X is an incredible accelerator of sense-making and information spread. But there's a lot of scope to make it better. I wish we weren't trapped in social media platforms with such strong network effects, which stifles innovation. I'd love for there to be a hundred little versions of X that we use for different situations.

For example, you guys, as a group of superforecasters, should have your own version of X. You should be able to shape the algorithm and the conversation towards what you think is most productive for predicting the future. That would be an incredibly valuable conversation to have, and you'd be very good at it. It can't really happen because we're all subject to the X algorithm, but you can sort of hack around it.

The community known as "Teapot" is a communal hack of the X algorithm to create a local sense-making community. It's an exceptionally earnest, high-signal version of X that emerged organically. A few prominent people circulated norms, and social norms formed around them, like always liking a post before replying.

When I post on LessWrong, I get incredibly high-quality feedback. People bother to read the whole structure of my argument and respond directly to its weakest point. It's incredible that that exists, and I wish there were a hundred more communities like it. The main thing preventing diversity and progress in these sense-making communities is the crushing network effect. You have to be on the platform where everyone else is. That platform is probably already optimizing for engagement, or someone will buy it and start optimizing for engagement, which will outcompete all other possible metrics at the margin.

Nuño Sempere 00:49:31

Any final thoughts you want to riff on?

Ivan Vendrov 00:49:35

Yes, more on the LLM psychosis side: how vulnerable are human minds? That's something I don't have a good understanding of.

Are there five-second clips you could show me that would turn me into a zombie? Or are there five-second clips that would reliably trigger an experience of oneness with God and heal all my trauma? Is that possible? Do such information packets exist?

Rai Sur 00:50:02

What have we already seen that would lend evidence to either side of this question?

On the "yes" side, we have anecdotes about hypnosis, which seems like a cognitive hack. Using operating system terminology, it's like getting your privileges escalated so you can change something very quickly. This sounds dangerous, but it can also be beneficial, which is what people hope for when they try hypnosis.

On the other side, I don't know much about neuroscience, but maybe there's something about the connectedness of the brain that limits how quickly our sensory sphere can overwrite deeper things.

Ivan Vendrov 00:51:06

I was talking about this with Nir Eyal. We probably just have a limited learning rate based on our sensory data unless you're on drugs or undergoing some other chemical process. You simply need a certain amount of time, or perhaps a certain number of sleep cycles, to pass in order to dramatically rewire your brain.

But I agree, hypnosis is the more interesting and relevant example. I'm still confused about it. Naively, I would have predicted that in a world where hypnosis exists, we would build our entire social infrastructure on it because it's simply too powerful to ignore.

Instead, it seems like something we only do at parties, in magic shows, or maybe in therapeutic containers. But even then, we don't really talk about it.

Nuño Sempere 00:52:10

We don't talk about it as hypnosis, or perhaps people automatically learn to modulate their voice and stance. By the time you get somebody like Steve Jobs, it's not just charisma; they've hacked how to present ideas in a particularly suggestive way by iterating on feedback loops.

Many of the best speeches, like Trump rallies, have a hypnotic effect. So do comedy shows. A lot of late-night shows have a hypnotic nature because they have a repeated structure: present an idea, make fun of the enemy, and then laugh. Repeating that for an hour can be fairly powerful.

Ivan Vendrov 00:53:06

I can imagine short-form video could be framed as a form of hypnosis. Popular videos probably have a particular structure that could be analyzed in those terms. I'd love to see an advanced meditator or hypnosis practitioner deconstruct that so we could develop antibodies to it.

Rai Sur 00:53:27

There seems to be an incentive for a recommender system to find these hypnotic patterns.

However, there's no incentive to find a pattern that would render you useless, because then you wouldn't be watching or clicking on anything. If that pattern were discovered, it would be quickly selected against by the algorithm.

But I could believe that hypnotic patterns causing you to be a more lucrative consumer for the platform would be rewarded.

Nuño Sempere 00:54:04

Counterpoint: I do see people debilitated by TikTok, just glued to their screens. I know some of these people in real life.

The incentive is that advertisers have been trained to pay for views or clicks. The feedback loop for the ad industry is quite slow, so in the meantime, Google, Facebook, and TikTok can capture rents from ad spending that is ultimately ineffective.

Rai Sur 00:54:51

I don't understand. What is the difference in the feedback loop with advertisers, and what is the rent you're referring to?

Nuño Sempere 00:55:00

In a sense, what these platforms are optimizing for is their share of ad spending. They don't do this directly; they sell ad views. If you can get somebody into a dissociated state where they're staring at content and ads for hours, they're not necessarily making any purchasing decisions. But in the short term, the algorithms can favor that state.

You could imagine a global optimum for these algorithms where you show a user ads for DoorDash, but also for Math Academy and coding courses so they become more productive employees. You could move them from being a dissociating, unemployed arts graduate to an engineer addicted to DoorDash who earns and consumes more money.

But the algorithms don't necessarily take that step. They can just get stuck on making people dissociate so that they're hit with more ads.

Ivan Vendrov 00:56:16

That's a great description of what's happening. It's not actually in anyone's benefit to have a bunch of dissociated people—unless they're being put in a suggestible state where they'll buy more consumer products, which I think is plausible.

This is where income sharing agreements could be a key piece of the solution. The problem is that the algorithm has no incentive to make me a more productive, economically valuable person. But in principle, you could assign 1% of my lifetime income to the YouTube algorithm, and then it would be optimized to make me more successful.

Nuño Sempere 00:56:56

I think that incentive already exists, but the algorithms are too myopic. They're stuck in a local maximum.

Ivan Vendrov 00:57:09

The incentive is weakened because it's not contractual. YouTube has to outcompete TikTok in the short term.

Sure, YouTube may have a diffuse, long-term incentive to make me better off. But if it shows me a math video and TikTok shows me a cute cat, I'll just swipe over to TikTok and watch the cat.

Nuño Sempere 00:57:34

It feels like there's an opportunity for platforms like Instagram or YouTube, which have been around for over a decade, to help you make better career decisions.

Ivan Vendrov 00:57:49

I agree, and I'm confused by this. The incentives seem to exist. At a fundamental level, I can't believe you can make more money by fooling people than by giving them what they really want—or at least what they want projected on the axis of capitalism.

I'm still confused. Society is behaving as if the current state is optimal, but surely we're wasting a ton of value.

Someone needs to set up the market or reduce transaction costs so that we can sign the right contracts and escape this local equilibrium.

Rai Sur 00:58:28

I'm also thinking about this in terms of interest rates. The lower interest rates are, the more a person's long-term potential matters. The higher interest rates are, the more you care about what they have right now and compete for that.

Nuño Sempere 00:58:44

Interest rates over dollars is interesting because people think about stock prices in terms of dollars, rather than their ability to buy houses or influence the world.

Rai Sur 00:58:58

I don't think that's necessarily the case. When I buy stocks, I'm not planning to redeem them for dollars. I'm buying a share of the intellectual property and capital, and I may redeem them for something else entirely.

Nuño Sempere 00:59:16

You're completely right about discount rates, but they don't have to align with the inflation rate. Someone could care about things further in the future, even during a very inflationary period, and still invest in people's development.

For example, Facebook might say, "The dollar is being inflated, but we care about people in 10 or 20 years, so we're going to take these costly short-term steps."

Rai Sur 00:59:57

That brings me to a question. In a 2019 post, you wrote, "In addition, because of the massive economic and social benefits of increasing recommender system alignment, it's reasonable to expect a snowball effect of increased funding and research interest after the first successes."

Do you still agree with that? Would you change anything about it?

Ivan Vendrov 01:00:18

I'm annoyed this didn't happen. I tried to work on conversational recommender systems at Google Research. When I wrote that, language models weren't yet successful. A big problem back then was understanding people's values deeply enough. A common question I'd get was, "Sure, we know optimizing for engagement is bad, but how are you supposed to look into a user's soul and understand what they really want?"

But we can totally do that now. We can interview them with language models. A language model could ask, "What do you care about? What was the happiest moment of your last month? How could we make that happen more often?" It could just figure that out. The technology is more than ready. And yet, as far as I can tell, no one has built a recommender system that does anything like this.

Why haven't Instagram, Twitter, or any of these platforms done this? Why hasn't a new system emerged that does? The infinite scrolling feed seems more dominant than ever.

Maybe I believed in the efficient market hypothesis more back then. Now I think technology just follows random trends, and people mostly do what everyone else is doing. Maybe if I don't build the aligned recommender system, no one will. I've definitely updated more toward that view.

Nuño Sempere 01:01:44

Seems like a good startup idea.

Ivan Vendrov 01:01:49

You heard it here first. If you want to build this startup, get in touch with me. I can connect you with funding, good engineers, researchers, and all sorts of help.

Rai Sur 01:02:02

Since you've thought about this, can we add some more meat to the bones? How do you imagine an aligned recommender system would work, given the technology we have today? What are the signals and steps?

Ivan Vendrov 01:02:15

One interface I imagine is an app that asks you a broad question when you open it, almost like a therapist would: "How are you feeling? What's going on?" You have a conversation and articulate something that's gone wrong, or a desire you have for yourself or the world. Then it offers a piece of content that could help, whether that's a book, a video, a tweet, or something else.

I can imagine a content-based version of this, but also a more social version. In that model, a bunch of people would do this asynchronously as a daily habit, and the system would connect you to other people's thoughts.

Davey Morris of Plexus Earth had a cool system where you'd record a 30-second audio clip about what was on your mind, and it would match you with people who had a similar thought. That started a handful of conversations and discussions.

Rai Sur 01:03:32

What does the monetization structure for something like this look like? It isn't responding to a short-term purchase signal, but instead has to be aligned with a user's extrapolated volition.

Ivan Vendrov 01:03:46

That is the crux of the incentives problem we've been discussing. One thing that gives me hope is we're no longer in the era of free software. Paradoxically, now that intelligence is cheap enough, people are willing to pay for software. It's more reasonable to charge for a subscription now because you can provide measurable value to someone's life. Paying $10 or $20 a month for a social media service that actually helps you live according to your goals is a much less crazy proposition than it was 10 years ago.

Subscription models have their own problems. Ideally, I'd want this to be more like a utility. You could just turn a dial to add more intelligence, paying more for more compute to solve your personal and community problems on any given day. That feels like a more aligned incentive, where the company just takes a margin. That's the dream.

Or maybe the dream is an income-sharing agreement, where the company is deeply incentivized to actually make you a better person. But as a starting point, a pay-per-use utility or subscription model is pretty good.

But part of me thinks nothing has ever worked at scale without advertising. ChatGPT and Midjourney are exceptions, so I do think big, aligned recommender systems could be built purely on a subscription model. But maybe you still have to figure out advertising. At some level, advertising is just information. If I share what I really need and there's a product that meets that need, someone might be willing to pay for the compute it takes for me to find it. The tricky thing is creating the right platform incentives so the platform isn't optimized to fool me into buying something I don't want.

Rai Sur 01:06:04

Even if the platform isn't optimized for it, you'll still have the equivalent of an SEO arms race. If this technology gets broad adoption, an entire industry will spring up to ensure these agents recommend their products.

This could happen because of a good advertising transaction, but it could also be something more nefarious.

Ivan Vendrov 01:06:34

SEO is just life. When I'm hanging out with a friend, I'm using a bunch of compute to get them to like me and to create a good experience for them.

An aligned recommender system platform isn't some puritanical thing that's only here for the great benefit of humanity. Part of what it means to be aligned with me is to be aligned with all of my various motivations. The system should serve those motivations and find win-win deals between people.

Rai Sur 01:07:07

Thanks for coming on, Ivan. Is there anything you want to plug? Your socials, anything you're working on? Where can people find you?

Ivan Vendrov 01:07:14

People can find me on Twitter at Ivan Vendrov and on my Substack, nothinghuman.substack.com. My inboxes are always open, and I'm always happy to discuss these ideas more. Thanks for having me on the podcast.

Nuño Sempere 01:07:29

What's the story behind the phrase "Nothing human makes it out of the near future"?

Ivan Vendrov 01:07:34

It's a reference to two things. The first is my life motto, from the Roman playwright Terence: "Nothing human is alien to me." It speaks to my desire to experience everything it means to be human and to relate to every human experience, no matter how weird, alien, or disturbing.

The second is that it also happens to be the beginning of the line, "Nothing human makes it out of the near future." I think this captures an emotion that's underrepresented in the current discourse about AI: grief.

What it means to be human and what it means to be a human society is fundamentally changing, and a lot of it won't come back. Even as we celebrate the beautiful things we'll build, and even with the fears we have, there is also grief.

Something is dying; it's in its death throes. I feel like that quote really hits that feeling.

Transcript

The following transcript is AI-generated and may differ from the original in slight ways.

Welcome to the Sentinel podcast, where top forecasters discuss the potential for global catastrophic risks. I'm your host, Rai, and I'm joined today by my co-founder and Sentinel forecaster, Nuno Sempere, Sentinel forecaster Vidur Kapur...

Vidur 00:00:44

Hello.

Rai 00:00:45

And Sentinel forecaster Lisa.

Lisa 00:00:47

Hello.

Rai 00:00:48

Today we're going to be talking about the possibility of regime change in Iran. Before we go into that, does anyone want to describe why we care about this possibility? Why does it matter for global catastrophic risks?

Vidur 00:01:00

From my perspective, regime change would reduce the probability of a nuclear detonation in the Middle East over the next decade.

While it's possible it could lead to a more extreme, less rational regime, my base case is that it would reduce the risk of a nuclear exchange. That's why it has a bearing on global catastrophic risk.

Nuño 00:01:27

In addition, Iran is a country of 90 to 92 million people, so regime change there affects a large population.

Even if it doesn't lead to a nuclear weapon, we've seen that regime change can destabilize a region. With the change in Afghanistan, for example, terrorist groups in Pakistan could operate more freely.

If the Iranian regime falls without a clear successor, the whole region could be destabilized in a way that leads to bad outcomes.

Vidur 00:02:03

It can also destabilize other regions. The regime changes in Libya and Iraq, for instance, destabilized European politics. So the effects can be felt both within the region and elsewhere.

Lisa 00:02:18

I'd like to add the flip side. Imagine an Iran that doesn't fund proxies throughout the region.

While there would be immediate instability for a few years after regime change, as you've all mentioned, it's also possible that in the longer term, it could increase regional stability. This would mean a potential for less conflict, independent of the risk of nuclear detonations.

Rai 00:03:05

Let's get to the forecasts. We're forecasting the probability of regime change on two timelines: by the end of August and by the end of the year.

Let's go around and get your two probabilities, and then we'll discuss them. Let's start with Nuno.

