D (2:53)

Amen stuff. Yeah, as you say, Ben, we spoke with David Kirsch in October of 2025 recorded that interview for this podcast episode. And we began by asking David where he was, what he was doing, what he was thinking on August 9, 1995, the date of the Netscape IPO.

A (3:20)

Well, so I was in grad school, I was writing my dissertation about the history of the electric car. And I was living in downtown Palo Alto, walking across the street on Saturday mornings to the farmer's market where the golden retrievers of Palo Alto were tied up on one side of the street and the, you know, exquisite Molino Creek tomatoes were handpicked and polished for everyone's election in the, in the little parking lot across across the street. And it, it was a big deal. People were really, wow, this thing is really happening. I had already been exposed to the World Wide Web at that point, as we used to call it, that thing www I, I probably had been using it since the fall of 94, I think was probably when I was first introduced to it. And I would go to the Stanford libraries and use, get on, I don't know, they had these weird like workstations. Maybe this was the next era of, you know, Steve Jobs Co. And I would log on to the Michigan Web service to look at weather maps and be just amazed at how, how powerfully the, you know, how quickly these, you know, sort of images would load and the color and you know, having been a very long time email user and even before that, you know, I'd used finger on UNIX in the early 70s. I mean this is going way back but, but this, this, I knew this was something weird and, and, and everyone was using Netscape. It was, it was the thing and you know, it was a big pop, that ipo really, everybody, there was a bunch of demand for it. And I remember at a, it was some event later that summer, I don't remember where it was, where people were saying, oh, I just need the Netscape to go up a little more. I needed to, you know, it was, it was, it was a kind of currency, you know, sort of like, you know, in, in, in 23, you ask people, what's your P doom? No one, no one talks about P Doom anymore, but because it's, it's, we're fucking here. It's doom.

B (6:21)

So were people like you buying into it? I mean, were your fellow grad students?

A (6:25)

I, I, I, I was not, I didn't own any stock at that time. I was a graduate student. I was in, I was, you know, probably paying off my, the loans on my computer for, on my little, you know, Mac duo that I financed so I could buy my wife an engagement ring, you know, just like. But I think, you know, I think it was, it was heady times. It was, it was crazy.

C (6:55)

And when did, what year did you go to grad school, David?

A (6:57)

I started in the fall of 91.

C (7:00)

Okay, so you were like four years in at that point.

A (7:02)

Yeah, I mean, I was dissertating and, and struggling and wondering, you know, as we all have, would I ever get a job? And, you know, what does it matter?

B (7:15)

But that's an interesting kind of swing of time too, you know, if you started in 91, that's kind of in the, in the recession and.

A (7:23)

Oh yeah, it was dark.

B (7:24)

But 95 was not.

A (7:27)

No. And it kind of popped and everyone. I mean, as. Again, by way of context, I grew up in Palo Alto. I was living on Forest Avenue. I'd grown up on Pittman Avenue, which was right next to Forest Avenue. I, I'd in many ways gone home for grad school, even though it was a very different. The place I came home to in 1991 was not the place I, you know, spent the late 60s and early 70s growing up in. So it was already very different. But, you know, 91, it was a little. It was pretty somber. You know, there was not a lot of traffic on, on, on the, on, you know, Bayshore Freeway, which is of course the, the true test of a bubble is traffic on Bayshore Freeway. And yeah, so it was. And by 96, it was crazy. In fact, the story I tell people is when we left our apartment, so we'd been in. We had this beautiful apartment in one of the old buildings in downtown Palo Alto. My wife was an attorney at Wilson Sonsini. You know, the, the, the handmaidens to every major IPO up till that point. And you know, the first real Palo Alto law firm. All the other law firms that came were sort of transplants from, From San Francisco. So we could afford this kind of nice, ish apartment. I mean, nice.

B (9:03)

It was a nice apartment. There's student salary.

A (9:08)

Yeah, yeah. I mean, whatever. As long as I paid the payments on the ring and the.

B (9:16)

So you pick the right time to buy the ring. If you're going on like, you know, how many months salaries of grad school is the time to do it.

A (9:23)

Yeah, yeah, for sure. It's. But you know, we, we moved in. It was this two bedroom apartment. It cost us fourteen hundred dollars a month, which seemed like a crazy amount of money.

B (9:35)

Yeah.

A (9:36)

Over the course of five years. They didn't, you know, didn't the rent didn't budge until the final year. They raised it to, like, 15 and a quarter. And we were, you know, shaking our fists, and. And then, you know, we moved down to LA for me to take this postdoc at. UCLA and the people they rented it to. So this was fall of 96. Paid, they went from 15 and a quarter to $4,500. And then we started getting. And so we filed the change of address form and all this, and we got some forwarded mail for a while, and then it stopped. And then we started getting these FedEx packages of our junk mail. And we were like, what's going on? And it turned out that the woman who had rented our old, you know, who moved in when we left was, and I kid you not, employee number three at eBay. Wow. And she had been a paralegal at Andrea's firm because she'd been sort of poached by, like, someone had gone from Wilson Sonsini to ebay. They'd then, like, hired her. And I forget, maybe it's not number three. So very early. Anyway, she's like, you know, kind of the. Basically, the staff at ebay, she had $90 million worth of options. And so she recognized our name on the junk mail and was FedExing us the junk mail, because of course, she could. So the 4,500 bucks was, like, didn't even. He couldn't even see it.

B (11:19)

It's amazing.

A (11:21)

So, I mean, it just, like, to kind of give you a. And. And that was within, like, 12 months of our leaving. Yeah. You know, maybe. Maybe two years max. Like, so. So really, we. We left right when things just exploded.

B (11:36)

So that, I mean, and that kind of gets us. Go ahead. Yeah. I mean, that kind of gets us around to what we were, you know, wanting to talk about with this. This. This bubble and the timing of it. And we. We started by asking you about the. The Netscape ipo, because you and others who have written about this have sort of placed that as the. As the symbolic, I don't know, beginning peak. I mean, how. And also maybe as a scholar of bubbles, how do you understand a particular moment and how. How do you say, like, this is the thing that mattered. You can tell anecdotes about FedEx and junk mail, but, like, how do you, you know, why. Why do we. Why do we fixate on something like the. The Netscape IPO in the summer of 95?

