TrueValueMetrics ... Impact Accounting for the 21st Century

Date: 2025-07-04 Page is: DBtxt003.php txt00028552

COMMENTARY PAUL KRUGMAN A recording from Paul Krugman and Nate Silver's live video PLAY YOUTUBE VIDEO Original article: https://paulkrugman.substack.com/p/live-with-nate-silver Peter Burgess COMMENTARY The format for this video does not work well for me. While I am confortable with much of what Paul Krugman has to say, the same is not true for Nate Silver. And to some extent ... this starts to raise questions for me about Paul Krugman! Peter Burgess
Live with Nate Silver ... A recording from Paul Krugman and Nate Silver's live video Paul Krugman and Nate Silver Feb 27, 2025 As promised, a live chat with Nate Silver without a single mention of either poker or sports — or, for that matter, monetary policy! Thank you Laura Fenton, Michael Chang, Marty Neumeier, J Dziak, Sonya Lea, and many others for tuning into my live video with Nate Silver! Join me for my next live video in the app. Transcript Okay, we are live. Okay. If you're doing this now, you're among the first 10 people in the first few seconds. I am Nate Silver. I am here with Paul Krugman. I am excited to speak with Paul. I think we have an agenda that includes talking about Trump and how it's going. I think we both will give you an answer that's not going that well, but maybe from slightly different perspectives. We want to talk about media stuff. We are both now former New York Times employees who have had not parallel trajectories, but have a lot of experience writing about geeky stuff in the media. And then I want to hash out some AI stuff too with Paul, but do you want to start on, on Trump? Okay. So, you know, I expected the worst. Uh, I was, I was, was and am terrified, uh, about where we're going. And, um, you know, uh, with the, um, uh, It's still a definite possibility that 2024 was the last real election we'll ever have. So, you know, let's be serious. But I did not expect – I expected a lot of bad stuff to happen in the first few months, but not quite like this. What I thought was that it was going to be a much more systematic consolidation. You know, in a lot of ways – the Trump one administration for most practical purposes, the policy agenda was set by Mitch McConnell was very aside, aside from a, you know, a tariff here and there was basically a standard Republic bin. Let's, uh, let's impoverish the poor and cut taxes for the rich. Um, and I thought that this time around it, that wouldn't happen, that it would be Russell thought there would be the project 2025 and it would be a systematic purge of independent civil servants, uh, loyalty tests and a right-wing agenda. But I thought it would be kind of a systematic march forward and not this wild chaos that we're actually seeing. And so that's come as a surprise. I'm almost tempted to say an agreeable surprise because I think the chance that we lose it all is less because they've gone so wild. But the role of Elon Musk and The line I'm using is that the country is being run by an ignorant, megalomaniac, chaos monkey, and that Donald Trump is like that too. Terrible stuff is happening, but in a really chaotic way that I did not expect. Yeah, I guess on the one hand, so one difference versus 2016 is that it was fairly predictable that Trump could win or would win. So they had more time to prepare, right? In some sense, they're smart in this narrow sense, which is like, they'll probably never have more political capital and less resistance than they have right now, right? It's like, I remember when I got hired by ABC News, you have a big mandate, or ESPN then, for 538. It's like, let's hire all the people right now because before long we'll recognize all the challenges of the business and how it awkwardly fits in this big corporate media empire. So hire now while you have political capital and then it's harder to like repeal stuff later. But I do think, yeah, I guess I'm surprised. So I have this column called 113 predictions about Trump. So I've tried to be probabilistic and quantitative about things. The one I'm most worried about and some were smart. Like I was more, I thought it was more likely that we would actually try to implement tariffs on Canada and Mexico, granted for like only like 24 hours or something, right? But like people were skeptical of that, right? I have one where it's like, what's the chance that Trump will eventually defy a Supreme Court order? And he put those chances at 10%, which I now think is too low. The watch was that Supreme Court's pretty simpatico with Trump, right? And so if it gets that level, but I'm not sure I put it at 50%, but like, More than 10, right? And, you know, the fact that we are required to kind of place a lot of faith in the court system and in Trump listening to the rulings, and there probably are some weird, I mean, Ezra Klein had a really good podcast with Quinta, what's your name, Jerbeck or something about, like, what's it like actually looks like to defy a court order? There's an escalatory kind of spiral. It's a very interesting phenomenon, right? It's, like, not quite as binary as you might think, but, like, but that worries me a lot, that if we lose respect for the rule of law, then that seems like a very dark corner to turn down. I think 10% is too low an estimate of that risk. And I also think that, like, you know, Elon has both been more empowered and more kind of addled, I guess, than maybe I would have thought. I mean, he's someone where my opinion's gone way down over the course of a few years. But I thought he would get bored and that, like, Doge would not actually have any sensuatory power. But instead, like, They're pressing a lot of buttons. And, you know, I think it's going to go badly because, like, not to use too direct a metaphor, but it's kind of like if planes start falling out of the sky and you start seeing, like, measles outbreaks and you start seeing eggs become more expensive and you start seeing tangible things. Like, I have a friend who was a provisional, what do you call it, probationary employee. She was hired by the government about nine months ago. And so, of course, not a political role at all, but she gets... Let go. Right. And so I think people will see things in a tangible way. If there are Medicaid cuts, as you wrote today, people will notice that. And so, you know, look around and find out the phrase. But I was going to what I thought, you know, Trump ran. I'm going to I'm going to bring down the price of eggs and bring down the price of bacon. A lot of that was stuff that he had no way of doing it. He never had even concept of a plan for doing it. But I thought that he would spend at least some of his time doing showy stuff that at least appeared to be addressing the things the people, you know, the swing voters supported him for. And there's been basically none of that. It's been all, let's go after DEI, let's go for purges, let's do all kinds of crazy stuff. And meanwhile, and the stuff that's happening, it's already, I mean, I wrote today about Medicaid, but Elon is saying, and it may even happen, I think to some extent may be happening, that they're going to cut the staff of the Social Security Administration in half. And they apparently are already closing a bunch of Social Security offices. And with the staff, you know, the federal agencies are not generously staffed. They actually are all kind of running on scotch tape and paperclips. And to begin with, this is going to mean, you know, your Aunt Millie is going to try to fix, she's going to, her check is going to get missed somehow. And