Jun 14, 2025 ยท 2025 #23. Read the transcript grouped by speaker, inspect word-level timecodes, and optionally turn subtitles on for direct video playback
Edit labels for this show, save them in this browser, or download a JSON override for the production folder.
Transcript Playback
20 Years of TechCrunch
Human Transcript
Timed transcript
Blocks are grouped by speaker for readability. Expand a block to inspect word-level timing.
Speaker 2
Saturday, June the 14th, 2025. This may go out on Sunday the 15th. 20 years ago, a certain Mike Arrington launched in association with Keith Teer, my friend, TechCrunch. He sent out an email to his friends suggesting... that he was starting, I guess at the time, it was called a blog focused on Web 2.0 companies. Seems a long time ago that history of Mike Arrington and of TechCrunch now is archaic. This week, TechCrunch supposedly shut its European office. It's changed hands many times. And it's an interesting anchor to begin our conversation, Keith, this week. 20 years ago does indeed seem a very long time ago in the history of technology. It's centuries, isn't it?
It's at least three eras ago when you look at developments in tech. It was, you know, the post Yahoo era, just coming into the Salesforce cloud era with Web 2.0 as the main transformation in consumer tech.
And you linked to an ex from Jay Wong, who's writing about this, saying people were just... People were talking about Web 2.0 back in 2005. It felt like Web 1.0 was just taking off with eBay and Amazon. I'm not sure what kind of memory he has about, but he does make an important point that it's always easy to categorize history retrospectively. Everything always looks perfectly neat in hindsight, but back then it wasn't very neat, certainly not the neatness that it has today.
Yeah, well, look, Web1, which is a loose label really to describe, you know, AltaVista, Yahoo, Excite, Infoseek, and Hotmail and things like that, Web1 was still massively dominant. So to focus on Web2, where the poster child was a not yet funded YouTube, So what's the difference between Yahoo and YouTube when it comes to trends? And what we were doing in 2005 was looking at YouTube and making the case that that represented the future more than Yahoo. And that's where the Web2 label came from, from that set of thinking.
Well, Riley ran with it. But anyway, I mean, the term is just a term. You know that YouTube was very much bound up. in the foundations of TechCrunch that Chad Hurley pitched YouTube at the first party in Mike Arrington's house, which was known as the TechCrunch house in Athens. YouTube, of course, now is owned by Google. You suggest that YouTube was the first Web 2.0 company. I would argue, and I think you and I have talked about this before and maybe disagree, I would argue that Google is really the first Web 2.0 company. Is that wrong?
Totally wrong. Google's a Web 1.0 company. It's just indexing the Web and creating a search engine. Web 2.0 is characterized by, well, look, Web 2.0 was latent in Google in that it indexed your website and presented the content somewhere else. And the essence of Web 2.0 was that content and its original publishing place were separated and that now content could freely float and be re-represented in new environments separate from where it was originally published. So in that sense, Google did do that. But it did it as a portal. Google was a portal. It was a destination. The essence of Web2 was that portals themselves were undermined by aggregation and syndication. And so Google was, in a way, a hybrid. It was kind of both things.
It was at the bridge. But you know that the comparison was between YouTube and Yahoo. Isn't the better comparison Yahoo, which was physically indexing information on Stanford's campus? through a website or imagining yahoo.com being the center of this information and google aren't they quite different and was google and the early history of google of course is very very interesting but google.com was never the heart of the the
the google ecosystem was it oh it was google didn't have anything other than google.com until it introduced um its advertising platform, which came after it was a public company, by the way.
Yeah, but this came before Mike Arrington founded TechCrunch. This came before O'Reilly and Docky invented the term Web 2.0. What year did that happen?
Look, you've got to remember, I was sitting in the room with Larry and Sergey during these years because we were partners with them at RealNames. And so I know intimately how they were thinking. And their goal was to attract traffic to Google.com and then get paid when it went off Google.com. The only difference with Yahoo is Yahoo was trying to attract traffic to Yahoo and wanted to keep it. So the distinction between Google and Yahoo was whether you monetized exiting traffic And to monetize exiting traffic, you needed an advertising platform for external visitors and the reward that came from sending them. That was Google's innovation after PageRank. PageRank was its first innovation. But in both cases, it was a centralized company. And the essence of Web2 was decentralized. So Google definitely- I take your point.