A (12:16)

Well, the interesting thing is, of course, we only see this in retrospect. I mean, this is why we need History and Kransberg's laws. Right. History of technology is the most important. I mean, we need this because you can't see it in the moment. I can tell you we're in a bubble, an AI bubble today. I can't tell you when it started and I sure as heck can't tell you when it will end, but only by kind of keeping the whole period in view that we have any hope of understanding what's going on. And so I think for, you know, the Netscape IPO is convenient. It's a convenient marker. So if you read, you know, John Cassidy's book, which was, went through a couple different names.

B (13:09)

This is Dot Con. Is that the one?

A (13:10)

Yeah, yeah, yeah. I don't think it was original. I think it was originally.com and then it became something else because all of a sudden everyone realized, oh, it wasn't such a con, it was kind of a cool thing, or maybe it became.com and started or something. Anyway, it had a couple names. But, you know, you, there's a, a kind of sweep to the period. So, you know, do you, do you market to Al Gore talking about, you know, how he invented the Internet in, in a Senate hearing in the, in what was that, 91 or the, you know, privatization, the plan, the passage of the plan to privatize the Internet over time, you know, or the actual launch of, or, you know, release of the World Wide Web, kind of an HTML protocols from cern. And you know, just there's so many things you could point to. And I think at least, you know, as a kind of scholar of markets and startups and firms, I was, you know, for me, that Netscape IPO was kind of, when all of a sudden the public had access to this technology.

C (14:36)

Oh, right.

A (14:37)

And so the theory of our, like our, our bubble theory is you need something to invest in or else you don't get a bubble. And Netscape was the first like pure play thing you could invest in that was just, here's this worldwide Web. This is how you, how you can buy into it.

B (14:58)

Yep.

A (14:59)

And before that, before that, you really couldn't like you going to what, going to invest in like IBM or a Compact or you know, what, you know, whatever else there was. All of those other plays were just gen. Generic tech companies that happen, you know, they were selling the generic picks and shovels, but not the actual sluice tool to separate gold from the, from the silt. Yeah, so that's, that's where I think where the Netscape IPO really kind of, it Kind of was the thing that people could look at. You could put it, you know, you could. This was a tool you were using every day to access this, this new thing. You were using it to get quotes about the price of the Netscape stock. You were, you know, so that's where I think it was kind of, it occupies a particular moment.

D (15:58)

So there are three things in what you said that make me wonder about terminology and sort of methodology and looking at these things, bubbles, in your assessment. One was looking back and saying, you know, like dot com, right? And you're looking back saying, oh, well, it, maybe it's not that bad. Right. Another is the AI bubble. You're saying we're in an AI bubble. I can't tell you when to start, I can't tell you when to add. I can only tell that in retrospect. And both of those things make me wonder about bubbles. Like if they don't pop, are they bubbles? Right. You know, we're in an AI bubble. But what makes it a bubble? Does the popping make it a bubble? And the not bad thing is the, the fallout have to have a certain damage that it does, or is there some more complex functionality of a bubble in sort of the arc of technological development? And the other thing that made me wonder is about the whole public thing, the distinction between public and. But I think that's for the other side of your address of sort of the terminology of bubble.

C (17:15)

Do you need a pop dab of bubble, David?

A (17:17)

Yeah, I think if it's just a boom, it's not a bubble, it's just a boom. But most booms over, you know, in most boom times, people get, you know, overinvest, they get too excited. They get, you know, people think it can do everything. They think, well, I want in on this, you know, I, I'm going to start my little AI company that's going to get squashed like a bug when chat, you know, when OpenAI rolls out that functionality. Oh, and there'll be a hundred of them because of course everybody's looking at the same evidence and doing the same thing. So I think, well, you know, I don't want to recapitulate the whole book on bubbles, but I, I, which I hope is over my shoulder. Yeah, it's somewhere back there.

B (18:10)

You can plug it, go it, do it.

C (18:12)

No, no, no.

A (18:13)

So, so, you know, we did write a book, Brent Goldfarp and I wrote a book on bubbles and crashes in, in Technological Change. And I talked to Lee about it on the podcast, you know, a couple Years ago when it came out. And we have a model and we think it's pretty robust and it certainly helps us understand the AI moment quite clearly. But I think there is always this overreaction and it's sort of natural because there's uncertainty. We don't know what's going to happen. The stories we, we lose our critical faculty. We can't assess necessarily the good ideas from the crazy ideas. And they're no experts because no one's been in this terrain, in this terrain before. So you know, your expertise is as good as anyone's. And I say it's going to do this. And if you can convince a couple other clowns to, you know, in the car with you, you'll get a. I mean, you know, the crazy thing is the scale, the scale today really is stupendous.

C (19:34)

It's larger than the dot com bubble a lot.

A (19:38)

Multiple orders of magnitude. Right. Pets.com, the famous, you know, kind of poster child of, of do Xs at its height, what the enterprise value of the, all the publicly traded, you know, of, of that company was $300 million. And if you figure that, you know, there are now a handful of AI companies that are valued at $300 billion and Nvidia is $4.5 trillion trillion dollars. We're looking at truly, you know, these, these are, this is, it's a big bubble.

B (20:21)

So David, you, you and I talked about this. I think a little bit happened Canada last, last summer, like in the, in the big sweep of economic history. How do you account for that? In, I mean is, is there just more wealth in the world today? Is it about inflation? Is this, you know, or is this simply that the, the bubble of the 90s wasn't as big relatively to the entire economy as we maybe thought it was at the time? Especially if you compare it to, to today.

A (20:52)

Yeah. Oh, that dot com bubble was so cute.

C (20:57)

Right?

A (20:58)

It really was a.

B (20:59)

On the end of your tongue, just a little.