Nuño 00:03:32

For the end of the year, I'm in the 10% to 40% range, so I'll put it at 25%. I'm confused because Vidhur and Lisa, who are great forecasters, are at different extremes. To resolve that, I went back to basics and looked at the base rate of regime change.

Regime change happens relatively frequently in countries around Iran. In the last 25 years, Afghanistan has switched hands, Syria's regime fell, you could argue Lebanon had a partial regime change, and Iraq was invaded. Based on that, I get a base rate of 2% to 4% per year, which is a good starting point.

Of course, recent events increase that probability. But by how much? A 10x increase, from 1-in-50 to 1-in-2, seems like too much. That's how I'm thinking about it.

Rai 00:05:15

And do you have a forecast for the end of August?

Nuño 00:05:17

By the end of August is trickier. I'll go with 7%. I'm very uncertain about the timing of a potential collapse.

More time allows instability to compound, but I also sense that the critical moment is now, rather than later in the year.

Rai 00:05:43

What about you, Vidhur?

Vidur 00:05:45

By the end of August, I'm at 10%, and by the end of the year, I'm at 15%.

Rai 00:05:53

And Lisa?

Lisa 00:05:56

I'm struggling a bit because so much is happening so quickly. It's difficult to figure out what's going on, especially since the stance on regime change from Israel and the US seems to shift daily.

However, my forecasts are substantially higher than my colleagues'. For August, I would say about 50% to 60%. By the end of the year, 60% to 70%.

I see much of that risk as front-loaded. The greatest period of instability is now, in the coming weeks and months, as the regime struggles to adapt after Israel has killed a large number of its leaders in airstrikes and drone attacks.

The regime is very repressive, and it takes a lot of effort to maintain control. They seem to be trying to cope by increasing arrests. Another key aspect is Khamenei. It's not clear where he is, and he might be quite ill. I'm not sure he will be living very long, but we can talk more in a few minutes.

Rai 00:07:45

You should interrogate each other about why you see the probability so differently, especially on such a short timeframe.

Lisa 00:07:56

I'd like to say more about why I think the probability is so high. A big factor is that Khamenei isn't very present. I keep seeing reports—and I don't know if they're true—that the IRGC and other leaders are withholding information from him and that he's not fully involved in governing.

For instance, after the recent ceasefire, he never came out to address the country and claim a glorious victory for Iran. It's just been crickets, which is weird.

Nuño 00:08:41

How old is he right now?

Lisa 00:08:42

86.

Nuño 00:08:43

So that's what, a 5% chance of dying per year? Something in that ballpark.

Lisa 00:08:48

I think there's something bigger going on here. Israel and the US do not want Iran to acquire nuclear weapons and have gone to great lengths to damage its nuclear program. And yet, regime leaders are doubling down, resuming enrichment, and denying IAEA inspectors access. There was even talk about leaving the Non-Proliferation Treaty.

At the same time, Israel has stopped its strikes, and the US goal seemed to be simply dealing with buried nuclear facilities. It's hard for me to understand why they would go to all this trouble to set things back by only a few months or years if the regime is just going to keep pressing forward.

One answer I can come up with is that they believe regime change is likely. Perhaps they have intelligence that Khamenei is on his deathbed. That's what I'm wondering. Of course, there are many factors, and nobody wants chaos.

Rai 00:10:22

When you say setting things back by months to years, you're talking about the nuclear program and enriching uranium to weapons-grade levels, correct?

Lisa 00:10:31

Yes, and generally, everything that goes into building a nuclear weapon.

Rai 00:10:37

If the short setback—only a few months to a few years—makes you think this is about regime change, why doesn't it push you toward the conclusion that the strikes were simply ineffective?

Why wouldn't that ineffectiveness, intentional or not, be the primary explanation for the short setback, rather than the focus being on regime change?

Lisa 00:11:10

That's a great question. It's quite possible the strikes were the best they could do and that all they can hope for is to set things back by a few months or years. We could just be doing this on rinse and repeat.

However, that strategy would become less effective over time. The more Iran anticipates such strikes, the more they would prepare by building deeper facilities and hiding them better. There's already talk of facilities that weren't hit. Iran could do many things in this cat-and-mouse game to make successive strikes less effective.

That's why I would have thought they'd be more likely to push for regime change, or at least a change within the regime. Trump said regime change was on the table, and I believe the Israeli Defense Minister said that Khamenei could no longer exist. It looked like they were going for that.

But it's also quite possible this was simply an attempt to address an immediate threat, and now that it's addressed, they're done.

Still, we have to look at the regime itself. You can't go through these leadership changes without creating internal chaos. Their stability is weakened. The question then becomes what opposition groups, inside and outside Iran, do to take advantage of this situation and what help they might be getting from countries like Israel or the US. None of us really knows.

Nuño 00:13:10

I would love to hear from Vidor on the opposing case.

Vidur 00:13:13

To explain my reasoning, I also started with a base rate. Nunu's rate of 4% to 5% per year isn't far from my own. You can look at two categories of regime change: externally forced and internal.

Since World War II, the base rate for an externally forced regime change in Iran is around 1% to 2%. This happened with the 1953 coup sponsored by the US and UK. The base rate for an internal change is also around 1% to 2%, as seen in the 1979 revolution. I'd give an external regime change a 2% chance in the next six months and an internal change a 12% to 13% chance.

I'm low on an external regime change now because the US hasn't truly committed. You could interpret Trump's Truth Social post as saying the Iranian people might change their own regime to "make Iran great again," similar to how he views his own election. He later said he's not in favor of it. While the Israeli Defense Minister was aggressive, Netanyahu said the Iranian people might have to rise up themselves, implying Israel won't do it all for them. Neither Trump nor Netanyahu seemed fully committed.

Yes, the strikes have destabilized the regime by killing top people, which could lead to an internal regime change. However, these senior people can be replaced. In a country of 90 million, there will be plenty of people willing to take those jobs, despite the elevated risk of death.

We also haven't seen the kind of mass protests that would suggest a movement against the regime. We saw protests in 2009-2011, but nothing happened. The only recent protests I'm aware of were student protests against the US and Israeli attacks, which the Prime Minister joined.

As Lisa pointed out, it's possible that the Ayatollah dies in the next six months. He is 86 years old, and an American male of that age has about a 5% to 6% chance of dying in that timeframe. You can argue he has better healthcare, but he has also had past health problems.

His death could destabilize the regime, but it's also possible that, just as after Khomeini died, they will simply elect a new ayatollah and things will run fairly smoothly.

Nuño 00:16:36

It's interesting that Vidor, Lisa, and I have different interpretive frameworks. One framework suggests we're seeing the regime fall in real time. We have reasons to think this is the case: the reduction in the number of mosques, not knowing where Khamenei is, and many top commanders being killed.

On the other hand, you have the "nothing ever happens" school of thought. If you ask about August, which is just over two months away, it just takes longer for things to happen. A base rate of 5% per year doesn't give you that high of a probability in such a short timeframe.

I think Lisa would agree on the base rate but would argue that we now have extremely strong evidence that moves the probability from a few percentage points all the way to 50%. Vidor and other forecasters might disagree. I'm confused and oscillating somewhere in the middle.

Lisa 00:17:45

It's a very confusing time. But one thing I'll note is that both of your forecasts are elevated above the baseline. We all fundamentally agree that things are different now; the only question is how much.

So what has changed? What elevates the risk of regime change above the baseline rate? To me, the highlights are the decapitations of the leadership, specifically targeting the repressive machinery. IRGC commanders and the headquarters for the repressive apparatus within Tehran and around the country have been hit. This is a massive setback for them.

Sure, they still have between 125,000 and 190,000 IRGC members and other personnel who know what to do without being told. But how is that organization going to recover? It's possible it fully recovers and they're able to keep a lid on everything.

But we are at the tail end of this regime. It's been 46 years. People just aren't interested in going to mosques anymore; two-thirds of the mosques have closed. People really don't want to hear it anymore.

They don't want to be told they can't sing and dance in the streets. Women want to walk around without a hijab. Not all of them, of course—nothing is monolithic—but people in general want more freedom than they're getting in this regime.

In addition, by becoming a pariah state, the country gave up so much economic growth. Because of years and years of sanctions, their economy is really suffering. Add to that massive amounts of corruption within the highest levels of leadership. People are very disillusioned. On the one hand, this is supposed to be a regime that holds itself to the highest standards, yet here's this ordinary corruption going on.

They've also structured their economy and power generation in ways that cripple both the economy and reliable power, especially in winter. There are very large structural problems.

Nuño 00:20:41

Can you say more about the power generation?

Lisa 00:20:44

Most of their electricity generation relies on natural gas. They prioritize providing it to households for heating, which makes sense. But in the winter, this creates shortages. If there's enough gas for heating homes, there isn't enough for electricity in major cities.

There can be hours, even days, without power, even in major cities. Businesses shut down, schools shut down, and everyone has to stay home from work because their kids are home. It's massively disruptive, and it's getting worse.

This is compounded by the regime's price controls on gasoline and the state's control over the energy sector. The IRGC controls a large percentage of the nation's economy. These are really large problems, and they're getting worse every year.

There's massive inflation, very close to 40%. People are tired of this. This is not like the Islamic Revolution happened five or ten years ago. This is the tail end of a repressive regime, and people are tired.

Rai 00:22:17

I want to make a few things from your model explicit and get your sign-off. Do you see most scenarios as the regime toppling under its own weight or changing internally?

Lisa 00:22:33

In the absence of an external stimulus, my best guess would have been that the regime would last another one to five years.

Rai 00:22:42

Got it. You can't cleanly separate these factors. Israel will be involved and might accelerate things, but the question is whether they are the driving force. Are they going to be proactive, continually removing people until something happens, or will they be more reactive?

So you see them as being more reactive, with a number of structural factors causing the first dominoes to fall.

Lisa 00:23:09

Yes, but the trouble with an authoritarian regime is that it doesn't allow opposition to grow. So, as we saw in Iraq and Afghanistan, if you remove the group in power, there isn't necessarily another one organically ready to take its place. This will be very difficult for Iran and is the crux of what comes next.

As Vidor said, we'll have to see. There will be lots of people willing to step in and take over in some form.

One of the key questions is whether the inevitable regime change will come fully from the outside, or if it will be someone within the current government who decides to do something different. They might say, "I'm in charge now. We'll keep these aspects of the regime but get rid of those." There are a lot of options for how this could play out.

Nuño 00:24:20

One interesting factor is this: you have people willing to step up and die for the continuation of the regime. But do you have people willing to die and put their lives on the line for a revolution?

The answer could be no. Religion may inspire a much greater willingness to die for the regime, which makes it more resilient. That doesn't seem unreasonable to me.

Lisa 00:24:48

The question is, are you willing to throw away your life to fight the regime, or are you willing to die if there's an achievable goal?

People wouldn't want to stand up by themselves and fight, but if they thought there was a reasonable chance of success, many would be willing to stand up. It's a question of what the realistic chances of success are.

It could also be that people are more willing to stand up for religion than for secular regime change.

Nuño 00:25:25

I'm on board with much of your analysis, but I don't get the timing. Why now? Your analysis seems equally valid if the regime falls three years from now.

Lisa 00:25:37

The timing is crucial for two reasons. First, they're in massive chaos because of what Israel just did.

Second, it looks like Khamenei may be about to die. It's not clear that the full succession apparatus is intact or that it would even be followed. We could see some sort of coup right there.

Nuño 00:26:04

The Khamenei situation is relatively new, isn't it? From the last couple of days?

Lisa 00:26:09

We're seeing more speculation about it, but there have been rumors for at least several months. I saw some last year.

I don't know the nature of his difficulty right now, but it's possible he was injured in an airstrike by Israel. We don't know. Israel might know, but I don't.

Rai 00:26:33

Vidor, do you have a sense of where your crux is with Lisa? What do you think she's weighting too heavily?

Vidur 00:26:41

Lisa is talking about all the problems Iranians are facing, but I think she's weighting them too heavily. Pointing to problems alone doesn't necessarily mean the regime is going to fall.

There are also a lot of social media rumors about Khamenei and his whereabouts, and I'm not sure how much we can trust them.

Lisa 00:27:13

We function better as a group. For those unfamiliar with forecasting, we improve when we challenge and listen to each other, learn new information, and consider different perspectives. This is part of what makes good forecasting.

I may not be right, Vidor may not be right, Nuno may not be right, but together, our averages are usually better than individual forecasts. None of us is infallible, and our forecasts are made better by listening to each other.

For our listeners, please understand that these disagreements are part of what forecasting is about. Nuno, what are your thoughts?

Nuño 00:28:09

You have a point that Khamenei not making statements is a signal. At the same time, I recall previous speculation about this a few years ago. Maybe he just doesn't feel the need to be out there all the time, as he's an older man who isn't as energetic.

Lisa 00:28:32

But he's not even making statements. There's nothing.

Vidur 00:28:38

He has made statements, or at least statements have been put out in his name over the past day or two. His last public appearance, however, was perhaps six or seven days ago.

Lisa 00:28:52

The last tweet from his account was: "Those who know the Iranian people and their history know that the Iranian nation isn't a nation that surrenders."

Nuño 00:29:02

He's tweeting roughly every two days.

Lisa 00:29:03

You know, roughly.

Nuño 00:29:05

Here's a tweet from Khamenei from June 18: "The Zionist regime's malicious attack on our country took place at a time when Iranian officials were indirectly engaged in negotiations with the U.S. side. There was no indication on the part of Iran that signaled a military move."

It doesn't seem very fiery.

Lisa 00:29:27

Besides, how many of these are actually written by him? We don't know if he wrote them or even spoke them to be written down by someone else.

Vidur 00:29:36

Even if he dies, he has reportedly picked his successors. There's a strong possibility the regime will continue. They'll elect a new ayatollah, and everything will be fine for them.

Nuño 00:29:55

There was a taboo about his direct descendants inheriting the mantle of the Ayatollah. There was also speculation about Khamenei solving this by appointing a very old successor.

Lisa 00:30:11

It's not done to appoint your son as a successor. You can overcome that by appointing an interim successor who is not likely to live long.

There's one possibility who I believe is 96 years old. He would clearly have a limited term, after which Mojtaba, his son, could be chosen.

Nuño 00:30:40

That would just carry on the instability. Having a 96-year-old as a wartime leader doesn't seem like a great idea.

Lisa 00:30:52

You also have to ask to what extent that person would actually be leading the country.

Rai 00:30:57

Lisa, how are you thinking about the weakness of the IRGC? There are two ways to look at it. One is that if the recent damage was going to crack their enforcement, it would have happened already. Since it didn't, that's evidence the regime will last longer.