she'll try to go to the local Social Security office and find that it's closed. and that you can't get anybody on the phone. And so we're already pretty much close to the point where people are not just going to say, I thought he was going to do something about the price of eggs, but what happened to government services that I was counting on? So that's very fast. I mean, we're five weeks in. This should be a period when he could be running entirely on smoke and mirrors and the belief that stuff was going to happen and instead we're already seeing actions that are going to directly affect people's lives adversely and there's a lot more of that to come yeah i tried to write down a list of things that trump has done so far that i like i like ending the penny i like restoring paper straws or restoring plastic straws right there's some stuff like i think on um There's an executive where they might ban gain-of-function research that I think is probably good. But there's other stuff. I think Trump has a mandate on immigration stuff and some of the DEI stuff. So it might not be my cup of tea, but that's what people voted for, right? But dismantling the size of the federal government is kind of this fairly radical project that people didn't really vote for, right? And I guess it's the most predictable phenomenon in the world, but you would expect a backlash, especially in an era where Democrats... have this electorate that turns out in the midterms in special elections, I mean, it seems like... And you don't need a message for special elections, right? The other guy's bad for midterms works perfectly fine. For 2028, then, okay, we can have a discussion about that. But I don't know. I think, at the very least, Democrats can concentrate their angst into election wins beginning fairly soon. There's special elections starting in May, I think. Yeah. And the... Yeah, people definitely did not vote for, you know, slashing the federal workforce. I mean, people have no idea how big it is, but they do like the services that the workforce provides. And they did not. And I don't think the voters care at all about DEI. And yeah, I agree. The penny, you know, if you think about what the purchasing power of a penny is compared to what it used to be. And it is kind of silly, but that is not going to win any votes. And I'm actually surprised at the absence of any kind of gesture towards kitchen table issues. I mean, even Viktor Orban tried to put price controls on groceries in an attempt to assuage the public. Now, it went about as badly as any economist would have predicted, but even Orban... felt that he needed to address the things that people voted him in for. And Trump has done none of that. And instead, and it's, I also wonder exactly, I mean, there's a lot of wondering what Putin's hold is, but what about Musk's hold? I mean, if you were paying attention at all, I think Trump is not. I mean, Doge issued this list of the big savings it's made. which, first of all, fell far short of its claims, but also then turns out, well, they accidentally counted $8 million as $8 billion, They triple counted another, you know, and they issued a revised list that is still full of errors. But this is the kind of thing that in the business world, you say, oh, my God, get this guy out of my organization. In fact, don't let him anywhere near my pets. I mean, and and yet he appears to have as much influence on Trump as ever. And I don't quite understand that. I mean, I think it's probably because Trump is kind of disinterested in the details. of governance right i'm sure he feels very validated by winning the popular vote and overcoming i mean you know the narrative arc is a remarkable comeback story in some ways but i think he likes the ceremonial the head of state not the head of government aspects of power and like if i don't know um the version of elon from five years ago or you're talking before what if what if michael bloomberg had become like conservative pill and we're playing the role of elon musk or basil i mean i think it would be more than marginally better, which is in part the way of saying that I think it's going more than marginally bad with Elon, who's a person who I think has remarkable accomplishments, but like has the combination of being poisoned by social media and power and whatever else he's taking. I don't want to speculate. I mean, that's, I think we've all, but the, well, I would say that even genuinely great level-headed business people, are often terrible on policy. I mean, it's just a different domain. I think about Jamie Dimon, who undoubtedly is a great banker and has managed his bank extremely well. The one time he made a big foray into the policy arena, probably nobody but me even remembers this, but about 10 years ago, he was really big on how America can never return to full employment because our workforce does not have the necessary skills. And we promptly got back to under 4% unemployment without inflation. And so, you know, it's not the worst thing in the world, but it just was showing that there's different skills. There's different kinds of things you need to know. So I think, actually, Bloomberg, there's not a chance that somebody like Bloomberg would be able to get along with Trump. But Bloomberg actually happens to be somebody who was a great businessman and is actually not clueless about policy. He actually was capable of being an effective, you may not like everything he did, but an effective mayor of New York. But Musk is not. Musk has always been a kind of a chaos monkey, which is something you can get away with up to a point in Silicon Valley. I think he may have overstayed his welcome there. But Wow. I mean, and the sheer amount, I mean, and I'm surprised, just to add, I'm surprised at how quickly people are noticing as well. I mean, we're now, I thought that Trump was going to have a five, six month window of very high consumer confidence numbers because people were going to believe, you know, first of all, Republicans, as people say, Republicans cheer louder and boo louder than Democrats. So just because Republicans would all give him 95 percent approval on the economy, that there'd be a high. And instead, after basically two months after the election when people were feeling somewhat optimistic, we're seeing now both of the major measures of consumer confidence falling off a cliff. So it appears that the public is aware that things are not going well. Yeah, the University of Michigan number, I think it's consumer sentiment, I think, is like lower than it was during the last year of Biden. And Biden lost in part because of public perceptions about the economy. And so, yeah, I think you and I is more experienced political observers know that if you're like us and we're following every headline every day, then, like, there's a lot of stuff. People aren't really processing it, though, on a day-to-day basis, right? The fact that, like, I mean, types of people rating is not awful. It's gone from plus eight to plus one. But it took Biden until, was it July or August, before he turned negative, right? Obama went from having an 80% of people rating in the ABC poll to having a very bad midterm. And so this is happening... pretty fast. Also, Elon Musk's ratings are way down, right? You see in Europe, I was looking at this for a blog post maybe to be written later, right? In Europe, he's like a net 60 negative approval rating. And so therefore you see Tesla sales way down in Europe. But like people, you know, look, people had a lot of pent up anger and anxiety at the Democrats over various things, inflation, DEI, immigration, et