A (21:01)

Oh yeah, there it went. Nobody got hurt because it was all, it was equity investment so people could, could afford to.

B (21:13)

Can you explain that for listeners? What, what, what equity investment means?

A (21:16)

You know, most of the dot com fund, those dot com companies were not leveraged. They hadn't borrowed money. This was not the, the, you know, I mean the housing bubble, 2000, I mean, you know, housing crash, 2007, 2008, that was so much worse because that was people's homes and they'd borrowed money and they borrowed money to do other things. So unwinding all of Those commitments ended up just, you know, as we saw, you know, basically destroying America and several other countries. Yeah, well, but destroying our politics in a way I would argue so much of we've totally, we're dealing with now is still the backlash from that and the dot com bubble. What didn't, didn't really hurt anybody except a few, you know, wealthy investors. These VCs, you know, they, they sort of. Okay, maybe a few. Yeah. If you were holding QQQ and it went from 5000 to 1100, you were bumming. But you know, where is it now? You're fine. So I just, these were. The leverage play at that time was actually in the telecoms. So you know, there were a lot of those Global Crossing, you know, the kind of Beverly Hills phantom startup that raised a bunch of money and built a lot of fiber that stayed dark for decades. You know, there were some real losses there. And maybe we could talk about, I don't know, maybe some, you know, people couldn't afford to, you know, put the addition on their Hamptons house because of it. But I just don't think that those, that the, the thing today is tech is a bigger part of the economy. It's a bigger, you know, a lot of, you know, a significant fraction of, of growth in S&P 500 in

C (23:18)

GDP. Like half GDP. Right.

A (23:20)

It's a percent. 1% of GDP growth is.

C (23:24)

Half of GDP growth basically is from capital expenditures around a data center.

A (23:30)

Yeah. So this is huge. And people are starting to lever, you know, this is not now just money that's being put in by, you know, the super sized funds like SoftBank and the Saudi, you know, wealth fund and whatnot. So the, we're all on the hook for this one for sure.

B (23:55)

Wait, is now the point where we can talk about Hyman Minsky?

A (23:58)

I don't know.

C (23:59)

Wait, yeah, let's say what that means.

B (24:01)

Do we need, do we need to educate Lee on Hyman on, on the Minsky moment?

A (24:05)

Go for it.

B (24:06)

Well, I mean, I mean, I'm not going to do it justice. This is sort of a seventies era economic heterodox economic theorist. But the theory of bubbles is that the, the tipping point, the, the bursting point comes when. David, correct me if I'm not phrasing this correctly, but it has to do with when you reach the point of maximum leverage, when you, when you people are borrowing to invest and the money that they're borrowing is, is contingent on the collateral of the thing that they've invested in. And it reaches a point that you can't predict ahead of time. You can only see it after where the, I guess is the borrowing costs just exceed what, what the, you know, the rate of increase. I'm not sure if that's an economically sound way to put it, but it's the moment, the Minsky moment is when, when the bubble can no longer sustain itself through borrowing. I see. How'd I do, Professor?

A (24:55)

It really comes out of the 20s. The, you know, in, in 29, the, you know, the finance, the sort of financial interests had actually withdrawn from the market to a certain extent and a lot of the borrowing was corporate and sort of non of, you know, these were non expert players. They were just borrowing, you know, the, the margin loans. And so it, it kind of became unsustainable and, and yeah, I think we might, I don't think that really ever happened with the dot com. I think what happened with the dot com is it just people realized that it was going to take longer than these valuations suggested.

B (25:52)

Yeah.

A (25:52)

And so then you know, if you look in say early 2000, the like Goldman Sachs Internet index starts dropping well before the peak in March of, of 2000. And I don't know, I may have shared this story with you guys at some point but you know, one of my historical collections is the, yeah. Is a company called Science, which was one of these I Builders. And the I Builder was this kind of company that was set up sort of down the street from the venture capitalist. So you, you know, you go in, you get your $15 million, you know, series A, you know, series B round or something from the venture capitalists and you go down the street and you sign over $10 million of it to this I Builder that's a basically a kind of newfangled consulting firm that builds your website for you. And what do they do? They bill out a bunch of English majors who've learned how to hard code HTML@ 200 bucks an hour to build you your, you know, state of the Art 1999 website to deliver dog food. Yeah, whatever that, whatever it is. And you know, I, I have this, I have the records Society failed. I collected a bunch of their records and what, what's really interesting is that Science understood that the urgency was something they had to manufacture if they couldn't.

C (27:37)

Oh, that's interesting.

A (27:38)

Keep selling the urgency. Your business will be disrupted. You might as well disrupt your own business. Someone's going to drive a Hummer through your business.

B (27:47)

That sounds very familiar, right?

A (27:51)

Yeah, these guys were and they, as soon as the kind of scales fell from the corporate's eyes, and they were like, wait a minute, this is urgent for you, but not for us. All of a sudden, the whole thing sort of just kind of unraveled and that's when the valuations fell down and that's when people realized, oh, actually, I don't, I can get dog food here or there and I could have it delivered, but as soon as they actually charge me the delivery price, it's kind of expensive. And, you know, stop spending and stop having VC money to subsidize looks totally different. So anyway, so I think that's when things kind of, you know, unraveled and obviously it was a big crash and 5000 to 1100 is a big fall. It's, you know, I think, yeah, we have a number in the book at 79% or something, you know, some 80 number. Yeah. Which, you know, hurt, obviously. That is a lot of wealth.

C (28:55)

That was. The NASDAQ fell 80%.

A (28:57)

Yeah, NASDAQ. But, but I think we're, you know, the, the issue, Joe, about the, the failure to.

C (29:06)

Me,

A (29:08)

I've actually really, I mean, you know, where are we? We're coming up on the 30th. We just passed the 30th anniversary of the Netscape IPO. We should have done it. Oh, yeah, like, holy cow, it was a great success. Like, look at the transformation. Look at all the things we can do that we can't, we couldn't do 30 years ago. All of those dreams, all of those fantasies of, of those, you know, e commerce entrepreneurs have come true and more.