But of course, the essence of Google compared to Yahoo was it was Automated. I mean, they didn't hire teams of digital librarians to sit on the Stanford campus and collate information for that.
Yahoo was a curated directory. Google was an index search engine from a crawler that did link counting. So architecturally very different, but from a business point of view, they were both centralized monetizers.
So at what point, this comes back to the key question, at what point did Google.com, was it replaced with, I don't know whether this is the right word, a distributed model where you didn't have to go to Google.com to search on Google?
But it's perceived to be decentralized, whether or not... whether or not you're really going to google.com, at what point did Google understand that it needed to essentially, I mean, it already downloaded the web. That was one of its more revolutionary qualities in terms of building a smart engine. But at what point did Larry and Sergey and Schmidt understand that they needed to essentially become the web to be ubiquitous? Anywhere anyone ever went on the internet, they would find a Google search button.
You don't find a Google search button anywhere on the web. That's a problem. So if you think about Larry and Sergey, what was their mantra? Organize the world's information and make it accessible. So yes, you're right. They had to put the web in a box. But so did Yahoo. Yahoo just did it worse. So in that sense, they were similar. They were both trying to take the web and put it into their place and have it become the authority. And Google was just better at it. And then they wanted to make it accessible to you so you could go somewhere. That was their original vision. By the way, they had no idea of monetizing it initially, none.
No, also not true. It was a guy called Bart Waskowitz, who is the husband of one of my employees, Barbara Gore, who ran sales there. He was hired as the first sales guy. And they tried to figure out how to monetize. And it took them quite a few years to come up. The actual course of events, and it's important for history, they partnered with real names and we were getting paid for clicks from Google. So the way it worked is, and this is before any revenue at Google, they would put my results at the top. And when that result got clicked on, I got paid. by the owner of the keyword. And that was the first time the penny dropped. Keith,
There we have it. Keith, you should describe yourself as
Words and timings
Therewehaveit.Keith,youshoulddescribeyourselfas
Speaker 1
By the way, they called it I'm Feeling Lucky. I don't know if you know that. It was given a name. There's a book that documents the whole thing. The book is called I'm Feeling Lucky, written by a Google insider. And it's all documented, so it's not made up. Well,
20 years ago, certainly things were dramatically changing, and they were associated in part, as Keith notes, with whether or not Google was a Web 1 or a Web 2.0 company is going to be very hard to resolve. Certainly, as you said, it's a bridge between the two worlds. It wasn't either conventionally Web 1 or Web 2, and it was the bridge, and more than the bridge, it became the road, extending all these metaphors for the Web 2.0 revolution. But the really interesting thing, Keith, is that we're back in a... at a pivotal moment, I think, in history. This week, you cover all sorts of pieces about the revolutionary quality of AI, which I guess in some ways is the Web 2.0 of 2025. These articles are almost predictable. The FT has something about whether AI will disrupt or displace jobs. Vinod Khosla has a piece and a speech about whether AI as a utopia or a dystopia. But the consistency, I think, is Google. There was a very interesting piece in the BBC, on the BBC website, ironically enough, about whether or not Google is about to destroy the web. Because one of the consistencies of 2005 and 2025 is the central role of
It's actually a super interesting story because the core of that BBC story, the word the web is used to describe the benefits that traditional media get from Google. So when this article talks about the web, what it's describing is the ability for traffic to flow to sites like the BBC due to Google's search engine linking to them.
Yeah, and it describes, and this is the way it puts it, and I'm quoting the BBC piece. The web is built on a simple bargain. Websites let search engines like Google slurp up their content free of charge, and Google search sends people to websites in exchange where they buy things and look at adverts. That's how most sites make money. They're assuming, of course, that... websites let search engines. I'm not sure they have much choice, but that's the simple bargain, right?
Yeah, and that describes the web as an advertising platform and a traffic and audience platform, which from a media company's point of view is what it is. Obviously, from a human point of view, the web's way more than that. So what the BBC is yearning for in this article, or at least the author, is the preservation of an advertising-based media audience platform. And what AI threatens is an information-based subscription platform.
And it's ironic that... because journalists have, for the last 20 years, been hostile to this advertising model. They've suggested that Google has destroyed the information, the traditional information ecosystem. So it's particularly ironic that, and again, I mean, we talk about the BBC, it's one writer on the BBC is suggesting that And again, I'm quoting the writer here. This arrangement, what he called the simple bargain, held strong for decades, certainly for two decades, between 2005 and 2025.