cetera, right? And like, resentment is a fuel that burns for a long time, right? So, you know, and will burn for some people right through until 2026 and 2028. But like, again, when people see tangible impacts, right, this is not abstract stuff. One of the things I always thought was difficult for like Biden is that like, because January 6th was like this near miss, it almost went practically wrong, but like Biden was ultimately sworn in. This is stuff people are going to see the impact of, or if they don't, then more credit, I guess, to Elon and Trump. But like, this is stuff that will notably impact people's lives. Yeah. And one thing that kind of bears, you're talking about European views of Musk. What, one thing that I think that Trump definitely doesn't understand, and I have no idea what Elon understands, but Trump does not understand how big the world is and how different other countries are, particularly that politicians in other countries are not like Senate Republicans. They are not easily bullied into submission. These are real countries with real leaders, real voters, and a lot of economic weight. So when I see him, well, even the... We're going to find out next week whether... whether the tariffs with Canada and Mexico are on. And it's still not clear to me exactly why they were put off for a month. Did somebody actually manage to tell Trump convincingly that this would be a catastrophe for the U.S. auto industry? Or was it just that he was impressed by how tough Scheinbaum sounded? But if he goes after the Europeans, he's going to discover, you know, there's... There's 300 million people there. Their GDP isn't much smaller than ours is. You have the incoming chancellor of Germany saying, my mission now is to make Europe independent of America. I'm not sure that Trump has any idea what he's tangling with out there. Yeah, that was one of those moments where it's very logical for him to say that, right? And by the way, I'm kind of jealous of Germany because they have like seven parties that occupy every part of the grid on conservatism. But like, but anyway, yeah, to have the new elected Chancellor Mertz saying my first priority is to make sure we're more independent from the United States, right? This is probably the most important country in Europe. And then other like little things like Canadian hockey fans, like doing the Star Spangled Banner. I grew up in Michigan, so I'm like kind of like you know, proxy Canadian almost like that was like, okay, this is kind of funny, but also not funny, right? If we're making an enemy of Canada, which is one of the most successful democracies in the world, I think in various ways, right? Um, What are we doing, right? I mean, if anything you want, if we're worried about like, oh, maybe we should take over Greenland or whatever. Okay, well, Canada is much bigger than Greenland. It's really probably pretty nice to have Canada and Mexico, you know, these two of the longest borders in the world. They're both allies. They're both amazing countries. I mean, that stuff, like, I don't know. There are a lot of different avenues. This is part of the point for people to get, like, annoyed by Trump. What I'm going to say is I go back, you know, think about sort of, you know, when John F. Kennedy was president, the U.S. was the overwhelmingly dominant economic power in the world, overwhelmingly dominant military power in the world. And yet we took great pains to treat our allies with respect. We set up international institutions that at least formally put everybody on an equal basis. We know it was as if. Denmark or Canada were equal partners in NATO with the United States. Now, of course, the truth was everybody understood who was kind of the boss, but we made a real point of preserving the dignity. And now you have us pissing off everybody. You have Trump saying yesterday that the European Union was created to screw the United States, which is we actually helped create that thing because we thought it would be a valuable ally. So when you have that, you just wonder, this is not going to end well because we do not have that kind of, we don't have the liberty of being that crass and rude to that really big world out there. Trump sees everything as a zero-sum system. relationship right and if you if you study game theory talk about a lot in my book what you find is like actually truly zero-sum problems are more the exception than the rule right there are more problems of coordination and cooperation where there are both rivalries and there are also reasons for cooperation and trump i think just i don't know if it's the real estate business or his background or what right but he just doesn't get that even though tactically he's kind of like effective in certain ways based on these wrong presumptions like he understands things like deterrence, I suppose. But like, why are we making an enemy of Canada? It's kind of... Yeah. I mean, it's a real... A lot of people have been... Actually, I guess the latest cover for The Economist is Trump as a mafia don. But I actually have been pushing back on that because at least of my understanding, which I have to admit is based mostly on reading Mario Puzo, but still... is that mafia dons at least believe most of the time in keeping their promises. And a promise from Trump is, you know, has a half-life of about 30 seconds. So who can trust us? Who will take our guarantees? Who will offer us anything that isn't sort of taken at the point of a gun or a tariff? No, I think, you know, parties have this, have you read that, like, revolt of the public, like Martin Gurry stuff. No, I haven't. Ezra Klein had a good interview with him and I don't agree with everything. Right. But like he was talking about the conditions, why it becomes more challenging for incumbents now. And part of it is that the media empowers lots of information, including incorrect information, right. To, to spread, but like, but the consequences people though, when you break your promises and, you know, fundamentally Biden promised they returned to normalcy and for various reasons, people didn't get it right and like trump kind of made the same in a different way i mean he promised revenge on the libs right and i guess people are getting that right but also promising like a return to normalcy and um and people aren't getting that either i think people are are more burned out than they were before and maybe emotionally it was a different thing to process in part because the election wasn't really much of a surprise um but yeah why why do you think i think you wrote at your newsletter a couple days ago that you were surprised that Wall Street hadn't reacted more strongly. Now, within the past week or so, the S&P is down 2.5% or something. I think the consumer confidence numbers have some impact. But what's your theory for why you haven't seen more reaction from Wall Street? Okay. And I don't actually have a good theory. And I did do an interview with Jim Chanos, legendary source seller, although he lost a lot of money. betting against Tesla, which I think in the long run will turn out to be right. But unfortunately, the law, you know, Keynes never actually said the market can stay irrational longer than you can stay solvent, but he might as well have. But and his his he talks a lot about how once the market has formed a view, it's very reluctant to drop that view, the cognitive dissonance. And he had this specific example, which I had real fun with that. And I did sort of a second follow-up substack on it. One of his biggest success stories was short-selling