C (29:39)

Yeah, Netscape's gone. This is something we wanted to talk to you about.

A (29:42)

But it's all. But, but those, those players, like cyant gone. Netscape gone. You know, but look what we're doing

B (29:50)

right now, we're in four different places having this conversation.

A (29:53)

It's right. It's crazy. So I think, you know, that's why the, the, the Valley folks are, are sort of pro bubble, right? They're like, give us one more bubble. We'll build something great. Yeah, we'll, we'll blow some things up. We'll, you know, some things will go south, some people will lose money, but in the end, you're going to get a bunch of cool shit and be able to do things and create value in ways you have not. You can't possibly imagine. And we can't even imagine.

B (30:26)

And that's a story, David, that goes back to like the railroads in the 19th century.

C (30:29)

Right?

B (30:29)

I mean, this is not a new phenome phenomenon.

A (30:31)

There's a lot of bad railroads that were built, right? Rail, non economic railroads, but who cares? You know, they, the railroads built like a few people, you know, some bond holders got, you know, foreclosed. I mean or lost their money. It doesn't matter.

B (30:47)

Those, those were again deprec panic of 1873, right?

A (30:52)

Well, no, I think. But. But I'm saying the people who were the principal investors in those ventures were not the ones who were harmed. Of course. Right. If you were again, maybe the Hamptons wasn't. You know, if you're the Fricks and you're living on fifth Avenue and you know, maybe it's one less painting on the wall, right. Then, you know, I, I don't feel too bad for you.

B (31:23)

Right.

C (31:24)

Joe, you earlier you had like a second question you're going to ask about the public. I wonder where you want to go there. I want to talk about like kind of pop culture stuff in a second, but why don't you.

D (31:33)

Yeah, I think it's related the. You're pointing to the Netscape IPO as being the sort of bringing it to the public's attention. It's like sort of out there now and that's one of the things that makes it a convenient touchstone. And I think that like brings out the importance of the narrative as fueling the bubble. Right. So like, is that like something that's, that's necessary to generate a bubble, this sort of like emergence into the public arena so that a narrative can be effectively spun?

A (32:13)

Absolutely. So strategic narrative is. Is the name of the game. I think a lot of folks have kind of caught on to narrative as the importance of narrative. Even some economists now accept that narrative matters, which is totally. I'm all on board. Robert Shiller's presidential address to the AEA was about narrative economics and the way narratives structure belief. So for sure. And you know, what was it, I think it's November 30, 2022 was that, you know, chatbot gets released by OpenAI and you know, dang, right. We've had, I mean we've been doing machine learning and you know, we've had AI, the Stanford Artificial Intelligence Laboratory that Sebastian's run left, you know, to go to Google and start the. I don't know, was that Google car or whatever that crazy thing he did or was that the. Their education? Anyway, Stanford Artificial Intelligence Lab was, you know, started in the 60s, right. You know, we'd had this kind of stuff going on in fact, listening. ACME is waiting. ACME was one of the first. Central computers at Stanford that had a multi user access. And there was a, you know, this was, it was a classic kind of tech startup inside the academy. This was, you know, way back when, but so that it's been there for decades, but November 30, 2022 brings it to the public attention and all of a sudden the narrative snaps in like, wow, this is what we can do with AI.

B (34:18)

Which it strikes me that one of the interesting things that historians do in this conversation, historians of technology in particular, but also business historians, is kind of push back against those narratives by being the kind of boring old person who says, yeah, but there were things before there was, you know, there's, there's this history that's not. This change is not dramatically overnight. Everything is different. Which hype bubbles kind of depend on. And yet I, My sense is that we're usually screaming into the wind when we're trying to say those things.

A (34:55)

Yeah, sure. You know, what's the.

C (34:58)

David's been screaming into the wind about Tesla and a lot of other things for a long time.

A (35:02)

Yeah, me too.

B (35:04)

Still waiting on those to pop. Right?

A (35:07)

Yeah. I mean, Tesla I still really can't understand. There's, there's clearly some, something weird going on there, but. And you know, it may be a little bit like, you know, the Trump coin. There, There may be some. Yeah, the fix may be in there. But again, an interesting question is sort of what, what conspiracy do you believe?

B (35:36)

Should we just start listing them?

A (35:39)

You know, like, I mean, for, like, for my wife, it's that Jeffrey Epstein committed suicide. She just, you know, she doesn't believe it.

B (35:47)

Uhhuh.

A (35:48)

She's not a love. And you know, she doesn't. That's not something she says in polite company. Although I'm not credit to you, but you know, so again, thinking about like, what are the things that we, that we can't quite believe. But anyway, but I do think that the, you know, the idea that, oh, AI is just another technology, it's just, you know, it's a technology like lots of other technologies. Okay. It's a general purpose technology like the railroad, like electricity, you know, you know, like the integrated circuit. Yeah. Like, you know, we're gonna, it's gonna create a bunch of opportunity. It's gonna, you know, undermine and, and disrupt a bunch of sectors, you know. Okay, so, so, so what? Let's do it, but let's figure out kind of how to do that. My, my little. I'll share this with you because why not? Here's my, My little Pet theory of the day. So I mentioned P. Doom.

C (36:57)

Right?

A (36:58)

And I. So P. Doom has kind of was. Was what people would talk about in 2023, you know, what's the probability that AI will destroy civilization? Something like that. And so my recent realization was, if you were to kind of just behind the veil of ignorance, think of what is the policy that of all powerful villainous AI would concoct and have the, you know, the free peoples of the world subject themselves to. How does that policy differ in one in any significant way from what we are doing right now? And my argument is it doesn't. It does. We are doing exactly what that all powerful villainous AI overlord would want us to do, which is throw every piece, you know, mortgage everything, leverage everything, build everything, drain every lake, you know, build every building, convert every, you know,

B (38:11)

piece

A (38:11)

of silicon into the service of the AI and all, you know, to race to get AGI. And we've done it to ourselves. This is, this is the bubble that is the proof of the bubble is that we're doing exactly like P. Doom is one. It's here. Well, without the villainous overlord, it's us.