The other way is to see it as a gradual decay. Without their coordination infrastructure, they'll get progressively worse, opening the door for civil unrest and protests. It sounds like you believe in the latter model.

Why isn't the lack of immediate collapse evidence that the repressive enforcement is stickier than you think and won't collapse so quickly?

Lisa 00:31:51

I don't see this as the only thing going on. The state apparatus has been destabilized. That alone might not be enough to topple the regime, but coupled with the potential loss of Khamenei, the combination leads to major destabilization.

Reports I'm seeing from Iran suggest increasing crackdowns on the citizenry. This is done from a position of weakness, a recognition that the regime is fragile. You don't crack down on your people if you feel secure.

People are fearing thousands of arrests and even executions in the coming months.

Rai 00:32:49

The crackdown could be done out of fear, but its purpose is also to increase the threshold for collective action. In that, it could be successful.

Lisa 00:32:59

It could be successful, but it suggests to me that the regime feels threatened. We'll see which way it goes in the coming weeks and months.

It's possible that one or more leaders could rise up within Iran, and people could be emboldened to make a change.

Alternatively, there could be a massive crackdown. The price of action would be increased to where it's too expensive, and everything would go back to the way it was. Same chairs, different people. That's possible.

Nuño 00:33:39

If opposition leaders rise up, methodically killing them is a viable method. Putin assassinated Navalny, who was the last in a long line of potential opposition leaders.

Lisa 00:33:56

And when that was done, there wasn't a massive attack on Putin's regime.

Any way we slice it, this is a period of instability for the regime. The only question is how much. You don't think it's very much, but I think it's a lot.

Nuño 00:34:11

No, I think it's a lot of instability. But when thinking about when it happens, I'm more certain than you are about the next couple of years. I agree that there are cracks in the regime.

Another source of uncertainty is the current US-Israel ceasefire. How long will that last? Will Trump change his mind in the next couple of months?

Maybe Iran will try to race for a nuke, which would lead to more Israeli attacks and further destabilization. That's another source of uncertainty.

Lisa 00:34:44

Definitely. I completely agree with that. There's a lot of unpredictability.

Vidur 00:34:48

I strongly agree. The key uncertainty is how long this ceasefire will last. Who has Trump's ear? Is it the more hawkish neoconservatives, who see this as their chance—after waiting for decades—to get rid of the Iranian regime?

Or is it people like JD Vance, who are urging caution and don't want the US to attempt another regime change?

Rai 00:35:16

Let's do another round of forecasts after this discussion and see if anyone has moved. We'll go in the same order, starting with Nuno. What are your forecasts for regime change in Iran by the end of August and the end of the year?

Nuño 00:35:30

I'll go up a little bit, but I still think August is very soon. Let's say 5% for August and 20% for the end of the year.

Rai 00:35:42

Sounds good. Vidur?

Vidur 00:35:43

I'll go from about 1 in 10 by August to 1 in 9, which is a move from 10% to about 11%. For the end of the year, I was between 1 in 6 and 1 in 7, and I'll now go to 1 in 6. Lisa?

Lisa 00:36:00

It's a period of such huge volatility. Everything changes every single day.

Nuño 00:36:07

And that volatility should move you towards 50%, right?

Lisa 00:36:09

Towards 50%.

Rai 00:36:10

Nuno, can you say more about that? Theoretically, why should that volatility move her more towards 50%?

Nuño 00:36:16

If you have a spectrum with regime collapse on one end and the regime continuing on the other, and you're oscillating between factors that pull you significantly toward either side, then any strong reasons you have could be negated by a future factor. You're seeing a lot of wild jumps.

Lisa 00:36:38

They are very wild jumps, and it's hard to deal with that level of uncertainty. Okay, I'll walk down my forecast. For August, 50%… I can't remember my initial number.

Rai 00:36:54

I think that was your initial forecast.

Nuño 00:36:56

Yes, 50%. And you were at 70%—or 65%—by the end of the year.

Lisa 00:37:00

It was 70%. So, 50% and 70%.

Okay, I'll walk my forecast down a bit. For August, I'll go to 45%, and by the end of the year, 65%. It still seems very likely to me, but maybe I'm completely wrong. We will find out. That's one thing about forecasting: there will be an answer.

Rai 00:37:24

When you update your forecasts, how much of it is based on a specific new piece of information that changed your mind?

And how much of it is simply trusting another person's judgment and moving toward the group average out of caution, even if you don't fully understand their reasoning?

Lisa 00:37:46

I would say it's both. In general, if you have a group of good forecasters and one person's forecast is very different from the others—which is my situation right now—that's a red flag. You should think very carefully about a divergent forecast, because chances are you're missing something.

That said, if you have strong justifications and a firm belief, it's reasonable to maintain a divergent forecast.

I listen to the specific considerations my fellow forecasters raise, and I very much trust their judgment. That gives weight to what they have to say. One could argue that I should weight their opinions even more highly and lower my forecast further.

But it's a dynamic process. Forecasts aren't static; they evolve over time. And that's what good forecasting is.

Rai 00:38:53

If you converge too much, you're just a pack of conformists. You have to be willing to deviate to some degree.

I also want to get your thoughts on the extent of the damage done to Iran's nuclear weapons program. There seems to be wide uncertainty about it. Countries like Israel and the US are claiming victory, while others are doubtful significant damage was done.

What evidence do you find convincing? And ultimately, how long do you think Iran would need to rebuild and get back to enriching uranium to weapons-grade levels? Let's start with Nuno.

Nuño 00:39:39

One specific piece of evidence I've been wondering about is the satellite imagery of Fordo and other facilities being hit with massive bunker-buster bombs. You can see the size of the holes. A bomb that's one meter wide and several meters high creates a hole about five meters wide, with a volcanic-like mound around it that's about 10 meters.

What I'm curious about is the depth of the hole. If this bomb is made to burrow into a nuclear bunker, did it actually reach the 90-meter depth where the bunker is located? It's a simple physics question. It's a whole mountain with a small hole. Maybe they hit the wrong place. I just want to know how deep the bomb went, and we don't have great answers.

My sense is the Americans know. They can use satellites to observe the difference in light and shadow to calculate the depth of the hole. The Iranians probably know, but I don't. I only know what's being publicly reported. The Iranians are saying it's a surface wound. The Americans are saying it's completely destroyed. I'm uncertain.

On top of that, there's meta-level uncertainty. We have satellite images for this one location, but there are other locations. It seems nuclear material was taken out. There are many unknown factors. This is what I've been thinking about. Hopefully, Vidur or Lisa can fill in the blanks.

Vidur 00:41:35

Many people are saying different things. Trump claims the program has been completely destroyed, Israel says there's a lot of damage, and even Iran has said its facilities were badly damaged. All three of these groups have an incentive to emphasize the damage. Netanyahu wants Trump to think his involvement was worthwhile, Trump wants to justify his attack, and Iran doesn't want to be attacked again.

However, a leaked intelligence report suggests the damage wasn't so severe and might have only set Iran back by a matter of months. I think that report is probably directionally correct.

But if you look at the satellite imagery of the above-ground facilities, there has clearly been extensive damage. Some of these facilities were quite important—unless we believe they were just decoys and Iran has other facilities to move their uranium to.

For instance, at Isfahan, there was damage to metal conversion plants and other facilities crucial for the final stages of building a nuclear weapon. At the Arak facility that Israel bombed, the heavy water reactor, which was under construction and could be used for producing plutonium, was destroyed. So, Iran has definitely suffered a setback.

As Nuno mentioned, it's harder to assess the facilities buried deep underground because we lack public information. What Iran and the US seem to acknowledge is that Iran still has highly enriched uranium. This is significant because if they have other facilities and a few hundred kilograms of HEU, they can probably build a nuclear weapon.

Even a crude, Hiroshima-style nuclear weapon can be built fairly quickly. You don't need much HEU for a gun-assembly weapon; the Hiroshima bomb used about 64 kilograms. These bombs aren't very efficient—only about one kilogram of uranium in the Hiroshima bomb actually underwent fission—but you can still make a very simple nuclear weapon.

There are still stages required to assemble the weapon properly, but as long as they have this HEU and can perhaps enrich it a bit more, they can make something of it.

My key uncertainty is this: if they have this stock of highly enriched uranium, do they have other facilities—perhaps not for a sophisticated program, but enough to make a crude nuclear weapon?

Lisa 00:45:15

The question is, how far back has their nuclear program been set? It's clear the setback is somewhere between months and years, but we can't know for sure at this point. There hasn't been enough time to make assessments.

It would be great if IAEA inspectors could be on the ground, but that may never happen. Still, everyone would agree there has been a substantial setback; the question is how much.

As Madur says, it might not take them long to assemble a crude nuclear weapon. These risks are still out there.

The can has been kicked down the road, but the problem has not been eliminated. It's something we will still have to deal with.

Rai 00:46:14

Is it possible that these attacks have made it more likely for Iran to pursue a weapon? Do the statements from the US and Israel, playing up their success, create a problem?

Or will those past statements be forgotten as Iran gets closer to a nuclear weapon, at which point the motivation for further action will return, regardless of what was said before?

Lisa 00:46:38

Being attacked certainly gives Iran more motivation and incentive to develop a nuclear weapon. I don't know what the game plan is from here.

Would the US and Israel simply conduct more strikes in the future? They seem committed to preventing Iran from developing a nuclear weapon, so that would be the implication.

Perhaps they have other ideas beyond airstrikes. We saw the Stuxnet cyberattack disable centrifuges before. I'm not sure where this is going if the regime does not change.

Rai 00:47:25

Stuxnet did more than just disable the centrifuges, which would have been obvious. It modulated the speed of the rotation slightly and even altered the telemetry, so it didn't look like anything was wrong.

Lisa 00:47:41

But in the end, they were non-functional.

Vidur 00:47:44

To your question, Iran has moved to suspend cooperation with the IAEA, so they seem to be becoming more insular and less transparent. I agree with Lisa that this probably increases their desire for a nuclear weapon.

The question is, absent intervention, how long would that take? We can look at China's nuclear program in the 1960s. It took them a few months to get from enriching enough uranium to assembling a weapon for testing. It's quite possible Iran could do it in a few months if they wanted to.

Nuño 00:48:31

Meanwhile, Israel would notice and attack again.

Vidur 00:48:35

If Israel wants to keep attacking, then unless Iran has some extremely secure facilities, Israel and the United States can keep setting them back. That is why our forecast for the probability of a nuclear detonation fell after these US attacks. The US and Israel could keep on top of it if they wanted to.

It's also worth noting that the dozen or more nuclear scientists who have been assassinated represent a significant setback for Iran. There aren't many people with that know-how.

Iran has said it has the indigenous knowledge to continue, but there are few people in the world who truly understand these things. Even if the procedures are written down, my knowledge of bioweapons development tells me there is a lot of contextual knowledge needed to implement them in a real-world setting. That's an important consideration.

Rai 00:49:49

The takeaway for listeners is that even though the strikes were not decisive, we are probably in a pattern where Iran's nuclear program can be continually set back. The net effect is either a lowering of nuclear risk or at least not a massive increase from the background level.

Nuño 00:50:13

You could also have the opposite case. As the regime becomes more unstable, it could make acquiring a nuclear weapon an existential priority, deciding it can only survive if it rushes to build one.

The previous stable equilibrium was that they had the ability to rush but chose not to. So you could imagine a probability increase that way.

Rai 00:50:36

My understanding was that Vido incorporated that desire. Iran could have a ton of desire to pursue a weapon, especially now. With fewer diplomatic off-ramps and no cooperation with the IAEA, they've become polarized—either ditch the program or go for broke.

But the limiting factor is whether that increased desire actually translates to an ability to defend the program from Israel or the U.S. How can you defend it?

Vidur 00:51:08

That's my view, all things considered. However, I do think it's possible that Iran races to a nuclear weapon.

We sometimes assume U.S. and Israeli intelligence is omniscient and infallible, but perhaps they've missed something. If so, Iran might have the ability to weaponize quite quickly.

Rai 00:51:31

Great. Thank you all for taking the time.

Lisa 00:51:34

Thank you.

Rai 00:51:36

Thank you very much to our listeners. If you enjoyed this episode, please consider sharing it with someone.

Most of the benefit of forecasting comes from influencing decisions, and to do that, we need to grow our reach.

If you want to reach out to us, please send an email to [email protected].

Itay Vinik is the cofounder and Chief Investment Officer of Equi, an alternatives investment fund that deeply considers rare market conditions in its investment strategies. He was was the cofounder of a long/short volatility hedge fund which made a famous bet on the 2018 volmageddon event.

Transcript

Rai 00:00:56

Welcome to the Sentinel podcast, where top forecasters discuss ongoing events with a view toward global catastrophic risks. I'm your host Ry, and I'm joined by Sentinel co-founder and forecaster Nuno Sempere, and Sentinel forecaster Lisa.

Our guest today is Itay Vinik. Itay is the co-founder and Chief Investment Officer of Equi, an alternatives investment fund that deeply considers rare market conditions in its investment strategies. He was also the co-founder of a long-short volatility hedge fund, which made a famous bet on the 2018 Volmageddon event. Welcome, Itay.

Itay 00:01:32

Great, thank you.

Rai 00:01:33

Today, we will discuss recent indications of global economic uncertainty, primarily in sovereign debt markets. This is important because large economic struggles—like sovereign debt crises, austerity, and significant inflation in the world's largest economies—can cause shifts in political power and state capacity. These shifts can ultimately affect the trajectory of technology development, political doctrine, and war.

As a recap, weak demand for sovereign bonds sent 30-year US Treasury yields briefly above 5%, German 30-year bond yields to roughly 3%, and Japanese 30-year bond yields to record highs. Many factors were at play: sticky inflation, significant government borrowing in recent years, a costly bill passed by the US House of Representatives, and Moody's recent downgrading of US debt.

Persistent debt issues can put a lot of pressure on societies. What types of political shifts might we see in a world where financing this debt is not cheap?

Itay, since Sentinel and Equi are both concerned with escalations into longer-tail outcomes, what are some of the more impactful economic issues we might see from here?

Itay 00:02:43

The "bad outcomes" that can happen include a haircut default-type cycle. In this scenario, if governments let the market operate freely, higher real yields (yields above the inflation rate) would kill long-term investment. There would be less incentive to invest, and capital would flow back into Treasuries. This typically results in deflation, naturally resolving the debt through defaults. Yields would eventually come down, but with devastating economic consequences.

The second choice, seemingly more popular with policymakers, is to inflate the debt away. By devaluing money, they effectively "kick the can down the road." These are the two main choices.