Is that fair, Keith? It is, it is. You could think of it as the social contract between search and media, just like there's a social contract between labor and capital. due to jobs, that social contract can't survive the transition to AI. It would be, if it did, it would be a setback because we would continue to monetize knowledge through advertising rather than have a value in and of itself. So it's scary as heck if you're a media company and it's exciting if you're, let's say an educator, And that happens at every juxtaposition.
Yeah, I mean, I'm not sure that an educator is not a media person either, or an education company isn't a media company. What the BBC says is say hello to the machine,
This bargain that held strong for a couple of decades, according to the BBC, as you say, you put it nicely, Keith, it was a social contract. Not everyone's thrilled with that social contract. They really accepted it. For Google, it was great. For media companies, it wasn't so great. They just didn't have much choice, right?
Yeah. Well, look, I would think of it in terms of media companies. Prior to Google, media companies with websites had to go and find traffic. So the digital media was much smaller than the physical media between 1994 and let's say 2002, maybe even 2005. Digital media didn't really begin to be as big as physical media until the 2020s. Look at the New York Times history as a good example of that. The revenue from digital is now bigger than from physical, but that's a fairly recent phenomenon. And so most people are not even in the position of the New York Times.
Yeah, and how does, in terms of this bargain, how does it work with, say, the Times? The Times makes some money on advertising and some money on subscription. How does the bargain work in terms of subscription?
It's a funnel. Inside the New York Times, somebody in the revenue department will have a funnel that looks at every click and how many of those clicks convert to subscriptions. And they'll do campaigns around that and they'll try to optimize their content around that. So it's basically just building funnels. And what the BBC article points to is that funnel is about to be kneecapped by Google. in Google's efforts to be relevant in the AI race.
Right, and this is the really important point. And I'm quoting again from the BBC. On the 20th of May, 2025, Google CEO Sundar Pichai walked on stage at the company's annual developer conference.
He said that Google is going further in terms of its development of AI. For those who want an end-to-end AI search experience, we are introducing an all-new AI mode. It's a total reimagining of search. What does that mean, Keith, and why does that... to put an end to this period between 2005 and 2025. Maybe we call it the Web 2.0 period.
So AI mode produces results from training, not from an index. So links to sources are not relevant because there is no single source for the AI's answer. It's coming from all of its learning. not from a specific New York Times article. And so there are no links in AI mode unless Google forces them in, you know, as an advertising ploy, which if it does will damage it, because no one really wants advertising in their AI results. So you end up with this binary decision, AI mode or search mode. And search mode, you can have this pipeline of conversion, in AI mode, there are no links, unless you explicitly ask for links, you will get some if you ask for them. And that means that the incidence of Google sending traffic to media companies is going to decline significantly. And the media companies will have to stand on their own legs for attracting an audience.
So in a weird way, does that, given that Google was the bridge, as you noted earlier, between Web 1 and Web 2, is this in an odd way a return to Web 1 or pre-Web 1? Or is it something entirely new?
It's something entirely new. Web 1 did have links and clicks and traffic. This is knowledge delivered to consumers for the purposes of learning and understanding. as a business model, probably it tends towards free because the cost of producing results, as Sam Altman talked about this week, tends towards zero. And so you really end up with free. And I think one of the key building blocks here is gonna be copyright law. My belief is that the legal system will say it's fair use for an AI to train on public information.
It comes back to, you noted YouTube's central role in the beginnings of TechCrunch. Of course, a lot of the content that was put up on YouTube was originally, quote unquote, borrowed from the web or borrowed from other people. It wasn't really owned by the people who put it up.
Yeah, that's true, but YouTube didn't steal it. It was people like me publishing, let's say, a music video. You did that, Keith? You were stealing other people's content? Sharing, I would say.
Borrowing. Borrowing. You're going to give it back. Maybe borrowing. But yeah, so I think we have to reimagine the internet as a knowledge engine. They call it machine, but it's really a knowledge engine, and it's a good thing from a human point of view. It's a very bad thing from a media business model point of view.