B (38:34)

So you're saying we've met the enemy and he is us.

A (38:36)

Yeah, that's it.

C (38:38)

Well, that's it. So that's it. You. You, David, you unwittingly did a perfect bridge from where I wanted to go from Joe to where I wanted to go. So it's, it's the role of like pop culture and, and the role of narrative, but the connection between narrative and pop culture and like shit like science fiction and stuff like that, right? I mean, you just spun out this like sci fi thought experiment, which is, you know, the, the effective altruism. People like to, you know, run these thought experiments all day long about, you know, like villainous AIs and all these kinds of things and existential threats. But, you know, like Sam Altman, Musk and all these guys, they so often are dipping into kind of like pop cultural science fiction narratives in, in, you know, when they're doing like presentation presentations at tech conferences or they're sitting on one of these stages on like cushy chairs with people like, they're, they draw on all these kind of very deep pop culture sci fi tropes. And it seems to be part of what makes their message so appealing. Is that fair?

A (39:47)

Sure it is. You know, the person I just defer to in this space is Jill Lepore. I think she's, you know, she has this whole series on Musk and On, on Musk's sort of the, his. How damaged he was as a kid and how sci fi sort of rescued him from, you know, from the. This kind of abusive nuclear family and you know, gave him sort of delusions of grandeur. You know, he was Tony Stark before, you know, in as an 8 year old or whatever. So anyway, so I think Jill Lepore is the person I, I kind of take my cue from on, on a lot of this sort of interpretive move. But I, but I do think it's, you know, it's not just Musk, it's all of these guys have a kind of. And it is, it's all guys. I mean, I don't know why. Maybe I do, you know, it's always about a girl. But, you know, so I, I think the, the culture isn't helping, that's for sure.

C (41:10)

Yeah, I guess another way to put it is, you know, I love your, I love your bubbles book. And I use it a lot in talks that I give. Like sometimes I get brought out to give talks about AI at other universities and stuff like that. And I use your book, your book is like very much a part of my frame. But one of the things I point out is that you guys are, you know, interested in financial bubbles because that's, you know, that's a core part of your book. But you're in. Your book's very helpful for thinking about that. But your book is also, I mean, when we think about like uncertainty, not knowing where we're going to go, like the, the, the presence of naive folks who, you know, don't know what's going on and are around kind of rubes and the power of narratives. Those things like setting aside pure play for half a second are also. There's like broader cultural bubbles or something like that, right? There's bubbles in a much broader sense than just the financial stock market aspect. And when I start, when you start to think that way, kind of like you're saying with Lepore, then all these other cultural tendrils come in, like pop culture and all these tropes that people kind of play on when they're spinning out these narratives.

B (42:23)

And there's an issue of a, a problem of imagination too. I mean, what strikes me about the sci fi stuff is that, you know, David, you, you threw out a thought experiment about what would, what would an evil nefarious AI actually do. And your argument is exactly what we're doing. But it strikes me that when the general public thinks about this, there is an inability to kind of Think about a reasonable range of possibilities because it's either the Jetsons or Terminator. You know, we just, we have this, I think pop culture and sci fi in particular give us a very.

C (42:58)

Yeah, that's the hype, critter hype to like to divide right there.

B (43:02)

Yeah. Whereas like, I mean reality can get really ugly and messy in human history and really devastating and deadly, but it's also often pretty boring relative to what we can imagine. And I, I, I worry that, you know, on both sides lee of the criteria thing, that there's just a, a lack of sort of ability to really appreciate a range of realistic options and then make decisions accordingly.

A (43:28)

So I think that would, I think that would totally track to a kind of AI as regular technology.

C (43:36)

Right.

A (43:37)

It's going to take, you know, Paul David had that dynamo in the computer. It takes 30 years for technologies to kind of cross into, you know, to have impact to diffuse. Maybe this is a little faster, so maybe it's 20 years, you know, or it just. But the problem is we say, we historians say when you accelerationists say this time it's different, we say, no, it isn't. History says it isn't different. But what if one of these times it is different?

C (44:14)

Right.

A (44:14)

And that, you know, then we're wrong.

C (44:16)

Black swan, right? I mean, yeah.

A (44:19)

Wow. AI really did, you know, learn to learn. And now, yeah, we can't turn it off or whatever. So I don't know. I, I don't get. I, for me, I do think the safe bunny is always on the middle, you know, something in the middle. It's not gonna, you know, we're not getting to the Jetsons. We're not gonna, you know, I don't remember any unhoused people in the Jetsons.

B (44:46)

It was the 60s. They were, they were in a solve poverty. Remember they declared boron.

D (44:50)

They were down on ground level.

C (44:51)

It. But David, to your point, this guy to, you know, I'm pretty, I'm, I've been talking about AI being a bubble for, since late 2022 when this whole thing started. But I, I've been writing about it recently. I've also become kind of like skeptical. People are too deflationary because they don't see how people are using this in everyday life to do things. Right. Like a lot of my people who do like mundane office work are using it, coders are using it. Joe, we use it for podcast work. Like Adobe has a pretty sweet AI based audio processing tool.

B (45:27)

History professors use it.

C (45:29)

Great. Right, right, right, right. So, so I mean, I What I like about you and Brent when you talk about these things, David, is you're like, no, these moments turn us all into, you know, ignorant, naive people. In a sense, it doesn't even matter if you're a bubble expert.

A (45:45)

Right.

C (45:45)

Like, you cannot see what's going to happen, you know, and no amount of study changes that. Right.

A (45:53)

No, we can't see around the corner. And yeah, I think it's. We need to be pretty candid about that. And for sure, that was, you know, we saw that so clearly. And the reason we wrote the book was because of what we saw in the 90s with, you know, going back to the dot com stuff. And so I think it's, you know, we're all talking about AI. I guess we're all AI experts.

C (46:30)

Right, Right.

A (46:30)

You know, we're all. We all know something, but it's. Yeah. You know, I think the, the weird part with it is to me is that it is kind of hard to disconnect ourselves from it.