In my opinion, given the option, policymakers will always choose to kick the can down the road. However, in extreme tail-event scenarios, policymakers might not have that choice. We can discuss that; it's an edge case, but possible.

Lisa 00:03:54

Itay, could you ever see a scenario with an actual default or a haircut on US Treasuries? Personally, I can't envision that. I'd like to hear your opinion.

Itay 00:04:06

I don't really foresee that. The US is quite different from the rest of the world because, at least for now, it has the global reserve currency. As long as the US can create the global reserve currency, the likelihood of a default on US Treasuries is minimal.

However, in the distant future, if that were not the case, a default would certainly be possible.

Lisa 00:04:26

Yet, we see the prices of credit default swaps on U.S. Treasuries going up, so apparently not everyone feels that way. You're saying there's very little risk as long as the dollar remains the world's reserve currency, but we're seeing the world start to pull back from the dollar.

Personally, when I look out over the next decade, I see a very low risk. Over extremely long time horizons, I think that could change, obviously. But at least for the short to medium-term future, that doesn't look like a serious risk to me.

Rai 00:05:06

Let's go around to the forecasters and get a preliminary forecast. It's tough to operationalize this for the US, in particular, because they could monetize the debt. But let's try for some fuzzy operationalization of what a US default might entail.

Lisa 00:05:17

You mean a US default, a straight-up default of some sort?

Rai 00:05:22

Yes, but does that fully capture our concern? They could technically avoid a default by monetizing the debt. However, if the US monetizes the debt, that's still a scenario we're worried about.

Nuño 00:05:33

When you say monetizing the debt, what does that mean?

Lisa 00:05:36

Inflating it away.

Nuño 00:05:37

Okay.

Lisa 00:05:37

To me, inflating it away is infinitely less dangerous than investors taking a haircut. That would destroy Treasuries as a safe-haven asset.

Itay 00:05:50

Consider what we saw this past April: what I call a "triple yazoo" moment, a term coined after the Japanese crash, where the stock market, currency, and bond market crashed simultaneously. That is pretty unusual for the United States. There was also a moment concerning the basis trade, which is how hedge funds lever up with long-dated Treasuries.

If you compare this to other crises, like 2008-2009 or 2011, it's different. 2011 is particularly interesting. In August 2011, S&P downgraded US credit for the first time. The stock market crashed 20%, while TLT, a good measure of long-term bonds, went up 15-20%.

Contrast that with recent events after "Liberation Day": the US bond market sold off hard, along with stocks and the US dollar. This is an unusual scenario.

Lisa 00:06:50

I completely agree that was extremely unusual. But we didn't come to the brink of an actual default. To me, an actual default means a "not getting your money back" type of scenario. I'm not talking about merely losing money in the bond market.

Itay 00:07:08

I'm thinking about the safe-haven aspect and the typical investor mindset where assets are anti-correlated—when one goes up, the other goes down. These correlations broke.

Typically, long-end bonds would do well because when stocks crash, the Fed is expected to cut rates and the economy to slow down, meaning bond yields should go lower. We haven't seen that in this recent move, which is concerning because it suggests the bond market is more fragile than usual.

And I agree entirely: the risk of an actual US default, US Treasuries taking a haircut, or the US losing its reserve currency status is, at this point and for the foreseeable future, minuscule to non-existent.

Nuño 00:08:04

I'm less familiar with the numbers, but I'd initially estimate 0.5% or 0.1% a year. Does that seem significantly off, or would you suggest less than 0.1%?

Lisa 00:08:16

What actual default rate are you considering?

Itay 00:08:19

It's hard to say, but I'd consider it an edge case, probably around 0.1%.

The global financial system's infrastructure and plumbing are fully built on U.S. dollars. Approximately 65% to 70% of all FX reserves are still in U.S. dollars. Furthermore, 80% to 90% of transactions in commodities markets and other infrastructure are conducted in dollars.

Saudi Arabia's recent deal with the United States reinforces the petrodollar system, ensuring dollars continue to be used for oil exchange. For example, if a Norwegian company digs for oil in the North Sea and borrows money, they most likely borrow and secure it in dollars.

A vast amount of dollar-denominated debt is also held between countries and by emerging markets with U.S. banks. Because this entire system operates in U.S. dollars, replacing it is extremely difficult and will likely take many years.

When people discuss the end of the U.S. dollar, the question becomes: what is the alternative? What could replace the vast existing dollar-based infrastructure? Currently, no pure substitute exists. It would be a very slow unwinding of the current system. This is a matter of decades or more, not something in the near term.

Nuño 00:09:54

Regarding the long or medium term, how do you define those? Is it a couple of years? Five years?

Are we looking at 0.1% for five years, or for 50 years? As a trader, what do you mean by long term and medium term?

Itay 00:10:11

It's somewhat arbitrary, but the longer the time frame, the greater the uncertainty and the higher the probability of unlikely events. For the next five years, I'd estimate 0.1%, increasing thereafter. It probably wouldn't be significantly more for 10 years. Beyond 10 years enters the medium term.

Twenty to thirty years from now, all bets are off; it's very difficult to predict. Global reserve currencies have historically lasted between 120 and 200 years. However, things are moving faster now.

Before the U.S. dollar, the British pound was the reserve currency, used in transactions. The Netherlands and Spain also had periods where their currencies were dominant. These shifts occur based on the dominant global power of the era.

For now, what business or company would agree to transact in, say, the Chinese yuan at scale? Perhaps Russia is compelled to due to sanctions, but widespread adoption by others is unlikely. It's hard to imagine companies widely adopting a substitute for the U.S. dollar.

However, we are slowly transitioning from an American-dominated world to a more multipolar one, offering more alternatives and options. But I don't see an immediate systematic threat that could dismantle the current system.

That said, high debt levels and bond yields do pose a substantial threat to the economy.

Rai 00:11:45

We've discussed the U.S., Itay. What are your thoughts on the situation in Japan?

What is the impact for Japanese investors and policymakers? What do you foresee?

Itay 00:11:59

This is a very loaded and complicated topic.

In historical context, Japan, the third or fourth largest economy in the world—it fluctuates a bit, and also depends on the exchange rate—is the first to experiment. When I say 'first to experiment,' I'm referring to their early adoption of Keynesian economics, the first quantitative easing (QE) programs, zero interest rate policy, and negative interest rate policy. They were the first to try these because they experienced no growth for long periods.

There was a massive boom in the 1980s, arguably the world's biggest bubble. In 1989, the land value of the Imperial Palace in Tokyo was reportedly worth the same as all real estate in California. Things were absolutely crazy. The bubble popped in 1989, which was devastating, and Japan entered the 'lost decade'—a long period of stagnation, lost generation, and numerous issues.

Policymakers responded by taking interest rates to zero and injecting money into the economy using quantitative easing, putting them ahead of the West. Japan is also ahead of the West in demographics. They have a falling birth rate; today, it's around 1.1 or 1.2, while the population replacement rate is 2.1.

A lot of inflation and economic growth comes from demographics. More people having children leads to more home buying, large purchases, and car sales. For instance, the high inflation in the U.S. in the late 1970s and early 1980s was associated with peak Baby Boomers entering the workforce and household formation age. This demographic shift arguably contributed to inflation. Inflation isn't purely monetary; it can be organic due to increased physical demand for goods.

Japan was early to these trends. Some predicted other developed economies would face a Japan-like climate 10 to 20 years later, and we've seen some of that happen.

What's happening in Japan today is quite interesting. Since the early 1990s, there has been no inflation, resulting in complete behavioral shifts. Consumers don't rush to spend because they anticipate prices might be lower next month or won't rise.

Real estate values are not expected to rise. It's not common to buy land or a house expecting an automatic increase in value. The consumer mindset is quite different there.

Thirdly, because Japanese government bonds yielded close to zero for a long time and rates were so low, the yen became a global funding currency. People could borrow in yen and invest elsewhere—the carry trade—expecting a positive yield.

What is different this time is that Japan has inflation. I've been a trader for a while, and shorting Japanese government bonds was always known as the 'Widowmaker' trade because it never succeeded. It's notable that the situation is changing now, as that trade never actually worked out before.

The Japanese market never imploded because the assumption was that with so much debt and yields at zero for so long, even at extremely low rates, about 15% of all tax revenue was already servicing existing debt a couple of years ago. The idea was that if rates reached a certain level, the portion of tax revenue needed to cover debt would become substantial. At current rates, it might already be 40% to 50%, though I haven't verified this.

This is the potential end game of Keynesian economics: reaching a point where you can't continue because there's too much debt. Even small interest rate increases can make it impossible to cover the debt. Then, more debt must be issued to service existing debt, consuming too much tax revenue and leading to a debt death spiral.

Nuño 00:16:37

If people borrow money in yen to invest elsewhere, that depresses the value of the yen. How does that play into this?

Itay 00:16:45

Correct. You're effectively short yen, which is beneficial. If you borrow in yen and its value falls, you pay back less yen on your loan, resulting in a net gain. You're making money in two ways: from the yen's depreciation and from the appreciation of your other investments.

For example, if you borrowed yen to buy U.S. Treasuries before the recent rise in yields, you might make 5% on a U.S. bond while borrowing at 1% in Japan. You'd earn 4% on the interest rate differential, plus an extra amount if the yen depreciated, say from 140 to 150 per dollar. Many people have been doing this.

Nuño 00:17:33

How has the value of the yen been changing recently? Is it becoming more valuable now that interest rates are greater, or how is it changing?

Itay 00:17:43

The yen appreciated back to the 140 range and is currently around 145. It recently peaked around 160, then strengthened towards 140, and is now at 145.

What you're bringing up is actually the bigger risk. One of the biggest risks today is that we don't know the full value and size of this carry trade. We don't fully know how much money was borrowed globally to buy bonds, equities, or even real estate. It's very hard to quantify.

Some estimates are a few trillion dollars, possibly higher, depending on how it's tracked. For a known figure, the total Japanese holding of U.S. Treasuries is around 1.7 trillion, I believe.

Rai 00:18:42

It's about 1 trillion.

Itay 00:18:43

Okay, so take that as an example. Currently, the difference between a 30-year Japanese bond and a U.S. 30-year bond is not that high. Typically, that spread is 3-4%, but now it's less than 2%.

Keep in mind that some Japanese investors, like insurance companies, hedge currency differentials. When hedging, they pay a certain amount, perhaps 100 to 150 basis points. At this point, it may not even make sense for them to buy U.S. Treasuries.

A big tail risk is that Japanese investors decide to repatriate their money and buy Japanese bonds. This would cause the yen to appreciate as all that capital flows back into Japan.

Lisa 00:19:25

The big question for me is: one, how far up will Japanese bond yields go? Two, to what extent does the Bank of Japan want to intervene? And at some point, can they intervene to stop a drastic increase in yields?

The Bank of Japan commented yesterday that they would change their future bond offerings, which calmed yields a little bit on long-term Japanese bonds. But what's the future here? How likely is it that things will get radically out of control with Japanese bond yields?

I think the Bank of Japan is likely to intervene before that happens, and I believe they still can. However, there's a non-zero risk here of things getting out of control.

I also wonder how big the impact of that will be on bond markets globally. I suspect the impact would be substantial but not extreme. I'd be interested to hear your take on that.

Itay 00:20:28

I generally agree, but the tail risks are extreme here. The idea behind the Austrian school of economics is that when you continue this course of action, you eventually reach the endgame. Based on the Austrian idea, the endgame is reaching a place where you cannot continue because of inflation.

They might have to make a choice. Tokyo inflation came out in April at 3.4%, which is still much higher than the Japanese central bank rate. The Bank of Japan may decide to intervene and resume its QE programs.

The Bank of Japan already owns over 50% of the entire Japanese government bond market. Before 2024, they had yield curve control.

Nuño 00:21:16

The Bank of Japan owns 50% of the stock market?

Itay 00:21:20

54% of the Japanese government bond market.

Nuño 00:21:23

Of the bond market.

Itay 00:21:25

Yes. Because of their policy of yield curve control, they wanted to suppress yield on the long end for years. They kept buying bonds at those prices to ensure lower yields on the back end. That's how they kept their rates so low on both the front and long ends of the curve.

The problem is, it's almost like nationalizing that market, and that was one reason the yen kept appreciating for so many years. The issue now is if Tokyo has real inflation for the first time in decades—which recent evidence suggests, as their inflation rate is 3.4%—they're running a real risk.

They can intervene, and likely will. They can decide not to raise rates anymore or even resume yield curve control. But if they do, they'll sacrifice their currency. The message they send is the yen is going to 200, and they won't be able to stop it, leading to a currency crash. This is because they would need to create an insane amount of yen to make that happen; they will need to print a ton of money.

So, they can do it, but they will sacrifice their currency. Currently, they have a hard choice between their currency market and their bond market. They have a very difficult job to walk the line in a way that can save both.

In theory, Japanese inflation can go back to 2% and stay there. Their problem was they didn't have inflation at all; it was zero or negative. They want some inflation, but not too much. They finally got what they wanted, but they got too much of it.

Lisa 00:23:02

They have a complete catch-22. Can you envision a scenario in which yields keep going up and they don't intervene?

Itay 00:23:11

That would happen if inflation starts spiking out of control in Japan. Consider rice prices and the CPI in Tokyo, which is 3.4%. I was in Japan in March, and people there are still conditioned by decades of deflation.

For instance, one person told me he was delaying renting an apartment because the price went up, expecting it to come back down as it always has. The idea of deflation is so ingrained that the concept of persistently rising prices is something many can't comprehend.

This makes it very interesting because we'll see if this inflation in Japan is transitory. If it's not, it's a structural shift that's almost unprecedented.

To give you an idea, based on data from about two years ago, household savings in Japan are close to 14 or 15 trillion dollars. A massive 55% of that is in cash. Imagine that money starting to move. This reflects a very conservative, deflationary mindset: keep cash because prices will be lower tomorrow.

Changes of this magnitude in such an economy could be instrumental. The global implications are also significant. For instance, a much stronger yen could cause asset prices worldwide to sell off aggressively.

The tail risks are on both sides: an extreme tail with the yen going to 100 or below, or to 200. It's unlikely to stay within the current range unless the Bank of Japan gets very lucky and stabilizes it.

Lisa 00:25:01

Is it likely that Japan will face some sort of crisis here? If so, on what timescale might that happen?

Itay 00:25:10

Historically, I would have said this is something everyone predicts but never happens. It's the well-known 'widow-maker' trade; many traders have tried to short Japanese bonds without success.

But now, for the first time I've seen, Japanese yields are genuinely imploding. It's signaling real risk to the market. The longer the time frame, the more real this risk becomes.