Wirecard, which was a German thing that was entirely fraudulent. And the thing is that the news, all of the information about his fraudulence was in the press for about two months. And yet the stock just hung in there until at some point, actually, it wasn't until the company itself admitted that we were a fraud that the stock plunged. And there's just a lot of people went in for the Trump trade and are not yet willing to admit to themselves that they got it wrong. And that's the best I can come up with. I don't I don't think there's a good story. that says that there's that there's hidden good stuff in the trump agenda that's sustaining stocks despite the chaos i think it's mostly just that people are not yet really ready to admit just how crazy things are i mean you see the importance of narratives to the street you know i play poker with some of these guys i know some wall street types right and they kind of have these stylized stories of the world particularly of politics right i mean you saw with deep seek where um There was one weekend, I guess actually the first weekend after Trump was inaugurated, when there's a big fall off in NVIDIA, all these tech stocks, because of concerns about China could build AI more cheaply and undermine AI hegemony for the US, right? That information had been known for a few weeks, right? But it's just like, oh, we're in the weekend, it metastasized a little bit. And so you have this massive sectoral sell off. And so I think that story of stylized narratives is right. I mean, look, being good for the economy, right? is not the same as being good for corporate profits, right? And so I'm not as certain whether, you know, I think maybe markets might say, well, Trump seems to back down from the terror threats ultimately, but I don't know. I mean, you know, the other thing is like, I think there's a story that you hear kind of centrist, I guess I'm a centrist, tell where like, well, markets will discipline Trump, right? The Dow will fall two and a half points if you're a few days in a row, right? You see the downward arrow on CNBC and then he'll, but like, I think that overestimates the agency of markets where it's not a person who's behaving strategically, right? It's a collective emergent behavior. And so like, you know, to get panic, you might have a real panic. You might have a real big sell-off and you can't like kind of faking panic doesn't work. I don't think, right? You're going to have real panic, which is real risks or, you know, businesses talk so much about uncertainty. This is a big thing under Obama. Well, okay. There's a lot of uncertainty if you're planning in many sectors in the future. Yeah, I'm thinking about that, actually, writing about that more or less as we speak, which is that you're a business trying to choose between two investments, one of which makes sense if Trump really does go through with the tariffs on Canada and Mexico, one of which makes sense if he doesn't. So which of those do you do? And the answer for the moment, I think, is neither. You expect to see a lot of investment plans put on hold waiting for some clarity. Something that Keynes actually did say, I think, was that in investing in many contexts, it is better to be conventionally wrong than unconventionally right. And at the moment, betting that things are going to be really terrible under Trump, even though there's a really good case to be made for that, would still be unconventional. And so people are not ready to do it. But the resilience of the markets is, as I said, it's a self-denying prophecy. As long as the market hangs in there, the craziness will continue. I do think some of the AI... I mean, this is why, if you literally ask me, recently I... have gone longer in the markets on things that are adjacent to AI. Maybe we'll talk about AI a little bit, but that's really boosted the market quite a bit. The semiconductor manufacturers are up 50% to 65% over the past year versus 17% for the S&P 500. Google and Amazon are doing relatively well. If you believe there's like Tyler Cowen, the economist George Mason thinks GDP boosts 0.5% per year from AI, which the AI folks thinks is way too low. But like, I think a rational assessment like that, I think actually can counteract drags on the economy. Well, I would say, I mean, that would be in this vicinity of what the internet did for about 10 years. You can actually kind of see even a little bit bigger, a roughly 1% boost in, I can't talk English here, total factor productivity. But, you know, the efficiency of the economy actually did get about a decade of accelerated growth, which then faded out in the mid-2000s. But the, and AI could be, on that order. It's unlikely to be, you know, it could be the singularity and then Skynet kills us all, but it could be the, but more likely it's sort of a medium-sized technological boost that lasts for a while. But there's a kind of separate question of whether that justifies the valuations. And I think that depends a lot on which companies we're talking about. The semiconductor manufacturers, yeah. because they're producing the stuff. I do not understand why this should be a positive for, let's say, Google. Because what Google is doing is they're spending vast sums on what are basically defensive investments they're spending to defend. So not that this may enable them to establish a profitable monopoly position. They have a profitable monopoly position. What they're doing is spending a lot of money to make sure that somebody AI-based doesn't come along and take it away from them. But this shouldn't be adding to the value of the company relative to before. And so, you know, or another way of saying, you know, NVIDIA is the, uh, is the guys selling picks and shovels and whiskey to the 49ers coming out to California in search of gold. And that might be a good investment, but I'm not sure that the guys hoping to strike gold are a good investment. And I don't think the market is differentiating between those roles. I chuckled at the kind of Skynet reference and the singularity. It is a weird moment because there are credible people who actually think that's what's coming with some non-trivial probability. You know, one of the things I rest with a lot in my book is like, because we've talked about kind of the empirical view of like technology is usually disruptive, but good for people. In the end, it takes a little bit longer to adapt than people assume, right? You know, so if you take the baseline view, then the idea that like AI could be an existential risk might sound absurd. At the same time, the subject matter experts are pretty convinced that this poses a real threat, that this is not like other technology, or at least it's the most sort of important technology since mankind, the Holocene started, right? And mankind starts to dominate, and womankind, its environment, right? And after spending three years in this book, I am still kind of agnostic between this kind of narrowly empirical view and the AI, I don't know if you call it futurist or alarmist or doomer view. Weirdly, I mean, both the optimist and the pessimist share a view about the transformative nature of AI often. How much do you use like ChatGPT or other large language models? I do it not at all. And for, and it depends, the things that I want to figure out, I kind of know what I'm doing. And ChatGPT does not, even though it has this vast database, it's, I've, You know, I've had friends take LLMs to summarize some of my own articles. And they fantasize. I'll have an article that is, you know, it'll bring in, you know, Brexit or something when I