C (46:48)

Right.

A (46:49)

If, if I were to say, okay, this week, I'm not using AI at all.

D (46:53)

Yeah.

B (46:54)

Right.

A (46:55)

Wow. I'm. I'm kind of like, what do I do? How do I. Yeah.

C (47:00)

Might be impossible in some ways given how it's being built into everything. Yeah, all.

A (47:06)

Yeah, yeah, exactly. And I can't, you know, I can't search on Amazon. I can't do this.

C (47:10)

I can't search on Google.

B (47:12)

Yeah, but if it was 1999, could you have said, I'm not going to use the Internet this week? You personally or me or some. Anybody else?

A (47:21)

I would guess not an academic. By 1999, I think we were pretty, pretty connected.

B (47:28)

I mean, email if nothing else, right?

A (47:30)

Yeah, yeah, but. But maybe it was easier then anyway, so I just think, you know, that's the part where I don't know if this really is the cultural piece you're talking about. But, you know, so, yeah, my bubble. And that my bubble story is about, you know, some villainous AI. But the thing that the villainous AI isn't telling us to do, but we're doing anyway, is actually reconfiguring our lives and building AI into it in, in ways that are, you know, consequential. And you know, we're thinking, you know, the AI agent applies for a job and the AI resume reader screens.

C (48:24)

Yeah, yeah.

A (48:24)

My, my resume that I've submitted, like, what's happening?

C (48:29)

Drafts the email and then some email summary AI tool, like, summarizes it.

B (48:34)

Yeah, yeah.

A (48:35)

And like, wait a minute, couldn't. So, okay, I Had a one, one line. You know, I remember this must have been before your guys's time, but there was this email thing called three Sentences that every email should be three sentences long to try and minimize the noise. And you know, so you just write something very simple. It was like three sentence es. It had some Spanish domain, so it wasn't.

B (49:08)

I hope this email finds you well. That's Star Wars.

A (49:12)

It is. That's one of your three. You better make the other two really hit hard. But you know, there, there are ways we could do this. We could be efficient, we could be if, you know, communicate more meaningfully. But, but we don't. So, you know, who knows, maybe half the people who listen to the podcast are listening to it at 2.25, you know, while they're like folding laundry or they're, you know, having their AI summarize it.

D (49:41)

Yeah.

B (49:42)

And we all sound like it is.

A (49:43)

It's just really weird.

C (49:45)

So.

B (49:45)

But this is, this is interesting to me, David, because I wonder if like one thing that you're pointing out is the stupidity of current practice, right? If, if, if AI writes the resume and AI reads the resume and AI makes the hiring decision, clearly that's just dumb waste of, of effort. But it also exposes kind of a stupidity in the whole process, which you would think would people would just eventually get wise to and, and realized if, if AI is going to write a 10 page report and then someone's going to read it and it's going to summarize it into three bullet points, that's just all silly wasted effort to start with, which has sort of shades of bullshit jobs. That strikes me as different than pumping tons of money into it as speculation. Am I wrong about that? I mean, is, is this. Is the stupid use function related to, you know, creating an actual financial bubble or is. Is there like a hype culture bubble that exists separate from a financial bubble that could pop with devastating consequences?

A (50:50)

Yes, I think they're related. I, I think you're right that the, the culture piece. And, and again, this, this, you know, if I go back to my. What would the villainous AI have us do? All the villainous AI cares about is feed the AI, right? That's all it wants. Feed it, build more every marginal dollar, more GPUs, more tokens, more, you know, get, Give me more power, give me more, you know, so feed me is all that the villainous AI cares about. And what you do with it is kind of like the fact that we're using some of that to create you Know, bloated vitas and simultaneously read, you know, summarize those bloated vitas is, is kind of on us. That's the, you know, what was the line I heard the other day? You know, broken AI won't fix a broken institution. Something like that. I forget who said it was a Princeton computer science guy. I think I'm probably butchering it. But, but, you know, something like this, this is the ins. The underlying institutions are. What are, are the things that don't work or are broken in some ways. And everyone is just seeing AI as the, you know, sort of the IT technology to fix it. And, and in that sense, I think that's, that is the sort of cultural logic of technology, right? Is that whatever is, is wrong. Let's, you know, whatever the culture tells us isn't working. Let's find, you know, use whatever the technology of the day is to fix it. I had some students pitching me a business the other day. I have to be a little careful because I don't want to, you know, rake them too, too hard here. But, you know, noticing that a lot of the lower class persons, you know, fresh people and sophomores and whatever, were feeling a little lonely on campus or having trouble, you know, I think they pitched it as like a compass, a social compass to help you find friends.

B (53:08)

It's friend AI.

A (53:09)

Yeah, exactly. And, you know, and their solution was an app. Let's get technology to do it. And I just said, you know, how can that possibly. Like the technology is the source of the loneliness. It's the source of the isolation. We all know it. You walk into class and when we used to walk into class, hey, how are you? You talking. There's noise, there's a din. It's a, it's. It, you know, it's, it's the opportunity, those five minutes before the boring lecture starts to kind of catch up with people. And now you walk into class and they're all just staring at their, at their screens. It's silent. And you know, so I think that in that sense that like we're, you know, we're, we just reach for the, we're lazy. You know, it's, it's. I feel like the other side heuristic. So we just, oh, it's like, here's a problem. Bang, grab that. Let's have AI. What can AI do for that?