And as the BBC says, this changes everything. As you know, a lot of websites are struggling. On a day like today, Keith, as we're talking... There's war in the Middle East between Iran and Israel. There's also an assassination of a democratic politician in Minnesota. This news will be interpreted differently by different parties. You'll have the presentation, for example, of the Israel-Iran war presented dramatically differently in Tehran and Tel Aviv. How does AI cope with that? You say that it will be machine produced, but will machines say, well, if you're in Tel Aviv, this is how we need to interpret the war. And if you're in Tehran, this is how we interpret it.
Well, the AI is clever enough to do both. And remember, AI is not proactive, it's reactive. So it depends what the user asks for. If the user asks for a balanced understanding of different interpretations around the world, it will do what you just described. If the user asks for a strong pro-American opinion, it will do that. So the AI is capable of all of the above.
But it depends how you ask it. I mean, very few people will say to an AI, give me a strongly pro-American version of what's happening in the Middle East. They'll put a question like, why did Iran invade Israel? Or why is Iran illegally trying to get a nuclear weapon?
Well, my experience of using O3 Pro this week, which is the new recently launched one, is you have to pay $200 a month to be able to use it. is it will comprehensively understand everything, including building in web search and research tools and give you everything. And it'll leave you to decide opinions. There's no opinion really in AI. There's the ability to answer questions and research questions and even analyze questions using programming and code. So it's a tool. At the end of the day, it's a tool. It's not a replacement for thinking.
Do you think, though, that when historians of information and media politics look back, they'll see the late period of Web 2.0, shall we say, 2015 to 2025 as a period of fake news where
the Google web of links didn't work and got replaced by AI and then AI cleaned that up and there's no longer any fake news? Or do you think it will be seen as a prelude to an age of increasing subjectivity?
So I do think that Google's link counting stopped working roughly around 2010 as social media grew. especially Twitter, Facebook, LinkedIn. And so you moved from a page-based internet to a post-based internet, and posts were short and ephemeral, and attention, went away from the portals to social media feeds. You remember the old phrase rivers of feeds?
The real time web was a phrase often used starting around 2008, 2009. So when that happened, Google began to curate search results from an advertising point of view. And the top third of a Google search page was not search results at all. It was curated by Google for revenue purposes. And so Google morphed from a pure scientific-based search engine doing link counting into a monetization engine. And that's been true for a long time. And the results got worse and worse and worse as SEO companies gamed the search results using... So, yeah,
and I think that's really important, Keith. I think, you know, sometimes they tease you for your... tier-centric view of the world, but I think it's a very important observation. What should it teach us today? We're all eager to understand the future. Everybody wants to know whether AI will disrupt or displace jobs. Everyone wants to know whether it's a utopia or a dystopia, but we're all impatient for a future that is barely in the process of being born. What does the history of Web2 and Google in particular, which is the dominant, whether it's the dominant Web2 company or the dominant tech company of the age, what should it teach us about our impatience of trying to figure out the future?
Even then, you know, you're guessing based on trends. So if you ask me, I'll have an opinion, but almost for sure your opinion and my opinion will, there'll be a Venn diagram. and the overlap could either be thin or thick, depending on how obvious things are. So if I say there are going to be robots capable of doing domestic tasks within the next 10 years, you and I might agree completely. If I said there's going to be robots replacing pretty much every manual job within five years, we might disagree.
Okay, I take your point, but you're not getting the question. I'm asking you, what does history teach us about how long it takes for a technology because you were already saying these kind of things. You and I had many conversations. Steve Gilmore got it. A few people got it back in 2010. But it's taken 15 years for the thing to manifest itself so most people begin to understand it. What does that tell us about this new AI age? How long is it going to take? Will it be another 15 years? Will we be doing this, hopefully, be doing this show in 2040 and talking about the reality of an AI age then?
Well, there's a famous way of thinking about that called an S-curve. An S-curve is a horizontal curve, like a figure S, and it basically describes the early adoption of a technology leading to extrapolations. We're probably there right now with AI. And then the actual product kicks in and usage kicks off. So the S curve goes up through adoption. And then at some point, there's a kind of a pause and it slightly goes down again while you digest it. And after that, it goes up again. And these S curves tend to last two decades. So you're definitely not at the end game at the start of the S curve. You're just at the beginning. But you can already see the whole play, if you will. The script's already there, but there's execution and operational excellence has to happen to turn potential into reality. For example, we all knew that TV wouldn't be broadcast in 1994. In 1994, when Warner Brothers was focused on interactive TV, I was writing stuff saying TV is going to be separated from broadcast and time won't matter anymore. In 1994, it really didn't become true until TiVo, And so you need actual companies to do things to make potential turn in reality, which is why the hostility to big tech is so flawed. Unless you have tech doing things, none of this stuff becomes real. And individuals who are driven by passion like Michael Arrington was with TechCrunch.