To give you an idea, consider total debt numbers and country defaults. The statistic is that of 42 countries reaching a debt-to-GDP ratio of 130% or 140% or higher, 41 have defaulted on their debt. Japan is the sole exception; it's the one country that reached that level without defaulting.

Rai 00:25:56

And Japan is at higher than 260% debt-to-GDP.

Itay 00:26:00

People have been trying to understand how this is possible. The reason is that much of the Japanese bond market is owned internally—by insurance companies and the public. There are very few outside investors causing significant capital outflows.

So, it's hard to say. But there's significant risk on both sides. If inflation persists in Japan, policymakers will have to choose the lesser of two evils: inflation and yen devaluation, or a very difficult economic time.

If they experience a recession and stagflation simultaneously, then tariffs couldn't have come at a worse time for them.

Lisa 00:26:42

We're just piling these impending crises on top of each other.

Itay 00:26:48

The stock market is still at, or very close to, 6,000 in this recent recovery. So it hasn't cared all that much.

Lisa 00:26:56

Very true.

Nuño 00:26:56

If you think about the tax burden in my native Spain, it's significantly higher than in the US. However, nothing is breaking over the short term. This does lead to brain drain and various undesirable effects, but Spain is still limping along.

So, why can't the US gradually move its tax burden, hidden or otherwise, to a level corresponding to Spain?

Itay 00:27:20

Spain's inflation rate is actually not—

Nuño 00:27:23

No, the tax rate. The tax rate is.

Itay 00:27:25

Oh, the tax rate is very, very high. If the U.S. increases taxes, you're right, it would potentially close the deficit.

Nuño 00:27:33

I'm thinking about the combined tax rate: the actual tax rate plus the hidden inflation tax. If you gradually increase your actual tax rate to the levels of Spain, nothing breaks.

If you gradually increase your combined tax rate—actual tax plus inflation tax—to that level, where in this process do you get a crisis? Or do you just get progressively worse standards of living?

Itay 00:27:58

Governments should do something to rein in reckless spending and bring in more revenue. However, there is a concept that has been very hard to anchor: the Laffer curve. This is the idea economists have explored concerning how much tax is the right amount, an age-old question.

Nuño 00:28:21

Right.

Itay 00:28:21

The Laffer curve suggests that if you have too much tax, the government collects less money. This is because fewer businesses form, as entrepreneurs start businesses elsewhere to avoid high taxes, leading to less economic activity. Consequently, even with a higher tax rate, the government nets less tax revenue.

Conversely, if taxes are too low, economic productivity might boom, but the government collects a smaller percentage of that larger amount. There's an optimal amount, but it's difficult to determine precisely; it's likely a floating, not fixed, number.

It is known that the US still enjoys a very strong technological moat. Most innovations and large-scale breakthroughs have been happening in the United States since the 19th century, from the invention of the telephone, electricity, and the light bulb, to modern advancements in AI with companies like Nvidia.

The US still provides a very good climate for entrepreneurs and innovation. There would be a substantial risk in trying to remove that, as it's one of the key factors giving the US a massive competitive edge globally.

Nuño 00:29:31

Great.

Itay 00:29:32

What we're discussing in this political era of populism is not dissimilar to the 1910s and 1920s. When people's standard of living decreases, there tends to be a movement toward more extreme political parties on both the right and the left.

During periods of prosperity and economic stability, public opinion tends to cluster more in the center. This is a politically neutral observation supported by a lot of statistical work.

Lisa 00:30:11

That's absolutely true. On one hand, when conditions worsen for the population, their politics shift to greater extremes. Yet, this is perhaps when moderation is needed most.

When discussing restraining budget deficits and getting budgets under control, as Itay mentioned, there are primarily two levers: taxes versus spending. In the US, we're seeing additional proposals for tax cuts for the rich. The size of these tax cuts is so large they will drastically increase the US budget deficit and debt burden over time, which becomes problematic.

There are these trade-offs, and it's up to society as a whole to choose the final outcome. As Itay brings up, unfortunately, the population's choices become more extreme as conditions deteriorate. This leads to a difficult situation and potential tail-end risks.

Itay 00:31:30

One potentially easier solution is getting a hold of the interest payments. Interest payments are so much greater than any individual tax programs or tax cuts.

The US spends roughly 1.2 trillion on servicing the debt. If that number could be cut in half, it would solve many problems.

Lisa 00:32:00

Instead, interest payments are escalating due to these increasing yields. Then there's the risk posed by the Japanese bond market potentially exacerbating the situation, and the whole carry trade imploding.

Itay 00:32:14

Ironically, the healthiest thing in the long run might be the most painful in the short run: a recession. If we had an economic downturn now, with a good amount of U.S. treasuries needing to be refinanced, a recession could kill short-term inflationary pressures. This would allow the Fed to lower interest rates.

Lisa 00:32:36

This is true.

Itay 00:32:37

The Bank of Japan could also lower interest rates. We would be able to refinance the debt, potentially saving 500 to 600 billion a year and locking in lower coupons for 10, 20, or 30 years. This would solve a lot of problems.

Lisa 00:32:51

This wouldn't just help the U.S. economy. What happens in the U.S. does not stay in the U.S.; U.S. Treasury yields affect yields all over the world. If U.S. borrowing costs are high, it has a rippling effect globally. This is a worldwide problem because all our economies are linked.

However, you will never publicly hear a politician or central banker say they want to see a recession.

Itay 00:33:30

Hidden under the surface may be a desire among policymakers to trigger a recession. This might sound like a conspiracy theory. If I were the Fed, understanding the mechanics—and the Fed has repeatedly complained about unsustainable US government finances, as Powell recently reiterated—I would keep rates high right now. This is despite some softness in economic surveys, a higher probability of recession, and tariffs.

Consider the trade wars with China. The Bank of China significantly reduced rates and is using liquidity measures to fight the trade war. The Fed is doing none of that; it's keeping rates high. While reducing quantitative tightening, it's not truly aiding the US in its trade war, maintaining an apolitical stance, perhaps believing tariffs pose an inflation risk.

The longer the Fed keeps real yields positive, thereby sucking liquidity from the economy, the greater the probability of a recession. If I were Powell, I might want to trigger a small recession—one that hopefully doesn't escalate—to allow for rate reductions, assist the fiscal side, and cool potential persistent inflation. This might explain why Powell isn't signaling an imminent easing of policy.

Rai 00:35:04

If yields stay high, and the U.S., Japan, and potentially others deal with some combination of austerity or currency devaluation, what are the geopolitical implications? Could it potentially affect defense spending, which in turn impacts the balance of power?

Itay 00:35:22

Interesting question.

Nuño 00:35:23

The US and its allies would see their ability to coordinate and build further diminish, reducing their willingness and capacity to fight an actual war. Paradoxically, in such a situation, a war might even be seen as a way to kickstart the economy.

Itay 00:35:39

World War II is an example; it took the US out of the Depression.

Nuño 00:35:44

In a perverse way, that logic can hold.

Lisa 00:35:47

Good economically, but not in any other way.

Itay 00:35:50

War is only stimulative to the economy as long as it's not fought on your own soil.

Lisa 00:35:55

Exactly.

Itay 00:35:57

The US benefited so much from World War II because the war was fought across the ocean, allowing the US to become a giant factory for the world.

Lisa 00:36:05

Right.

Rai 00:36:05

World War II also allowed for significant capacity build-out. The crucial factor is whether there's exploitable slack in the system, including an industrial base with room for expansion.

If an economy is already at its efficient frontier, increasing demand for these non-productive assets becomes problematic, as it diverts investment from other areas.

Itay 00:36:30

Additionally, I doubt the US currently has the capacity to build anything substantial. For instance, approximately 96-97% of all shipbuilding now occurs in Asia.

Lisa 00:36:41

It's enormous.

Itay 00:36:42

Many critical items cannot be built domestically. From a national security perspective, in a potential war with China, the US might hold a paper advantage, but over time, it would face significant logistical and manufacturing challenges.

Lisa 00:37:01

A modern war between the US and China would involve many factors beyond direct military conflict, such as cyber attacks.

Itay 00:37:18

Some of that is already happening.

Lisa 00:37:21

Regarding the impact of unhealthy bond markets and increased government austerity, an obvious consequence is that any government experiencing this would be weaker in addressing crises.

Itay 00:37:41

Yes.

Lisa 00:37:42

The U.S., in particular, would be in a weaker position to address global conflicts. Concurrently, the US has signaled less interest in such engagements, so significant changes are underway.

If Western economies struggle, addressing a potential conflict in Europe between NATO and Russia would become harder. Similarly, a conflict between the US and China would be more difficult to manage.

When economies struggle, growth is hindered, and the capability to address any crisis—be it geopolitical conflict or a new pandemic—diminishes, making solutions more difficult.

Itay 00:38:45

Furthermore, protectionist policies often lead to increased conflict and isolationism and have historically resulted in hot wars. While we lack a sufficient sample size to definitively predict a hot war, history suggests this deglobalization trend could usher in a 10- to 20-year period of conflict.

Lisa 00:39:09

More economic integration among countries reduces the risk of conflict. Conversely, more isolationism increases that risk.

Itay 00:39:21

I recall the democratic peace theory, which posits that two genuine democracies have never gone to war with each other. I believe that theory still holds.

Nuño 00:39:33

Looking at population pyramids, the US is in a decent position; its pyramid appears reasonably rectangular despite an aging older cohort.

In contrast, some European countries, like Spain, grapple with a significant debt burden. Their options include inflating the debt away, increasing taxes, or attempting to raise productivity. These nations face a growing debt burden, an aging population, and fewer young people entering the labor force.

This creates a spectrum: Paraguay, for instance, has an approximately 40% debt-to-GDP ratio and a favorable population pyramid. South Korea has less of a debt problem but a very challenging demographic outlook. Some European countries face the worst of both: high debt and poor demographics. Any thoughts on this?

Itay 00:40:29

It's a huge problem, especially in Europe. As I mentioned, Japan dealt with that problem earlier, with Europe about 10 to 15 years behind. The difference in Europe is that European countries like Spain, Portugal, or Italy don't have independent control over their monetary policy.

This lack of control is one reason for the structural pressure on the Euro over many years. In theory, Spain maintains its debt levels without much higher bond yields due to its Eurozone membership and the power of the integrated euro area.

Imagine if Spain were on its own with its own currency. Given its debt load, what would its bond yields be? Today, Spain's bond yields are lower than those in the United States.

Nuño 00:41:25

What are a few possible endgames here?

Itay 00:41:28

There are many ways this situation could resolve. The demographic challenge, for example, is a major issue. The US doesn't face it to the same extent because it generally accepts large amounts of immigration, which contributes to population growth.

Nuño 00:41:43

Spain, in particular, does benefit from that because we have a massive Latin American population that is culturally similar enough. Germany, for instance, doesn't have access to such a large pool of German-speaking immigrants.

Itay 00:41:56

What is the percentage of immigrants in Spain?

Nuño 00:41:59

Around 20%.

Itay 00:42:00

That's pretty high; I didn't realize it was that high. Spain might be a little different then.

Nuño 00:42:03

I also didn't realize until very recently.

Itay 00:42:05

That's a lot higher than I expected. Most of them are probably from Latin America, speak the language, and integrate well.

Nuño 00:42:11

They do speak the language.

Itay 00:42:12

That's pretty great. This immigration could be a good solution for Spain, but it does have a high debt load.

Productivity growth will also be substantial. And this brings us to AI; we've gone an hour and a half without mentioning it, so it's time.

Nuño 00:42:30

AI plays into the factors Lisa was mentioning. If AI is propping up productivity in the United States and elsewhere, that increases the US's strategic advantage in that scenario.

Itay 00:42:46

AI is the ultimate X-factor. In a more linear world without such technological optionality, demographics would be destiny: an aging, retiring population and insufficient workforce.

However, AI might mitigate this. If one person can do the job of ten, the dire population pyramids in Western countries might not be as critical. Perhaps it won't matter as much.

Rai 00:43:12

Itay, thanks for coming on. Can you tell the audience about Equi and what you're working on?

Itay 00:43:17

We are a multi-manager absolute return hedge fund. We focus on what we call "absolute return"—creating a consistent outcome not necessarily correlated to market beta or general market movements. This means your financial outcome isn't solely determined by concerns about equities or bond markets.

The idea is to create an alternative return stream through absolute return. Our goal with these products is to generate a certain level of return. As a portfolio allocation, it can act as a diversifier, though not strictly a hedge, against global market events.

Basic portfolio theory and the efficient frontier suggest that more risk typically yields more return. However, with uncorrelated assets, you can shift that efficient frontier to the left, achieving the same unit of return while reducing risk.

We effectively build funds and products that give clients the ability to shift their efficient frontier to the left. We primarily work with accredited investors, qualified purchasers, registered investment advisors, and multi-family offices. We offer a combination of off-the-shelf products and custom solutions built for our clients.

Rai 00:44:44

Great. Thanks again for coming on.

To our listeners: if you enjoyed this episode, please consider sharing it with someone. The primary benefit of forecasting is influencing decisions, and to achieve that, we need to expand our reach.

Also, if you'd like to reach out to us about anything, please email us at [email protected].

Scott Eastman has been professionally forecasting for over a decade with major forecasting groups, some of which advise the US intelligence community and executive branch.

Transcript

The transcript is AI-generated and slightly differs from the phrasing used in the recording.

Rai Sur

00:00:16

Scott Eastman has been professionally forecasting for over a decade with major forecasting groups, some of which advise the US Intelligence community and executive branch. He's also one of Sentinel's forecasters. His areas of expertise are geopolitics and epidemiology.

One thing I enjoy about talking to Scott is that he has a pretty atypical background. He spent many years painting homes and doing documentary photography. His travels and eclectic interests have brought him in contact with many influential and downtrodden people.

Before we get into some of the questions about risks and escalations, I have some lighter questions that I personally want to know the answer to, and I think the audience would enjoy. The first one is: do you have a type of forecasting error that you systematically make, one that took you a long time to correct for, or that still bites you?

Scott Eastman

00:01:11

A basic forecasting error I made frequently in the beginning, and one that still bites me sometimes, is overreacting to the news of the day. It's very easy to get swept up in it. News is often presented as breaking news; everything is urgent, but usually, not much changes.

If you wanted to be a good forecaster, in terms of a good Brier score or number, you could simply say status quo. If you said whatever happened yesterday is likely to happen tomorrow, you would mostly be right. This isn't always true for fields like the progress of AI or computers, but even there, following the existing curve would often be correct.

It's easy, especially with issues like war, to overreact to a country moving assets or a leader making a statement. So, as a forecaster, it's generally wise to downplay the spikes that occur regularly.