C (54:10)

Yeah, I feel like the other side of your question, Ben, is like, so David mentioned Paul David's the dynamo and the computer essay, which I turned a lot to in these moments, too and you know, just to reiterate its argument is that with, with both the dynamo and the computer, it took a quite a bit of time for these computers to, to diffuse out and become built into everyday practices and become useful and have economic consequences. And I think what happens in, during that process is a lot of experimentation of using technologies that don't pan out. There's a lot of stupid shit that happens frankly as people try things. And you know, so we see recently we saw this kind of MIT study come out that found that like 95% of gen AI experiments happening in corporations right now are failing according to this survey they did and the study they did. And you know, people like the computer historian Jim Cortada, who worked at IBM for like his whole career, he said this is not at all surprising. This is what you'd expect actually this is how it works is when you introduce a new system. Like the early wave of approaches mostly fail like by huge amounts. And so I think that this is part of bubbles too in the sense of when you have new technologies emerge and there's these kind of, as David and Brent point out, these powerful narratives develop about how awesome they're going to be. You know, it forces firms and individuals to try things out, but a lot of it turns out to be fucking stupid, you know, and a waste of time. And I think a lot of like the most egregious examples we're seeing today of gen AI use just won't be around in 15 years. I mean this is where people will be trying to apply it and it's just not going to stick because it's a waste of time. It doesn't, you know, it doesn't save us. It, it decreases quality, all these kinds of things.

B (56:14)

So this is a very optimistic kind of view of creative destruction of capitalism, right Lee, that you know, these things will shake themselves out because you know, the market rational than that or whatever.

C (56:26)

Well, I mean, you know, I think I'm very interested, you know, you know, so thinking about Joseph Schumpeter, the kind of classic, you know, the father of innovation theory and you know, his picture of innovation, I'm very interested in how he thought of recessions and downturn, downturns and cycles because apparently somewhere someone said that he referred to downturns as douches, as in douchebags. Like so like in these, in like these expansion cycles you have a bunch of mal investment, there's creative destruction happening, but there's a lot of people trying things out that are not going to work, that are Just dirt, they're just garbage. And then you have downturns just kind of wash it all out. And that this is part of, you know, some shumpetarian evolutionary economist thinkers have thought about this way too. I mean, and so the idea is, you know, and this is, this is part of David's totally right. We're seeing now like a lot of the Silicon Valley folks, I just saw a piece in the Financial Times yesterday or the day before, they're pro bubble for this reason. They say no, like with new technologies we need a bunch of capital coming in here and we're going to burn it. A lot of it's going to be waste, but we're going to give you these awesome things. So I mean, you know, it's very interesting how there's probably some truth to it in the economic sense and the kind of shumpeterian sense. But now in a kind of self reflexive way it's become part of the hype. But the, you know, it's become kind of like what these hype mongers are up to themselves. They've kind of like adopted the way of talking.

B (57:59)

I wonder how much of that is actually a reflection of how we understand the history of the 90s and the dot com bubble. Because as you pointed out, David, it's cute. It's like this little, little thing that seemed like such a big deal at the time, but in the end, you know, it didn't really hurt that much. I mean there was a recession, you know, but then 911 happened and everyone sort of forgot about, you know, about this. And then the things that survived kicked ass and took names, right? Amazon comes out ahead, Netflix comes out ahead and Google comes out ahead. And all the fiber optic cables are laid and all the infrastructure is set and web 2.0 happens in the early 2000s and suddenly the web goes both ways. And you know, Time magazine decides that you, the user are the person of the year in 2006. And all of this kind of explosive Internet capability transforms modern life. And we get to say, oh well, you know, the bubble wasn't all that bad. The bad bubble comes five years later in housing, which is just pure greed, speculation and stupidity. And when it bursts, it destroys livelihoods and it throws politics into a fascist tailspin and it does all this horrible shit. But if we kind of glamorize the dot com bubble as more creative than it was destructive, then we look at AI and say it, let's do it again.

A (59:21)

Yeah. So the question would be, what's the section 230 of. Of AI.

C (59:27)

Oh, say that, let's explain what you

A (59:29)

mean so that, you know, like section 230 is this, you know, carve out that Congress gave the, you know, these Internet publishers that gave them immunity from being responsible for the content that's shared on their sites. And so if, you know, that was meant at the time to allow this fledgling industry to kind of take root and not have to content moderate every single post of Jenny's cat picture or whatever.

B (60:07)

I mean, just to clarify, this is in the mid-90s that this is enacted.

A (60:11)

I think it's a little, a little later. I don't remember the exact date I thought it was. I don't know what the date of the. I think it's dcma, the Digital Telecommunications Committee.

B (60:22)

Communications Decency act or the Telecommunications act,

C (60:25)

the Digital Millennium Copyright act.

B (60:27)

Okay.

C (60:28)

It's 98.

A (60:30)

Yeah. So it's pretty late. But these entities are here and you're sort of like, oh, let's help these little fledgling guys that Section 230 is now the thing that allows X and, and Facebook and Instagram to be the sewers that they are, right? That you know, they can't be, you know, they, they're immune from prosecution for, for slander, for, I mean, you know, I mean, except in the most extreme cases and you know, after 2016, after Cambridge Analytica, you had, you know, Zuckerberg and, and Facebook saying, oh, we're going to take content moderation seriously. We're not. And you know, all the while, of course, you know, the, like Facebook was the newspaper of, of Myanmar. Right? The Rohingya were being slaughtered on basis of, you know, these conspiracy theories and, and missing disinformation that was being spread on Facebook. And you know what, where are we 10 years later, right? We're basically Myanmar. Facebook is treating us like it treated the rohingya. And section 230 permits that.

C (61:53)

Right.

A (61:55)

And so the question again, if we're thinking about this kind of the bank shot the long term, then I'd say, well what do we need? What are the things we need to sunset now, like have a date certain that they will go away to make sure that we are, we. We aren't just a kind of locking in the assert a set of AI overlords, if I may, that who are, who can't be dislodged because of some, you know, quirk in the, in our institutional architecture that we put in place, you know, a couple years ago when it looked like oh hey, this thing might be kind of interesting.

B (62:44)

So Just to throw out for future reference, there's a whole sub theme to what David, you were just talking about, which has to do with, you know, information silos and the, and the fragmentation and stratification of politics and news goes back to the Fairness act and things like that and augmented through section 230. And Lee, I just wanted to throw out that we could think about the history of the 90s in terms of information siloing and things like that in a future episode.