never mentioned it and it clearly wasn't relevant because they're just sort of scooping up this stuff and going by correlations. And so far, at least, I haven't found any situation in which I do better. I actually... uh, often start my searches with minus AI to, to turn off the AI part of the search because it, it ends up being more, um, harm than good. Now that may make, you know, what I, What I do, what you do, I don't know if you're using these things, but what I do is clearly not typical. And there's a lot of stuff that, you know, one thing I say is, look, some people say that this is really just souped up autocorrect, which, you know, is not totally unfair, but a lot of highly paid jobs are basically souped up autocorrect. Yeah, I was talking to, I guess, cognitive neuroscience, He studies how the brain works in very mathy ways at University of Chicago, right? And he's like, we don't know if AI thinks, but we don't know how humans think or how our cognition really works either. I mean, if I were able to give you a homework assignment, Paul, I would just say spend, first of all, use the latest version, use 01, right? Because you're somebody who I think like me is both gifted verbally and gifted mathematically, right? I guess we're bragging here, right? But you are, dude. I think if you spent more time with the current versions of the models and just maybe not for work, just kind of played around, you would be interested in the way it kind of matrices all this information and language. And I think it's a pretty fascinating technology that it literally did kind of begin as autocorrect, right? It's like kind of like, okay, I am going to the blank, the store, right? The market. And then somehow these transformer models seem to have an incredible amount of power. They're kind of like the Ozempic or whatever of technology. But I think if you spent, I think they're pretty cool and impressive. Like I had AI give me, I was flying from an Asian vacation over January, right? I'm like, give me a tutorial on how to distinguish Korean from Chinese from Japanese characters, right? Give me a quiz. I mean, that's really cool, right? Because it's really good with language type stuff and translation type stuff. Actually, where I use it is, the one place where I actually do use it is books. If the AI summary of what book reviews say is, you know, nothing critical is hinging on getting it right, but it will summarize. You know, there may be 500 reviews, uh, of a book and the AI summary gives me a, it doesn't say this is a good book or a bad book. It says some reviewers said that the, the fundamental concept was interesting, but that the pace was too slow. And that's, that's useful information, but you know, it's, uh, and again, even if it turns out that we're nowhere near artificial general intelligence, we're nowhere near robots replacing people and, and, uh, and killing all of us. Um, um, saying that it's a souped-up version of, I mean, we already had sort of, you know, translation became, machine translation became actually usable a few years ago. And the, you know, you could, like I say, you could say that heavy earth-moving equipment is just a souped-up version of a guy with a shovel. But in terms of what you can actually accomplish, it really is a qualitative change. Okay. What if AI, because I have this model that's very simplified, right? But say, okay, divide jobs into 25% quartiles, right? And one quartile AI completely displaces, one quartile AI displaces all but the top performers. One quartile AI is mediocre. It kind of is replacement levels term we use in baseball. And then one quarter, it's totally unable to do maybe more physical tasks, right? What's the economy look like if AI is Very impressive, but lumpy and not great. I mean, what's the equilibrium look like? Well, I'm not sure there's an equilibrium, but there can be a long period. I mean, if our notion is that improved technology and higher productivity is always good for everybody, then you really haven't been paying attention to economic history. The Luddites were not idiots who hated technology. They were skilled weavers who had been displaced by power looms. In fact, the 19th century data aren't great, but it does kind of look as if real wages in Britain stagnated for the first 40 years of the Industrial Revolution. Britain became much richer, but the technology was replacing labor with machines. And so it did put downward pressure on the demand for ordinary labor. And in the long run, yeah, by 1870, the British working class was in better shape than it had ever been. But, you know, 40 years later, feels like a long time to people as they're saying that, you know, in the long run, we were all dead. And, uh, so the idea that, and we, we, we're getting some sense, it's just very preliminary, but some, you know, middle managers appear to be seriously at risk. Um, um, coding all, you know, just a few years ago was, you know, learn to code. And now it turns out that coding is one of those things that AI is pretty good at. And it's not a great job. Uh, I have no idea what's happened to translators, but that was one of the first things where you really got a transformational change. I mean, at this point, apparently, it's actually good, but good enough so that you didn't really need to get a person to translate stuff. That's been around for a while. And so, no, major displacements, maybe. But that's not saying that the price of of Google or even of NVIDIA makes sense. So there's a big difference between, you know, stuff that can actually make a big difference to the economy and stuff that actually justifies a PE ratio of 137 or whatever. If you were talking to a rising, like, graduate student, the next Paul, right, a very bright person, age 28 or whatever, how would How would your advice differ on how they should say they're pursuing a career in academic economics, right? How would, how should they be thinking about AI? Oh, that's a, I hadn't thought about that at all, but let me, let me see to my parents, come up with it. I think what, um, what AI is, is good at is summarizing and synthesizing, uh, lots of data. Um, what, um, Well, we've been advising now for actually for several decades now. You know, there's been a long, for a long time, been telling people, do not do what I do. Because I grew up, you know, I grew up professionally in a era when clever little models, where you took a few stylized facts that seemed puzzling and tried to put together a story that made sense of those facts. When that was a really productive enterprise, but, you know, that That seam has been pretty thoroughly mined out at this point. And for quite a long time, it has been work with data and preferably not with the same data that everybody's already, you know, if it's on Fred, don't use it. People don't know, Federal Reserve Economic Data is this incredibly useful website that anybody doing macroeconomics instantly turns to. But if it's on Fred, then it's probably been worked to death. So find interesting untapped data. Find stuff about Indian railways in the 19th century, there have actually been some economic prizes handed out for that. The fine stuff on very micro level stuff, natural experiments, just, and those, finding the data is going to be, I think, up to many people. who this is still going to require human judgment to know what might be an actual interesting place to look, which often involves actually knowing some history, among other things, knowing something about other countries. But then, at least for a first pass analysis of what's in that data, I think that's where