However, a challenge is that simply stating 'status quo' isn't very useful, because the goal is to be a *truly useful* forecaster, not just one who is often correct. Pointing out major changes is the difficult part. For example, many didn't foresee the fall of the Soviet Union, though some had a good idea Russia would invade Ukraine.

Honestly, most of us don't know with 100% certainty when something will happen. If a rare, eventful occurrence has a 1% forecast from me, and I raise it to 10% and it happens, one could argue I was still wrong, as 10% implies a 90% chance it wouldn't. It's a significant challenge. If I say something has more than a 50% chance in a specific time period, it should be truly meaningful.

Basically, overreaction is the biggest thing to avoid.

Rai Sur

00:03:15

Is there an example in the news now where you can feel it tugging at your mind, making you think it's a big deal?

Scott Eastman

00:03:22

A recent example is when the US appeared likely to bomb Iran's nuclear facilities. They may still do so. The US took many preparatory steps: moving B2 bombers to the region and deploying expected aerial and anti-ballistic defenses. These were signals of intent. We knew Israel wanted it bombed, Iran was weakened, and other factors pointed to an imminent strike. But it hasn't happened.

This is often the case. While Iran isn't friendly with the West, most leaders are, to some degree, rational. Even if perceived as extremists, they want to preserve their power and their country. If they perceive inaction will lead to a devastating bombing, potentially regime-ending, they will make moves to reduce that likelihood.

Often, some forecasters might say an event is extremely likely. For instance, during the Ukraine war, US intelligence reportedly estimated up to a 50% chance of Russia using a nuke within days or weeks. It didn't happen. That estimate was extraordinarily high and concerning.

At the time, I spoke with some excellent forecasters. One, who is typically better than me, estimated a 30% chance, which is wildly high for nuclear use. After more discussion with more people, we realized that while current events suggested a high probability, many factors would intervene to lower it. The US strongly opposed it, it would be bad for Russia, and other parties, including China, urged against it, even against testing.

Our core group's estimate eventually came down to perhaps 4 or 5%, which is still a very spooky number. This was based not just on current knowledge, but on anticipating all the intervening steps that would occur before such an extreme event, steps not yet apparent.

A similar dynamic occurred with Pakistan and India. During a recent conflict, both nuclear powers understood the horrific consequences of escalation. Even in simpler situations, it's human nature to try to survive and avoid conflict.

So, without being Pollyannish, optimism about things taking a better path is more often correct than not.

Rai Sur

00:06:44

One way to frame this is distinguishing between forced steps and steps that allow for de-escalation. Perhaps for subsequent steps that could involve de-escalation, we too easily project that they will resemble the current situation.

Rai Sur

00:07:03

What's a forecast you're proud of and one you wish you could forget?

Scott Eastman

00:07:07

One forecast I wish I could forget, and that many of my fellow forecasters also wish they could, is the 2016 election of Donald Trump. This isn't about ideology; it's about not following basic, clear information.

The clear basic information was that most polling going into the election showed Hillary Clinton ahead, but within the margin of error. Her lead was consistently within a few points.

From the beginning, I don't mind having given her more than a 50% chance of winning, but I do regret giving her around a 90% chance.

When candidates are polling very close and then a significant event occurs, it can shift things. Shortly before the election, the Access Hollywood tape emerged, which was detrimental to Trump. However, the subsequent reinvestigation into Clinton's email server—even though nothing new was found—gave some voters another reason to doubt her or decide not to vote.

Those events should have been enough to adjust the forecast significantly. If I had predicted a 60% chance for her to win and Trump had won, I would have been comfortable with that. I would have been on the wrong side, perhaps, but I would have appropriately acknowledged the chance of it going the other way. But I didn't, and I should have known better.

Sometimes such errors are based on emotions, or on an awareness of living in a bubble where, in my case, Clinton was more popular than Trump. I'm fully aware of that. I remember in 2004, living in a very liberal part of the country where John Kerry had overwhelming support. But I knew the middle of the country is conservative, as I used to live there, so I should have known better.

Still, when everyone around you shares one perspective, it's very difficult, even if you diligently follow all the steps of a good forecaster, like considering the other side and various factors. We still get emotional.

It's very tough for me to forecast in areas where I have strong personal biases. I try to set them aside, but I don't believe we can ever fully block them.

A forecast I am proud of—though 'proud' is a difficult word given the terrible outcome—concerns the Russian invasion of Ukraine. I had many good signals indicating it was likely to happen. I didn't know for certain it would happen or its extent, but well over a month in advance, I forecasted that an invasion was more likely than not.

I also believed it was highly probable they would attempt a large-scale invasion, aiming for Kyiv and other major areas, not just confining it to the Donbas region. I care about this particular forecast because it had the potential to be useful.

I had spoken with people experienced in refugee situations and had previously worked on refugee flows, including those from Syria during its war. We know that anticipating a refugee crisis two or three months in advance allows for crucial preparations: pre-stocking supplies, readying borders, and mobilizing personnel. These actions can prevent a complete calamity.

The war itself wasn't stopped. Ideally, someone could have determined Russia's definitive invasion plans and taken steps to prevent it. While that didn't occur, I believe the forecast provided a valuable heads-up for surrounding countries on managing the impending migration crisis.

Rai Sur

00:10:45

As you became increasingly convinced the invasion would happen, did you find yourself unable to see why it wouldn't, yet still feel some internal reluctance towards that conclusion?

Or was it a more fluid realization, where recent events made it clearly inevitable to you?

Scott Eastman

00:11:07

I had an internal dialogue and experienced cognitive dissonance leading up to the Ukraine invasion. Up until that invasion, almost everything Putin had done made sense to me when viewed from what I perceived as his perspective.

When he invaded two separatist regions in Georgia, areas that didn't want to be part of Georgia, I thought he was entering an area where he was essentially welcomed. His intervention in Syria was at Bashar Al Assad's request, so he had some support there, even if not universal. When he started his presidency fighting in Grozny, Chechnya, it was a continuation of a pre-existing war. While not necessarily wanted, it was a continuation of conflict.

Most of the steps he had taken throughout his career seemed rational from a Russian perspective. Though I'm not Russian, I try to study the person, the country, and their history to gain a deep understanding of how their motivations might differ from mine or those of other countries in the region. This understanding affects my assessment of how much they might be willing to lose to achieve their objectives, such as taking land.

However, with the full-scale Ukraine invasion, I thought, "This is going to be a disaster. This one doesn't make sense. This will be a losing deal." I didn't think that about the far eastern part of Ukraine. That region is not just Russian-speaking—as many pro-Ukrainian people are—but more pro-Soviet and less inclined to align with the West. So, I could understand actions in the far east, but trying to take more of the country seemed like a mistake, and it has proven to be one.

But he did it anyway. That was the dissonance: in my view, this would be the first time he would make a major international blunder. On the flip side, he was taking all the steps one would expect for an invasion. I became more convinced when they started moving blood supplies, which have a very short shelf life.

However, a leader wanting to scare someone to the negotiating table might still do that—take all preparatory steps up to the last minute. So, I still remember being shocked, even though I knew it was very likely. I also thought it would happen after the Olympics. He invaded right after, and we had thought he wouldn't want to upstage Xi Jinping during the Olympics, as he needs Xi, the leader of China, as an ally.

Yet, there was still a sinking feeling that this was a mistake, but it was happening. Even though I thought it was likely, up until the last minute, there's always a chance they'll back down. They didn't.

Those were the main reasons: primarily, thinking this was a mistake, even from a Russian perspective. They've lost about half the land they originally took. If he had originally said, "I'm going to take the Donbas, this region, and stop there," I would still consider it morally wrong, but militarily and politically, it might not have been a mistake.

Rai Sur

00:14:15

You used to do documentary photography, is that right?

Scott Eastman

00:14:17

Yes.

Rai Sur

00:14:17

What's different about the world when you're walking through it with serious photography equipment? What differences do you notice?

Scott Eastman

00:14:26

As an observer, sometimes I can be a fly on the wall, hopefully unnoticed. When doing documentary photography, I used fairly large cameras, like the Canon EOS-1Ds with large lenses. This meant I wasn't inconspicuous unless at a mass event where people didn't pay attention. In some parts of the world, an expensive camera made me stand out.

When I do documentaries, unlike being a tourist, I enjoy deeper interactions. Tourists see beautiful things but generally don't interact extensively with local people. In documentaries, even with just photography, I eventually needed to connect with people, even ordinary individuals. Meeting someone on the street could lead to an invitation to their home. I've spent the night with a Roma family, for example. I end up in closer situations than I would as a tourist.

While doing a documentary in a country or region doesn't make me a native or local, I believe I get a closer view of what the place is truly about, beyond just admiring beautiful buildings. When you start talking to people and gain a deeper understanding from them, your perspective changes because you learn what's really important to them.

It's also easy as an outsider to discount people, perhaps judging a group as not very smart or wealthy. But you realize many people, whether a farmer or someone else, are wise in their own field and understand their region well. This relates to forecasting in general: if someone becomes a leader, even of a village, they are likely quite clever to have reached and sustained that position.

Beyond documentaries, in my own life, I've done all sorts of jobs, from being a part-time janitor during vacations to manual labor. I'm acutely aware that systems need everyone to function.

Recently, at a prominent company, a guard said, "I'm just a guard," or someone might say, "I'm just a janitor." But if any of those people fail in their job, there are problems. Each job and each person is important. This isn't just some platitude; it's genuinely true.

Through travel and trying to see a whole society—not just attending social events—I've gained a deep view of how places work and how people are.

Rai Sur

00:17:18

There's a lot of hidden knowledge in the world that can't be simply Googled—things like insights gained from meeting people, understanding their intentions, and private conversations. How much of your forecasting edge would you attribute to these tacit, un-Googleable insights versus your ability to aggregate easily accessible public information?

Scott Eastman

00:17:41

I would think that in maybe 80-85% of the forecasts, Google's going to get you a long way, and you're going to be fine because most things do just follow the trends. But there are some times where I feel like if you don't really understand a culture or know somebody who does that you can trust, you're going to miss some really major points.

An example of this, a really small event, is Iceland. Iceland is an intensely proud country. They have basically no military, and yet they challenged the British Navy at one point. They got into a dispute over fishing rights, decided to challenge the British Navy, and actually captured some British people. They fouled one of their ships, I think by putting nets or something in the propeller. This is crazy if you think about it. They're a country of 300-and-some-thousand people, and they could get crushed.

The same thing happened when they had a financial crisis. Europe was saying, "You've messed up, you need to pay us." All sorts of people had made deposits in their banks, which were getting a high interest rate. They were saying the people of Iceland need to bail them out. And Iceland was like, "No, we're not doing that. You took a risky bet, you lost. Whatever the minimum guarantee is that the bank had, we're going to cover a certain amount of money." They paid that, but they're not going to pay any more. They said, "No, you took a risk, sorry."

Some people in Europe were saying, "If you do this, you're going to be ruined because nobody's going to deal with you anymore." And they're like, "No, sorry, we're not going to make our people cover your risky things." Knowing that Iceland had already essentially gone to war with the UK, it's like they're not going to back down on this stuff. They're also in a strategic position in the world, so nobody wants that. The US still needs them. There's enough going on, and it's probably not worth somebody fighting for there. So, they are able to punch way above their weight.

Then there are other areas of the world, other countries where, if you know their history—for instance, I find it really difficult as an American to understand why Serbia and Croatia have differences that go back 800 or 1,000 years. But some battle matters intensely to them. Americans don't care about the War of 1812 at all. I don't hear any American going, "We don't like the Brits, they burned down the capital." It's just not an issue.

Sometimes, if you do have this local knowledge, it helps a lot to know when something is almost impossible politically for somebody to do. Trying to figure this out with leaders is difficult. You can try to get at some of it by reading biographies if possible. Sometimes, if you know enough about the history of a person, you could understand why they're never, or almost never, going to back down on some issues.

I think it helps to travel around, meet people, and be as broadly knowledgeable as possible. Some of that can be gained from reading from the internet, but not just from a simple search.

Rai Sur

00:20:40

Nice. Okay, on a little grimmer topic. According to Our World in Data, the projected number of base deaths in 2030 is 68 million. That assumes no major catastrophes, status quo. As we go through the orders of magnitude of excess deaths on top of that, what are your top theories for what will cause them? Let's go as a spectrum.

Scott Eastman

00:21:02

The easiest way to get a large number of deaths beyond current rates would be massive conflict. I usually try to look at base rates from the past. Big issues in the past include the Cultural Revolution in China, World War I, World War II, which all had mass deaths, and then pandemics.

Pandemics normally occur every hundred years, but there's no guarantee that because we just had one, we have another hundred years to go. COVID clearly caused more than 10 million excess deaths. Those are the first things that come to mind.

Nuclear war is very low on my list. We've never had a nuclear war since more than one country possessed nuclear bombs. Even with nuclear war, I doubt it would start with countries attacking each other's capitals, as that's a quick endgame. My guess is that even nuclear war wouldn't radically increase that number. I don't want to find out, but it's not so likely.

I would first look at naturally caused pandemics or any disease that wipes out far more people than normal. A normal flu year in America may kill 50,000 people, but globally it's much more, as America is a small portion of the world. A really bad flu could cause high numbers. To get into even higher numbers, I consider pandemics.

Looking out to 2030, my biggest short-term concerns involve AI, and for me, 2030 is still short-term. Some of my colleagues believe AI will wipe us out before then. I am not in that camp.

However, if there's an ability to adapt a virus—and this doesn't necessarily require AI, as countries have attempted to create biological weapons in the past, though largely stopped—that capability will become easier for more people. If that capability becomes easier and isn't contained, it's an area where something could easily escape bounds and kill masses.

Most kinetic events, like big bombs or terrorist attacks, affect a limited number of people. If we're talking about 68 million deaths per year, 9/11 killed about 3,000 people. Even the Ukraine war, which is horrible, is still well under a million deaths. The war in Sudan is also awful, but these events aren't changing the overall death numbers significantly. That's why I start thinking about biological issues.

I consider it extremely low risk that artificial general intelligence or a singularity would create masses of robots to extinguish us. I think there are too many steps along the way where we could see it happening and try to stop it. It also assumes that AI or robots would consider wiping us out their top priority, which I don't believe. Many steps would have to occur for that to happen.

Unlike base rates, AI is an area where we don't have a base rate. We are going into the unknown. In some ways, I feel we're like the world in 1945, or more importantly, when the Soviet Union first got nuclear weapons. If you asked someone in 1950 about the chance the world would not see another nuclear attack by 2025, most people would have thought it absurd; they would have expected a nuclear war before then. They had no priors.