C (63:15)

Yeah, well, actually that's. Thanks, Ben. I mean, that really sets it up. So, David, one thing we want to do, and it builds on kind of the way Joe has been thinking about our media production stuff for a couple years now in terms of not just using these podcasts to kind of popularize and synthesize thoughts that have already happened, but use it for setting up future inquiry and new knowledge creation. Like, you know, when you think as a historian of bubbles, but also a business historian and you think of like the dot com bubble moment, and we should say that you're also kind of like, we didn't go into this, but you're also kind of fascinatingly like an archivist of the dot com bubble and bust with your funky digital archive you have. But anyway, you know, how do you think about like, if you were going to approach it as like a serious historical topic, like examining the dot com bubble and bust, you know, what are some questions you ha. You have as a thinker and also what are some, like, how would you do it? Like methodologically.

A (64:24)

Yeah, that's, it's an interesting question. You know, for me, the, the, the business plan archive that I started in, in 2000, 2001, really, when we got it started, was a response to my writing a book about the history of the electric car where I had no records of this damn failed electric vehicle companies. Okay. And so here I was in 2000 writing a book about these electric vehicle companies, 900 year old electric vehicle companies. And so I kind of thought to myself, well, what are the, what is the historian of 2100 going to want as an evidentiary base to tell the story of this, you know, boom and bust period? In the same way, no one would say that automobile was a failed technology. It was a fantastically successful technology. There just happened to be a few dead ends along the way for, you know, perhaps institutional reasons, like why did we have distribution of refined petroleum but not electricity? And that mattered a lot for the types of uses that people wanted to put cars to. So it was kind of maybe a Little bit of an accident of history that we ended up. That I ended up not having the records of the electric vehicle companies, but only having the records of the internal combustion ones. And so that was kind of my. That was my lodestone in, you know, in 2001, 2002, 2003, was. Gotta save these records. Not for me, but for the historian of 2100 who.

B (66:17)

It's so cute that you think there'll be historians in 2100.

A (66:20)

I know that. How quaint, how naive. You know, the AI overlord will have it and can tell the story however they want to, but. So I think you're asking the question in a way. Well, what evidentiary base? So anyway, so I can tell you the questions that I was interested in about the dot com era and what was the role of business plans and how did. I remember going to Cafe Verona on Hamilton Avenue as a. A student. And it was walking distance. You know, walk past the Golden Retrievers, turn left, and you're at the. You know, you're at Cafe Verona. And I was just there, you know, probably reading Tom Hughes. But a lot of other people were there, like, pitching, and, you know, there were stories about, you know, you. The VC goes, you know, sits on the can and the Cafe Verona bathroom, and someone slides a business plan under the. Or it was, I mean, all kinds of stuff that was going on there. I don't know that we could really capture that. But my thought was, well, if business plans were the coin of the realm, let's capture some business plans and, and try and understand how people were using them. And what, what did they do? Were they, you know, just like a kind of boundary object that people, you know, a sort of thing you exchanged and you kind of weighed it and, and, and like, looked at it or did. Did people actually read them? Who knows? So there were a whole set of questions like that, I think, you know, maybe for. If we're thinking about how we tell the history of AI, what. What do we need? Maybe we need,

B (68:10)

I don't know, search logs, screenshots. Lots of screenshots.

C (68:14)

Right, right, right, right. Well, I mean, David, you're pointing that's the history of, like, use, right? And that is. That shit is hard, actually. Right. I mean, this is, this is why we have so many histories of invention and firms and. And stuff like that. Like, those records stick around sometimes. But how Joe Schmo was using some technology in work life or home life is. Is much harder to get at.

B (68:37)

And it gets into the thing we were talking about. Earlier about narrative and the. The kind of limited imagination that comes from, for example, sci fi or even historical understanding. Because if. If all you have left over is sort of evidence of discourse as opposed to actual everyday use of technologies, you're all, you know, is kind of what people were saying about it. And as we know today, what they're saying is often completely ill informed or it's. It's warped by a lack of imagination about what things could be. Can you imagine reading the 2025 discourse about AI and thinking that that reflects reality?

A (69:18)

Yeah, right. Well, and I was sort of thinking, what should I have, you know, in 2023. Right. My little P Doom insight. I should have been running a survey or something. So what's your P. Doom like? Hey, all you engineers, right? You know, and instead all we're going to get is like P Doom usage in the ending.

C (69:39)

Right, Right, right, right, right, right.

A (69:41)

It's not gonna. It's not really gonna help us very much. And so, and, and we're also now so. So much of what we read is AI slop. So our text analysis, all of all the things we've relied upon, the. The kind of text was kind of a human product. Has to be rethought. It isn't. You know, so much of it is slop. And so then how do we kind of sift, Get a human signal from the slop is going to be pretty hard too. I'm sorry, I may not have answered your question.

C (70:21)

No, dude, I think that Dark Note is precisely the right place to. To stop this conversation. David, thank you. Thank you for hanging out with us and talking to us about this stuff. It's as good as always.

A (70:36)

Great to see you both. Joe, nice to meet you. Yeah, you as well.

B (70:41)

Great to be with you. Thanks.

D (70:47)

Well, that's the 411 home slice for us. This has been all that and. And a bag of chips.

A (70:54)

Word. Word.

D (70:57)

Mad props for today's guests and contributors and for all of the hands that helped to make this episode a thing. Our hosts have been Ben Waterhouse of UNC Chapel Hill and Lee Vincel of Virginia Tech. The show is created in cloud spaces across the ether, but is anchored right here in the Newman Library Athenaeum on the Blacksburg campus of Virginia Tech. The whole thing chillaxes under the auspices of Virginia Tech Publishing and Press, and the New Books Network distributes our show to all of the places you'll find your podcasts. Along with myself, JM Lamb and Graham Conway, our editors, Sarah Beatty, built the art and animation. Any YouTube listeners may encounter. I am producer Joe Fort, and this has been the iHeart 90s history podcast. As a final note, I'd like to say that our deepest and most earnest gratitude has been reserved for you, our dear listener. Thank you very much for listening. Peace out.

A (72:15)

Sa. Sam.