AI might actually pay off. Yeah, I think ground truth becomes very important in a world that's dominated by kind of synthetic data and and facsimiles and ai right that if you have tangible evidence of something real and that trust is low this is maybe more metaphorical than it needs to be but like but yet any original data that's verifiable and it like is not sympathized synthesized by claude and chat gpt and whatever else i think becomes especially valuable yeah i don't know if you have a view but there is this question about whether the These models are trained on the stuff that's out there, and as increasing proportion of the stuff that's out there is stuff that's generated by the AI models, do we have the AI apocalypse not in the form of machines coming, but the machines basically choking on their own nonsense? Yeah. Look, one big question. I think within five years, some company will declare that we've achieved AGI, and I'll do that partly for PR reasons. Okay. Can you extrapolate beyond the data set? Can you self-improve when you only have this much training material? I think that's less clear. It does seem like these models deteriorate in performance until they get a reboot. Although, one thing I would say, there is a lot of human capital going. If you talk to people that work for Anthropic or OpenAI, these are people that are very well-trained and bright. They're poaching people from academia. They're poaching people from the private sector. from other countries, right? There's a lot of talent there. And that's one reason why I wouldn't short it. You can debate the leadership talent, but like the engineering talent is like, are very good, thoughtful people. I'm sorry. Yeah. I don't know where you were going to go with that. But what I was going to say is, you know, what I see in it, what do I know? But is that in a way, narrow domain looks from, to me, like something that has a lot more potential for right now. That is, instead of training something on the entire universe out there, you train it on, to some extent, I understand that's what DeepSeek does, but it's just, you know, we're already talking, my wife and I have one of the best-selling principles textbooks out there, and we're already talking about AI, you know, Students can't ask us questions about the material, it's too many, but a sort of AI version of Krugman and Wells that answers your questions, that is basically just trained on the textbook, that becomes a real possibility. Yeah, the queryable nature of AI is what's interesting, right? It's where it has, you know, by the way, if you ask AI about AI, it's pretty smart. about itself actually but like the query building it's never going to replace like you paul right because you know it kind of smooths out all the spikiness that make people interesting and kind of creative i think it's interesting by the way that like it might be the ai language models are with qualifications very good whereas the ai image models are quite shitty right i think that's interesting and maybe you know maybe is a a Proof that it's going to be kind of lumpy and spiky where AI is good and where it isn't. But I don't know, spend more time with ChatGPT. It's come a long way and it's pretty cool. And do you want to talk about some media stuff now? Yeah, let's talk about media because we're already, I think we should probably try to hold this down to an hour if we can. Media. I don't know. Why did the New York Times not want you to publish a newsletter? Yeah. I still don't really know. I mean, it was really strange. I mean, I thought that, you know, so what actually the story for people who haven't read it, I was getting frustrated. I used to have a blog via the New York Times, which was great, but they terminated standalone blogs in 2017. And Twitter threads were a really poor alternative. And I opened a Substack account in 2021 just to have a place to put the technical stuff. They freaked out totally. I said, well, I need something. And so I got to do a newsletter. And the newsletter was far more constrained than what I'm doing now on Substack. But it was at least something. And then a couple of months before the election, they said, well, we think you're writing too much. And so... And what we decided to do is to put your newsletter on ice, which is really weird because I think the newsletter, I'm not sure, you know, they would never tell me how many people I had, but I think it was a pretty popular newsletter. And then, and I hadn't, I can't tell how much of that was. I really, I mean, I have theories, but they're, they're, they're, they're not very nice to the people involved and maybe I shouldn't share that. It was just, it was just pretty weird. and that they would take somebody who clearly was, you know, I think I was contributing something that nobody else at the Times could do, which, you know, there are plenty of people that are doing things I can't do, but for them to can that was a very strange thing. And basically at that moment, I decided I'm going to last through the election and then leave because it was clear that they just did not understand know what i was trying to do and what i was bringing to the table so uh but it was a very odd thing now you had your own experience i i have i do have a theory about a problem that media has but uh i'd like to hear your your take first um what happened my experience is that some of what i was doing was running counter like the political reporters on the times news desk in 2012 which is i think a less sophisticated era than now i think the times's coverage has improved right But I'm kind of saying all this stuff is silly because Obama's way ahead in the swing states and it's been a very stable race and all of this is noise and you're making something out of nothing, which kind of undermines their message, right? And then when the election is over and the contract expires, there's a leadership change where Mark Thompson picks over, who's now at CNN. And whenever you were brought in by the previous regime and you were an achievement of the previous regime but a controversial achievement, then – that's going to go badly right they offered a contract extension but it was kind of a little late i mean i don't know i mean were you eyeing but the times is weird like they would get they would have random like there was some column about like that was very innocuous comment about like john huntsman they vetoed for some reason right i wrote something about poker whereas other times they were like very free but like the it was kind of this random editorial hand that like you sense that there are internal conflicts they're that are opaque to you that they're trying to preserve in the building. And it's, it's kind of like randomly once in every 10 things we get veto. And the other times it was fine. Sometimes they would put headlines on things I didn't like, but like, do you think they wanted to like move away from like a certain style of like resistance, anti-Trump or what areas were sensitive for you? Well, certainly, I mean, it was not just carrying the newsletter. There was also a lot of editorial intervention on the columns, which had never happened until my final year there, always flattening it and both sizing it. But to the extent, if I thought it was just about, well, I'm too controversial and too obviously liberal, then how do we explain? I mean, they had two economic newsletters, me and Peter Coy. And Peter is great, but does very, very different stuff. What he does is a lot of here's something interesting that's happening in the economy or the business world and makes