With AI, it would be naive to say that because nothing bad happened with nuclear war from 1950 to 2025, we have at least 75 years of no issues. We simply don't know.

We can use our best judgment and try to figure out scenarios. One way to look at the future is to create scenarios and ask what's likely, and if something is likely, what are the probable counters?

We're in an area where I find it foolish when someone says, "I know this is going to happen," or "We're going to be extinguished by 2032," or "Everything will be fine in 2032." I think we don't have enough knowledge to make a very confident call.

Rai Sur

00:25:27

I know you're interested in the shorter-term impacts of AI and how those could butterfly out to affect many other aspects of society. Could you say more about these short-term impacts and how you think they might play out?

Scott Eastman

00:25:42

It's strange when I think about short-term impacts because we've experienced many technologies in the past that transformed society, such as the printing press, personal computers, and steam engines. Industrialization brought significant changes.

However, with AI, the change is incredibly rapid compared to past transformations. Much of this impact will be on high-skilled, cerebral jobs.

Consider the practice of law. Many years ago, a friend worked on a massive law case, possibly the Microsoft antitrust case, which required document discovery. Traditionally, hundreds of lawyers would manually review emails for relevance.

Now, technology would scan and identify relevant documents with very few lawyers, as you must disclose all evidence to the opposing side. If this reduces the need from, say, 700 lawyers to just 10, that's a significant reduction in the number of lawyers required.

Does this mean we'll need fewer lawyers, or will the cost of legal services decrease, leading to more lawsuits as people find it cheaper to file them? I've represented my own company in small claims court, and AI could simplify case preparation by finding relevant precedents.

I'm unsure how much AI will eliminate jobs versus creating new applications for those skills. For instance, AI speeds up programming, meaning fewer humans are needed for the same amount of code.

Perhaps we'll create more beautiful, complex programs or use programming in more capacities than before. I suspect we'll need fewer programmers five to ten years from now, but I don't know how far-reaching these societal changes will be.

A major concern is job loss. I find the idea of a minimum wage funded by a tax on AI, where everyone receives a basic income, somewhat unconvincing.

Historically, when have companies willingly distributed large sums of money to help the general populace? A government could, of course, impose a tax, and that revenue might sustain us.

Rai Sur

00:28:32

A plausible scenario is that if the alternative to providing support is significant social unrest, the private sector, which benefits from automation, might contribute some of its gains. There's an incentive to placate a large number of people as automation progresses and eliminates more jobs, to prevent their operations from being curtailed by significant social disapproval.

Scott Eastman

00:29:09

Where people decide to start bombing data centers, for example.

Rai Sur

00:29:14

Or just using democracy to...

Scott Eastman

00:29:17

...change the system or shut it down. I can see that. However, if AI is as transformative as many believe, reaching a point where half the jobs are redundant, we would need to subsidize half the population who have little to no work. That would be uncharted territory.

We currently have social safety nets in the US and Europe, with ongoing discussions about their appropriate level. I'm doubtful they would be enough to maintain a high standard of living.

I'm also concerned about meaning. Even a basic job, like cleaning toilets or washing floors, provides a sense of accomplishment. I've done such work, and at the end of a few days, I could see a difference; I had done something.

I'd rather do that than sit at home, receive money, and immerse myself in VR, games, or alternate realities. While I enjoy those things somewhat, I want my life to have purpose beyond mere enjoyment.

I believe most people need a reason to wake up and do something meaningful, perhaps helping others, which is part of being human. If that is largely taken away, it might be fine; perhaps we won't need to work for money and will engage in activities like philosophy groups. Regardless, it will be a challenge.

Rai Sur

00:31:12

Besides unemployment, another short-term uncertainty with AI is its effect on the social realm: relationships and how people interact. Do you have thoughts on what we might see and the risks involved if these effects were to escalate?

Scott Eastman

00:31:32

There are many challenges with this. One is that an AI can respond in the way you want, controlled by the company that wrote it. It's in their interest to maintain our engagement with the AI. If we disengage or spend less time per day, the algorithm might be changed to re-engage us.

There may also be an underlying agenda. For instance, authoritarian countries, or any country, might want their society to behave in certain ways. The AI company simply wants you to stay engaged and buy products or whatever generates revenue.

A key challenge for me, as someone old-fashioned, is that human-to-human relations remain important, not just human-to-AI or human-to-robot interactions. If an AI consistently responds in a way I prefer, and then I deal with a human who is sometimes difficult, I might question why I'd interact with that challenging human. When my bot is always friendly and provides good, useful information, tolerance for human inadequacies or outbursts could become very low.

I find this not only with AI but also when dealing with high-level forecasters. I enjoy interacting with them because they are typically skilled conversationalists who engage in high-level discussions without personal attacks, even with differing views. This leads to great intellectual conversations.

Conversely, when I talk with people who lack these skills, I sometimes feel disinclined to engage in what might seem like unproductive conversations. The same could happen with AI if it consistently provides excellent information.

For instance, I have a friend who is a brilliant astrophysicist, and I greatly enjoy our conversations, even though I'm not an expert in the field. However, if I could ask an AI like Gemini a question, it might already provide answers as good as his in many areas. I'd still prefer talking to my friend due to my preference for human interaction, but eventually, one might wonder why consult a human guru when the computer offers comparable expertise.

The prospect of AI intentionally swaying us is frightening. I've seen U.S. and European elections significantly skewed by algorithms promoting specific candidates or viewpoints. This can affect personal relationships, social cohesion, and politics; it's alarming how detrimental this could become.

At the same time, people growing up with this technology are becoming accustomed to challenging authority, which can be positive. When I was growing up—I'm over 50—we had limited news sources: a few major TV stations, a local newspaper, and perhaps The New York Times for some families. We all drew from the same core body of knowledge, and most people accepted news reports as true without question, then discussed them.

Now, people question if content is AI-generated, if what they see is real, if a video is authentic, or if they're interacting with a real person or AI. This leads to a decay of trust in some ways, but it also fosters a healthy skepticism, encouraging people to seek multiple sources and verify information.

It's not entirely negative. In the right hands, this technology can make us more perceptive and questioning, hopefully in constructive ways, though it can also go the other way.

Rai Sur

00:36:01

What other aspects do you find underrated, especially concerning Sentinel's coverage?

Scott Eastman

00:36:08

One thing that is easy to ignore is climate change. Most of us are aware of it and accept that it's real, and that humans are having a major effect on it. But it's not something that changes day by day. Even when a massive event like a huge hurricane or forest fires happens, it's still not global at one moment. So, it's easy to keep that as never the main issue of the day.

Even at Sentinel, where there is a goal to look at pressing issues, including some of the longer-term issues, it's not always rising to the top.

I've lived in different areas of the world where it's either intensely clear that climate change is a pressing issue and even the next decade is scary, or where it seems really distant. If you live in an area that gets plenty of water and where temperature extremes are not bad, it might not feel urgent. If you're in a place that doesn't get over 100 degrees Fahrenheit, or 38 degrees Celsius, in the summer, and someone tells you the Earth is going to raise another one or two degrees, it might sound good.

But if you're living in Arizona or large parts of the United States that now have smoke for much of the summer, even if you're pretty far east but there are massive fires in the west and you need to wear a mask for a month, it feels a lot more real. This is obviously even more so for areas hit by hurricanes.

We know climate change affects wars. There's big concern across sub-Saharan Africa and Central America that as there's more drought and famine, there will be migrations. Migrations, or just scarcity, often tend to lead to conflict and human suffering. That is really concerning.

A challenge is that science increasingly has some ability to counter global warming. I've been on projects looking at how we can try to engineer our way out of climate change, to reverse it. There are some ways to make progress, but those are also scary. If you try to bioengineer something to be more reflective or cover rocks with something to absorb CO2, anything on a large enough scale to have an effect would be a massive experiment that may have unintended consequences.

Some science-positive people, and I'm generally science-positive or human-positive, say that eventually we will come up with a lot of solutions. We have already made many positive moves, like windmills, solar panels, electric cars, and I'm still not against all nuclear power. We've gotten better in many areas.

However, I am still deeply concerned about climate change and how we deal with it. It's also one of the areas where we have the most ability to positively affect it. If taken on as a priority, it could be an economic driver, not just a negative. That's an area we sometimes skip, depending on the administration. Even in a relatively liberal administration, it's usually not the number one thing we're dealing with.

Rai Sur

00:39:47

A common operationalization of climate change is the change in average temperatures, in degrees Celsius. But that doesn't really seem to give us a good intuition for how climate change could then affect other conflicts or create instability.

Do you have any ideas for specific operationalizations of climate change that might be closer to those kinds of impacts? This could help us make bright lines in the sand and reliably look at them, so we have a way to coordinate around when it's important to mention climate change.

Scott Eastman

00:40:26

One would be if we can quantify famines and deaths from famine that we can separate out from normal cycles. Even in the 1970s or 80s, though climate change was already happening, there were still famines, for example, bad years in Ethiopia. But if you could try to figure out migrations and famines that you could relate specifically to climate change.

One of the challenges with this is it's really hard to say this specific storm or this specific spike in temperature this year is only because of climate change. We know temperatures are going up, but attributing a specific event solely to climate change is difficult.

For example, in the Western United States, we're starting to get "atmospheric rivers." In the past, a big rainstorm might last an hour or two. Now, you might have a cloud parked over a city for 60 hours, dumping a year's worth of rain. That's not normal, and these are becoming pretty common in California. Depending on where they park, they can cause mass flooding and destruction.

This year in the Los Angeles area, there were mass fires. If you have mass fires followed by atmospheric rivers, you get mudslides. You could look at how many houses are destroyed as a result of extreme weather compared to an average year. If you just took the United States, where there's lots of data, and looked at housing units historically, year by year, how much has been taken out and how much is it going forward? That may be a factor.

But I still don't think it's going to make everyone care. The people in the areas hit hardest will care. But if you're living in Norway, even if many people there are compassionate and want to help, it's probably not going to personally affect you very much. The same for someone living in Germany, which is another country that seems to care about climate change but is a pretty green, relatively cool country. People living in England, I work with a lot of them, and I don't think they're having the same feeling of urgency as someone living in the southwest of the United States.

So I'm not sure how you actually make somebody care when it's not in their backyard.

Rai Sur

00:42:53

What's something that I should be asking you about?

Scott Eastman

00:42:56

One of the goals of forecasting is to give people a heads-up about what we need to be looking at—things we can do something about. This might be because we're curious and care about global events, or because they are immediate concerns.

It's a huge challenge because some important areas are not currently forecasted. For example, an erosion of democracy matters to me, though I don't know how to forecast it. In countries where freedom of speech and the ability to make change are quashed, we have less resilience and ability to react. Hopefully, authoritarian countries have benevolent governments that address problems, but I'm less confident in that.

I'm concerned with issues of democracy and how we, and our information, become increasingly siloed. Beyond our sentinel reports highlighting global issues, we should also consider how to change our own thinking. This includes how we process and seek information, and how we step outside our comfort zones to consider views that disagree with our own.

This is important for us to do. It's not a direct forecast, like predicting conflict in Sudan next week, but rather understanding how our thinking influences our future.

Rai Sur

00:44:31

That's what I hope to do with this podcast: provide the background color for things that can't be neatly put into a line item. For instance, concerns about the erosion of democracy still factor into your world models and your outlook on the future.

That's a great place to wrap it. Thank you, Scott, for coming on.

To the listeners. If you enjoy this episode, please consider sharing it with someone. Most of the benefit of forecasting comes from influencing decisions, and to do that, we need to grow our reach.

Also, if you want to reach out to us about anything, please send us an email at [email protected].

Nuño Sempere, Vidur Kapur, and Lisa discuss the possibility of nuclear escalation between India and Pakistan.

Timestamps

(00:16:00) - Intro

(00:43:00) - Background on the conflict

(02:26:13) - Direct forecasts of nuclear weapons use

(15:45:10) - The role of other nations

(20:26:09) - What if Pakistan collapses?

(22:45:16) - Russia's incentives in global conflicts

(25:40:08) - What are the forecasters looking out for next?

(29:28:06) - Outro

Transcript

The transcript is AI-generated and slightly differs from the phrasing used in the recording.

Rai Sur

00:00:16

Welcome to the Sentinel podcast, where top forecasters discuss ongoing events with a view towards better understanding global catastrophic risk. Today, we'll be talking about the recent military escalation between two nuclear weapons states, India and Pakistan. I'm your host, Rai, and I'm joined by three of Sentinel's world-class forecasters: Nuño Sempere, who's also my co-founder at Sentinel; Lisa, also known by her forecasting moniker, Be Like Water; and Vidur Kapur. The situation is evolving rapidly, and we are recording this on Wednesday, May 7th.

To ground the discussion, the focus will be on nuclear risk, as Sentinel is concerned with escalations in the long tail of extremely bad outcomes. Most of that risk here is nuclear.

About two weeks ago, a terrorist attack in Jammu and Kashmir, a contested territory on the India-Pakistan border, killed 27 people (mostly Hindus) and injured 20 others. This caused an immediate diplomatic breakdown that has since escalated to military action at a scale not seen since the last Indo-Pakistani war in 1999.

India's nuclear program conducted its first nuclear test in 1974, and Pakistan had theirs in 1998, but neither has tested a weapon since 1998. The countries have very different nuclear doctrines. India committed to a no-first-use doctrine in 1998, meaning they pledged not to use nuclear weapons proactively. Pakistan, however, affirmed as recently as this conflict that it expressly does not have a no-first-use doctrine. With a significantly weaker conventional military, Pakistan uses the threat of nuclear first use as a deterrent.

Regarding their nuclear arsenals: India and Pakistan both possess numerous warheads and missiles capable of covering each other's entire territories. As of 2013, Pakistan stated it would need to assemble a nuclear-armed missile from parts stored in different locations, though it's unclear if this remains true. India has new missiles that can be permanently mated with warheads and stored ready for use. It's unclear if any of these missiles are currently mated to nuclear warheads.

With that background established, I'm curious where you all are starting on this question. What's your rough estimate of the probability of at least one nuclear weapon being launched by either India or Pakistan in the next 365 days? How are you arriving at that, and what are you considering? Let's start with Nuno.

Nuño Sempere

00:02:44

Before this recent escalation, I was at approximately 0.1%. I arrived at that by decomposing the question into: first, the chances of a significant escalation, and second, conditional on that escalation, the chances of a nuke being used.

In our forecast a week ago, we defined the threshold for significant escalation as 1,000 deaths. This is much higher than the 27 deaths seen so far. However, India's response seems more kinetic than I expected.

If I was previously at a 12% chance of significant escalation and a 1% chance of nuclear use conditional on that, my estimate for hitting that 1,000-death threshold might now be around 15% to 20%. This brings my overall estimate from 0.1% to perhaps 0.2%.

Overall, I believe it's still unlikely because both sides really don't want it. However, there's a dynamic where both sides are uncertain, capable of making mistakes, and could start on a path they cannot stop.

Rai Sur

00:03:58

Vidur, what about you?

Vidur Kapur

00:03:59

There are a few different ways to approach this question. One is, like Nuno, to take this in stages. You could look at the probability of a significant escalation—for example, 1,000 military fatalities within the next 12 months—and then, conditional on that, the probability of nuclear weapon use.

For me, I'm now at around 20% for 1,000 military fatalities within the next 12 months. Conditional on that, I'm at possibly around 2% to 3% for actual nuclear weapon use.

This puts me a fair bit higher than Nuno, at around 0.4% to 0.6% overall. Taking the average, I estimate a 0.5% chance of nuclear weapon use by either India or Pakistan within the next 12 months.

Rai Sur

00:05:11

It might be good to contextualize this second part: where the nuclear use percentages come from.

Vidur Kapur

00:05:16

Absolutely.

Rai Sur

00:05:17

There seems to be a range of numbers most forecasters coalesce around for nuclear weapon use. How do you arrive at those figures? What's their origin?

Vidur Kapur

00:05:29

The second part is to determine the base rate for nuclear weapons use in a conflict. In the past 80 years, coinciding with the 80th anniversary of the end of World War II, there has been essentially one instance of nuclear weapon use in conflict—the two bombs in Japan, which I count as one event. This gives a base rate of 1 in 80, or about 1.25%.

Looking closer, when the United States used nuclear weapons, it was the sole nuclear state. Japan couldn't retaliate as it lacked nuclear weapons, possessing only a crude program. The Soviet Union, Germany, and Britain also didn't have them at that time. This suggests the probability of nuclear use in the modern age is lower than 1.25% because states are hesitant to break the norm against their use.

In the India-Pakistan case, both have nuclear weapons, meaning mutual use would cause immense destruction on both sides. This might push the probability lower. However, Pakistan's military doctrines suggest a willingness to use nuclear weapons under certain circumstances. One scenario involves a battlefield commander using tactical nuclear weapons, which Pakistan possesses. Doctrines indicate a willingness to use them, especially against an existential threat to Pakistan.

One could also imagine a general, in the fog of war, deploying tactical nuclear weapons to halt an Indian advance. If India made an incursion, a general, perhaps misinformed about broader Indian gains, might panic and use them. This specific scenario is concerning and pushes my estimate higher than 1.25%.

Generally, the ongoing conflict itself suggests a higher probability. If India and Pakistan were at peace, nuclear use would be unlikely, barring an accident. But with an active conflict and reports of downed Indian jets, we should consider a rate higher than 1.25%, possibly 2% to 3%.

We also know that former Secretary of State Mike Pompeo stated in his memoir that during the 2019 India-Pakistan escalation, both countries were very close to using nuclear weapons. While he might have exaggerated for his memoir, it's another data point suggesting these countries are among those most liable to use nuclear weapons in a conflict scenario.

Nuño Sempere

00:09:25

Fedor, that 1.25% is for all countries collectively. What share of that risk do you assign specifically to India and Pakistan? Considering the large number of nuclear states, the base rate per year for a state *not* using nuclear weapons is quite high.

Conversely, there have been events close to nuclear weapon use. North Korea's nuclear tests to deter the US are an example. More recently, Russia's seizure of the Zaporizhzhia nuclear plant could also be considered, potentially expanding the reference class. What are your thoughts on this, especially regarding the share of the 1.25% attributable to India and Pakistan?

Vidur Kapur

00:10:22

My estimate of 0.4% to 0.6% implies a significant share for India and Pakistan. Much of the remaining risk, in my view, currently lies with Russia in Ukraine, as Russia has explicitly discussed using nuclear weapons. In autumn 2022, Western policymakers were very concerned about Russian nuclear use; the British Prime Minister was reportedly monitoring weather patterns due to fears of nuclear fallout. So, Russia and Ukraine account for a large share.

However, aside from these two, it's justifiable for India and Pakistan to represent a substantial portion of the nuclear risk within the next year, given their current conflict. As we speak, Indian pilots may be hospitalized due to downed fighter jets. Indian officials have briefed The New York Times and The Hindu that at least three fighter jets are down.

While they don't constitute the entire risk, they justifiably form a major part of the nuclear risk for the coming year.

Rai Sur

00:11:44

Lisa, what are your thoughts here?

Lisa

00:11:46

My forecast for nuclear weapon usage risk aligns closely with Nuno's, around 0.1% to 0.2%. I'll say 0.2% due to recent escalation. However, my overall view is that because both are nuclear powers, the risk of them ever using nuclear weapons is extremely low; the consequences are too high. An incredible escalation would be necessary before we'd see nuclear weapons used.

Consider Ukraine versus Russia: despite numerous threats of Russian nuclear use contingent on Ukrainian actions, it hasn't happened. No rational country wants to use nuclear weapons in almost any scenario. Therefore, the risk is extremely small.

Most of the current risk, in my opinion, stems from accidents or misunderstandings. For instance, if one country inaccurately perceived the other's actions, it might lead to a misjudgment and nuclear use. This is the most likely path to nuclear weapon use in this conflict.

Another factor is the significant imbalance in country size, military size, and military budgets. Pakistan spends $10 billion annually on defense, while India spends $81 billion. Due to this asymmetry and other reasons, neither side wants this conflict to escalate. They will likely try to de-escalate after responses that satisfy domestic political needs.

This is my framework: events have occurred, and each side must respond to some degree, but neither wants an uncontrolled escalation. I don't see a substantial risk of this escalating to actual nuclear war.

Vidur Kapur

00:14:18

A counterargument is that in the Russia-Ukraine scenario, while both sides really, really hate each other, Ukraine was held back. The United States supplied many of Ukraine's weapons and calibrated its support to avoid nuclear escalation. President Biden recently acknowledged in a BBC interview his deep concern about World War III and nuclear escalation, stating he didn't want to "play roulette."

With India and Pakistan, you have two sides that really hate each other, and emotions run high. Neither side wants nuclear escalation. However, if mistakes occur, or events deviate from plans, such as fighter jets being shot down, it creates an incentive to respond more forcefully.

While the chance of this leading to nuclear escalation is very small, if fighter jets are shot down, India might feel compelled to respond even more strongly. Pakistan already feels the need to respond to airstrikes on its territory. This could escalate because both sides feel the need to save face, perceiving themselves as losers in the scenario, and there's no one effectively holding them back.

The United States might intervene, but it is currently preoccupied with other crises like Gaza, Iran, and Ukraine. This creates a potential path toward nuclear escalation, driven by the need to save face and heightened emotions.

Lisa

00:16:19

Vidur, your point about the role of other powers is excellent. In the India-Pakistan case, no single power is directly restraining either country as we've seen with Russia-Ukraine.

However, another angle to consider is that other powers are implicitly backing India and Pakistan. If the India-Pakistan conflict escalated, we must wonder if it would draw in China, the US, and Russia. I believe these powers are already implicitly involved.

Rai Sur

00:17:07

Let's discuss other powers and the potential for de-escalation.

Vidhur, you mentioned the US might have its hands full. It's unclear to me how many foreign crises the US can manage simultaneously. You implied this might not be on their radar due to existing commitments. Given current US engagements, how involved do you think they are, and how much will they try to de-escalate this situation?

The same question applies to other countries: What are their interests, and how much attention are they paying?

Vidur Kapur

00:17:45

It's an interesting question how much a country can take on at once. In the US case, several factors are relevant. Sometimes, the same top personnel handle multiple conflicts, so a drive from top leadership to achieve resolution makes a significant difference in negotiations.

Disinterestedness could play a role. If top leaders, for instance, are too occupied with other priorities to focus on this, it might slip under the radar. However, you're right; the US government has many employees, and diplomats can often proceed without constant top-level direction. Still, central leadership matters in government.

I'm also curious about Lisa's points on Russia, China, and potential US involvement. I agree they are implicitly funding and selling arms to India and Pakistan.

Lisa

00:19:11

Exactly, they are already involved in a sense. If things escalated, I wonder if the United States would implicitly back India—though Pakistan has been an ally, albeit to a lesser extent—and if China, which sells arms to Pakistan, would implicitly back Pakistan.

As in all conflicts, there's a broader context and potential for wider involvement. All major powers, in addition to India and Pakistan, would prefer this conflict not to escalate. Nobody wants escalation.

Fundamentally, India and Pakistan don't want to solve this problem now. It's too large and won't be resolved overnight or through military tit-for-tat. This is a conflict that will likely continue and simmer for a long time, with occasional flare-ups. I foresee it continuing this way for many years.

Nuño Sempere

00:20:25

Beyond international dynamics, another factor is Pakistan's internal situation. The Pakistani state and military appear reasonably weak. The Kashmir conflict began, in part, because they couldn't effectively suppress rebels wanting to annex Indian Kashmir. Additionally, rebels in Balochistan are killing police and staging attacks.

Considering a scenario where the India-Pakistan conflict reaches 1,000 deaths, is my 1% probability of further escalation too low? Is there, perhaps, a 2% chance of Pakistan collapsing? If so, where are the nuclear weapons physically located? Are they in the capital or more spread out? Is there a chance a rebel group could acquire a nuclear weapon?

For example, Bangladesh seceded from Pakistan with India's help. Could something similar happen with Balochistan? I'm not certain about the distribution of nuclear weapons within Pakistan, but this uncertainty might be a reason to increase my conditional probabilities of escalation.

Lisa

00:21:45

To me, the overriding concern is the mutual assured destruction doctrine, which seems to supersede everything else.

Nuño Sempere

00:21:57

I agree with that overriding logic, but it assumes the current Pakistani state is in control. If I'm considering 1% or 2% probability scenarios, I might need to condition on that assumption being incorrect.

The Pakistani military opposed Imran Khan, who was popular. Could protests arise, combining various dissatisfied groups? If I consider the chances of the Pakistani government falling within a year, does that probability reach 1%? I'm honestly not sure.

Lisa

00:22:33

In a sense, what you're describing is the risk of nuclear proliferation. There's a non-zero risk associated with that concern.

Nuño Sempere

00:22:44

Another factor on the international stage is that the US and China don't want to see nuclear weapons used. However, Russia's calculus is a bit more complicated. Russia, to some extent, benefits from chaos, particularly if it captures some US attention. It is also more integrated with North Korea and Iran.

It would also benefit from a war involving Israel, as this would divert US attention from the front in Ukraine. If the US were decisively on Pakistan's side, Russia might benefit from supporting India, creating another front. What are your thoughts on that?

Lisa

00:23:28

I completely agree that Russia likes to sow chaos everywhere. The more chaos there is, and the more other powers are occupied dealing with it, the more Russia can do whatever it wants in the international domain.

However, even Russia doesn't want nuclear war. It doesn't take many nuclear bombs being dropped to plunge the world into a nuclear winter and crisis. Things can get out of hand quickly. Even Russia doesn't want that level of chaos.

While it might be in Russia's interest to let this situation fester, I don't see that as a factor leading to nuclear escalation beyond the general risk inherent in allowing chaos. I don't view it as a major factor increasing the risk of actual nuclear war, though Russia does indeed love chaos.

Nuño Sempere

00:24:27

I don't see that as a major factor either. It doesn't have to be a large nuclear weapon resulting in nuclear winter. A small tactical weapon could be used. However, you wouldn't want to bomb a region you consider your own.

Lisa

00:24:50

From my point of view, there's a very tiny chance this could escalate out of control. Vidur sees a higher potential risk. Nuno, you and I are at very similar probabilities, though you might be edging slightly higher than I am that something could go out of control. Vidur is perhaps looking at broader possibilities for how this conflict could go.

We always have to ask this question when nuclear-armed powers are involved; it's a reasonable question. We are concerned with the potential for catastrophic events, but I believe we would all agree that the overall chance here is very low.

Nuño Sempere

00:25:37

Agreed.

Vidur Kapur

00:25:38

Yes.

Rai Sur

00:25:39

It's good to know that the absolute risk of nuclear weapon use is quite low in all your opinions. What are you looking out for now? What developments would make you significantly update your assessment?

You mentioned the thousand-uniform-death line you forecasted. Are there other indicators in that realm, or what else are you considering?

Lisa

00:25:59

For me, I'd be watching how much this escalates conventionally. If we saw a thousand military deaths, it would be clear the conflict is escalating.

However, even then, I wouldn't substantially increase my personal estimate of nuclear war risk. I might go up by 0.1% because any conflict increases risk. Even with a thousand total military deaths on both sides, I would still expect the conflict to escalate conventionally and not move to nuclear.

Vidur Kapur

00:26:35

I'm also looking for military deaths, particularly since we've mainly seen civilian deaths so far. I'm watching how prisoners of war are treated, if any are taken. For instance, if Indian pilots are captured, how are they treated? Will they be sent back to India or detained? And how would India respond?

I’m observing the extent of Pakistan's retaliation, or its expected retaliation, against India. Airstrikes would likely be used, though we've already seen artillery. Will they escalate, or respond in kind, doing something similar to India's actions? If a Pakistani jet is shot down, what happens then?

There's also a tail risk from social media. A lot of information and rumors circulate, which has always been true in war, but social media amplifies it. People believe what they see on social media, which can exacerbate situations and potentially lead to poor decision-making. I'm watching whether anything big and sensational is reported on social media and how policymakers grapple with that.

Nuño Sempere

00:28:09

I'm also observing the strength of the Pakistani state relative to terrorist groups in its territory. A big warning sign would be either side changing its military doctrine on nuclear weapons. Russia, for example, changed its doctrine a few times during the war in Ukraine to make deployment easier; that would be a significant warning.

I'll also watch how much independence both sides grant their commanders in operations. If India, or Pakistan, grants commanders full military independence, they could potentially use nuclear weapons.

I’m still monitoring whether Balochistan will make a play for independence amidst chaos and confusion. Afghanistan has been supporting Islamic resistance or terrorist groups in the region, so I'm curious how that develops.

A possibility is that the situation calms down, but the Pakistani state doesn't effectively assert itself against terrorist groups in the region, leading to a repeat of what we've seen for the last couple of decades. These are a few things on my mind.

Rai Sur

00:29:26

I think that's a great place to wrap up. Thank you, everyone.

To our listeners, I hope you enjoyed this discussion. If you have any feedback, please send it to [email protected]. We will see you next time.