it a certain news you can use feel to it. But and certainly not something that would offend anybody who isn't determined to be offended. And they I mean, I left. He was let go. And I which totally struck me as bizarre. So I have, let me give you my, since we don't have that much time, I, aside from, you know, the ideological stuff may be playing a role and we have this horrific, I mean, Jeff Bezos has just killed the Washington Post opinion section and, and possibly its reputation for all time. But, but there's also, there, there's a middle ground between this is my opinion and here are the facts that I think is really undercovered and which I could not get the times to understand. I mean, we had, you know, I, when this, the newsletter thing came in, I was, I was shocked and probably inarticulate in my immediate response. But I said, you realize that I'm pretty much the last person you have on the opinion page doing policy. And they said, that's not true. And they rattled off several people who express opinions about policy. I said, no, that's not what I mean. I mean, they don't I'm the only person who knows how to read a CBO report, you know, and they just did not understand the difference between. Now, they have economics reporters, but they are very much constrained. by the convention of, you know, the voice from nowhere, the no, no opinion. So instead of saying Trump's terrorists will have a disastrous effect on auto manufacturing, the best they can do is say some economists say that Trump's terrorists, you know, and, and they, they just, the, we really need some, you cannot do decent, um, analysis, economics, I know, but I think in lots of other fields without having a point of view. And they seem to have an inability to get at the idea that you can have people doing that, that yes, people have opinions, but also know their stuff. And instead you just get people who have opinions and people who know their stuff, but have to be very, very careful in expressing it. And that leaves a vacuum. No, what I love about newsletters, um is that each newsletter should be more than the sum of its parts right you see a point of view and some things are some things are more takes some things are more hands-off right but it's part of a whole and you're able to kind of use your voice and kind of train i guess i don't know that term the audience but paul i assume that you i know you write well i assume you also write pretty fast yeah that's uh um I was the fastest writer in academic economics and just, and when I started doing journalism, I was wondering, okay, how do I compare with journalists? And it turns out I'm fast by their standards too. And if you ask the question, how to, why? And the answer is, I have no idea. Uh, but it's, uh, but yes, I'm, I'm a fast writer and it's, um, which can be overused. I sometimes think, you know, it's this old line, Napoleon would have been better off if he hadn't been quite so good a general, because then he wouldn't have tried to invade Russia. So sometimes I feel like I overstretch because I'm counting on fast writing. But yeah, I don't know about how you feel about it. And I think that, you know, there may have been You know, I have lots of stuff to say and I'm able to write fast enough to say it. And the Times didn't want that. And now I can. And yeah. And it's one of the things, by the way, the one that I got a lot of pushback when I tried to follow up. You know, I wrote a column and then there was more that needed to be said. And they would be, oh, but you've already written about that. As if, you know, one standalone 800-word piece should be the last word you have on a subject. And that, fortunately, I don't have to, I'm not in that position again. You know, every PR person in the world understands that sometimes to sink in, it's the third time, it's the fourth time, it's when you say it a slightly different way, it's when you kind of meme-ify it. a little bit, right? And also, you know, people love you. I mean, with a newsletter, you probably get, I hope, people coming up and saying, hey, I enjoyed that. It's very personal. It's in people's inbox, right? But you also get people saying, oh, you've been writing a lot and I haven't read your last five things, right? And so it takes a while to sink in. But you're kind of, I admire the fact that, I mean, you were, when you were brought into the Times when it was like 2000-ish? I started on January 2000. Was there anyone like you who was like a serious, yeah. They reached out to you and they reached out to them. How'd that relationship start? Well, it was actually kind of funny because the Howell Reigns, who ran editorial opinion at the time, said, we have five guys writing about the Middle East and nobody cares about that. Turns out it was in a really bad call. But we have nobody writing about the economy. And they were looking for somebody who knew economics who was not a terrible writer, which is actually a pretty small intersection on that Venn diagram. And I'd been writing for Slate and Fortune, so they kind of knew that I was capable of writing more or less comprehensible English. But it was very different. And it was – actually, one thing that happened is to go on to it. But they had no idea what people actually – cared about or writ at that point because it was still mostly a physical newspaper and they've only really started to get digital tracking of what people were paying attention to about around 2003. And I had been getting a lot of grief because I was an early we're being lied into war guy. I was pretty much the only we're being lied into war guy on a major editorial page. And management wanted me to stop and I offered to quit and it was a pretty stressful time. And then by the end of 2003, they found out how popular my writing was. And all of a sudden, although it's not always encouraging. It's great that controversial, interesting opinions get a lot of readers, but it's also true that if you look overall at what people really care about as judged by views, the answer is food. But- Yeah, no, I mean, it's like food and Wordle and the real estate section, the luxury homes in the Hamptons, like those pay a lot of bills for, foreign correspondents in Sudan or things like that. And I mean, you know, that is one thing. So people like you, Paul, are probably being underpaid relative to your marginal revenue product for the New York Times is a nerdy way to put it, right? Because you see that it's very power law, top heavy, right? You know, columnists don't require, you know, spotters, they go abroad and things like that. They're, you know, they're just people that you're sending a paycheck to. Do you worry about, like, I mean, on the one hand, like, it's nice that if you don't like the editorial direction of management, and ironically, in this area where Trump's elected, you see a pullback, I think, sometimes for more cynical reasons. I think, you know, Bezos is a different case than the Times, where I think they feel more like, oh, we kind of did a certain type of anti-Trump blogging last time, and now we're moving on. And believe me, I think the Times is great. I freelance for them sometimes. And it's a better run company. I've talked to every media company at some point. It's in the 90 plus percentile of how well-run media companies are. But still, I mean, you, I guess you and I, are empowered by having this alternative in sub-stack. And I'm not sure that what they did was a bad business decision. I mean, the Times in particular, years ago, they told me that their business strategy was last paper standing. And to some extent, they want to be the newspaper for everybody. And that I'm not sure that that's actually a sustainable position, but it's at least it makes sense. And OK, some more controversial figures. And there's also there's a lot of internal politics going on in ways I don't understand because I've never worked at the Times, only for them. Right. I never had an office there. But the but yeah, there's this there clearly is a big demand for stuff that that the legacy media organizations, for whatever reason, don't feel that they can supply. And so look, it. The the big concern that I had, I'm I'm 72 years old. I'm we live frugally. We have one of the best selling principles of economics textbooks. So money has never been a concern. So I was never worried about about losing the job, but was worried about not being able to reach enough people. Yeah. And no building there. And it's, you know, I try not to check my subscriber count more than twice a day. But that's the main thing is, and what's happened is there clearly is a strong, you look at some of the other, you look at Heather Cox Richardson or something, you look at people who are just giving feedback information in a way that legacy media just won't. And I mean, in the end, assuming that, I mean, with the reporting, there's still nothing quite like having a guy on the ground in Lebanon, right? And you or I are not going to be able to do that. And you still need these news organizations to do that. But it turns out there really is a significant number of people who want sort of analysis in a way, opinion as well, but mostly I think it's just raw opinion. This is what I think. Not many, rightly, not many people are that interested, but here's my opinion. Let me give you a bunch of charts and tables to explain why that's my opinion. There turns out to be a lot of people who want that. One thing I like about the newsletter too, is that like, I am basically just taking the stuff that I am thinking about naturally, right? Or, cause I'm just a curious guy, right? I, I, have some questions, I'll dig into something, get some data, start formulating some thoughts, outlining things. And that output gets used in a way, you try to edit your newsletters and try to think a little bit about the balance. But the fact that you're using the whole animal, right? And you're not taking these interesting Paul Krugman thoughts that for some reason some muckety-muck at the times was worried about offending somebody else and therefore the newsletter doesn't get published and so either goes nowhere or maybe it becomes like a passive aggressive tweet or something right it's like why why waste why waste a Nobel Prize winner's thoughts on important issues that he's put a lot of time into and there's what I'm finding so far now I'm only uh my, I took my thing out of hibernation, I think on December 11th. Um, so I'm still quite new at this thing. Um, but the, um, uh, I have not yet had a single day where I wondered what can I write about? It's, uh, there's just so much. The, the hardest part has been finding the, the, uh, the musical coda to end the pieces. And, uh, the, you know, the, uh, but, and, and, and, and, Although, yes, your freedom, there's no space limit. It's not 5,080 characters, including spaces, which turns out to be the actual length of a New York Times column. But the risk of overloading people with just too much detail is always there. And so it's a lot of pruning. And I have to send my wife is very good at saying you really did not need to include that chart. You know, you really did not need to include a reference to the learners' macroeconomics. But it's, yeah, there's a lot of stuff that, and one last thing, low production values. I find to be a real virtue. I mean, one of the problems of writing for a place like the Times is that they insist that every chart be beautiful and not a screenshot from some OECD report, which is good enough for a newsletter. I have a friend who used to work in kind of like marketing and social media strategy for like a nonprofit media company that would fundraise, right? And you said that like pictures that went out that had typos in them, actually got more contributions right because it demonstrated the scrappy authenticity of this company as opposed to things that were like too polished i think there's something to like the lo-fi nature of of subsec where it's where it's where it's unpretentious right and it's just it's just a person's point of view now i think you're also in the very very high percentile again people who write well naturally and you know i just want to endorse the idea that like when you write more you flex that muscle more and you're immersed in ideas more and like You're never going to run out of ideas. You said you're 72. I'm surprised you're that old. You look younger, but you're never going to run out of ideas, Paul. If you wrote Substack every day for the rest of your life, you're never going to run out of ideas, and people have this terror that they're going to. The challenge is always that there's not enough time to write up all the things that I do think would make worthwhile posts. Yeah. I am working harder than I've ever worked in my life, I will say. I didn't quite anticipate that, but it's complete freedom. I'm working 70 hours a week, but I can do them whenever I want. It's not quite that bad, but it is. So it's been a great experience. And we probably, speaking of burdens, we probably should be bringing this to a conclusion. So no, I mean, I, I mean, I know you've done, it's obviously been liberating for you that, you know, you've had run-ins with media organizations and at the same time became, you know, so this was from a previous election and I'm sure it helped on this one, but, you know, you know, take two aspirin, read some Nate Silver and call me in the morning. But it's, it's a, it's, You know, if we get through this, if democracy survives and all that, I think we will be glad to have an enlarged media landscape. It's at least interesting. Give it that much, right? I'm not sure I'm sleeping this soundly at night for various ways. And I worry about, again, the collision between AI stuff and geopolitics. But it's at least good for content, I guess. But yeah, do you actually find... I guess this is the last question. Do you find it harder to, like, do you feel more restless now? Or are you kind of like, well, what can you do, right? No, I said, well, I don't know. My reference point. My last year at the Times was a nightmare. I... Mondays and Thursdays, which were column submission days, I woke up in high anxiety. What are they going to do to the column? And then spent the afternoon in a rage after what they did to it. So removing that has made me a lot more relaxed. It's just, I have to say, the pace I'm maintaining is probably, I will need to slow down at least some time. But but I'm not having a lot of anxiety because it's there's plenty of material. And, you know, I'm actually always happiest when the bad policies have to play right into my strengths. So, you know, tariffs, more tariffs. From a personal point of view, the worse the trade war gets, the easier my job gets. Okay, Paul, if you're a Silver Bulletin reader, I'm sure you know about Paul anyway, but please check out his, it's just called Paul Kugman's Substack, right? It's just called Paul Kugman. There's no dumb name like I gave mine, but obviously check it out. It's great. And Paul, it was a real pleasure to speak with you. And Silver Bulletin, also. Mutual recommendation. Okay, it was great to talk to you. Awesome. Talk to you soon. Okay.

SITE COUNT