Yeah, although I'm not sure everyone would agree to use the word passion, maybe other words. In other words, when the BBC says say hello to the machine web, it's going to be a long hello. It's going to take years. And you talk about big tech. My interview of the week this week is Karen Howe. Her book, Empire of AI, Dreams and Nightmares, and Sam Altman's Open AI is already a bestseller. It's having an enormous influence. She's very critical of OpenAI and its imperial aspirations and consequences. I'm not sure if you had the chance to look at the interview with Karen or her book. So look, in the early... You've always been very hostile to critics of big tech, but doesn't she have a point here? I mean, especially given the last 20-year history of the way in which big tech, Google, and Facebook have corroded democracy in our information ecosystem.
You're saying she doesn't understand what's happening, even though she's an experienced tech reporter. She worked for the MIT Review, Wall Street Journal.
right that's that's in the nature i've written i've written two books um in my career and when i was writing those books i wasn't really doing anything i was summing up what i understood to date and interpreting it that's a different skill it's not i'm not being against it by the way you write books it it's a very important thing for society that some people do that but they're not going to make history They're going to interpret history and maybe observe history and maybe influence the next set of thinking if they're really good. But the truth is AI today, to be a believer, you have to be probably doing it. The most optimistic people about AI, not insignificantly, work at OpenAI because they can see stuff they haven't launched yet and they know what it can do. And so optimism comes from knowing, not from observing. And Therefore, you've got to be, as the phrase goes, in the arena to be an optimist. You can also be a pessimist, by the way, because sometimes things don't work.
Well, I don't agree with that. I mean, back in 2005, every journalist was writing a book about the promise of Web2. In fact, my book, Cult of the Amateur, was the first really corrective on that. It was the first pushback against that optimism. So I don't think all journalists write optimistic books. Sorry, not all journalists write critical books. They're writing them now like Karen Howe's Empire of AI, but they're not always. And people, viewers should... Look at the interview with Karen. She's particularly good and articulate, and she knows her stuff. She disagrees with Keith on OpenAI.
And I want to give her credit, Andrew. I said she represents probably 95%. So I'm not knocking her. I'm just positioning her in the S-curve. as being the early majority that's against them.
Is there a difference, though? Back in 2005, everyone was writing utopian books about the web. Now they're all writing dystopian ones. Is that a difference, or is it basically the same thing?
It's a reaction to reality. The more real AI becomes, the greater the dystopian belief will get a megaphone. And then the S-curve The second S-curve kicks in, and eventually everyone becomes adopters. And we're early in that, like you're already an adopter. I'm a deep adopter. The adopters will grow massively, and eventually people like her will join the adopter group.
Well, she's an adopter. I'm sure she uses AI. She's a catered technologist. One can use AI and still be critical of it. In the same way as you could be critical of social media and have a Twitter account.
I don't think she's just critical. We all should be critical because things aren't perfect. She's more than critical. She is a doomster, if you will. And so she's not just critical. She's at a binary level against.
Well, people should read her empire of AI, look at my interview with her, make their own minds up. Finally, post of the week, ironically given the fact that you said it was that changed the original 2.0 revolution and corrupted them, and at this point is Josh Koppelman. He posted on X, Facebook made social networks, TechCrunch made startup networks 20 years ago. I received this email from Arrington, and ever since... TechCrunch has been an open tab in my browser. It's hard to underestimate the impact it's had on the ecosystem. Is TechCrunch, will it be one of the victims, if you like, of the AI revolution? This week it's making news, as you note in the news, because it supposedly shut its European office, although it claims it didn't.
I think if they do, they have to adopt the TechCrunch playbook. If you characterize what Mike Larington did with TechCrunch, it wasn't really ever a journalistic effort. It was what I would think of better as a participant observer, as we say in sociology. That is to say, it both impacted and observed at the same time. Mike's opinions were always very strongly articulated, although often not felt as deeply as they came across because he was able to learn as he went. But TechCrunch became the organizing center of Web 2.0. This startup network that Josh refers to was literal. Startups came together at TechCrunch events, met each other, talked, learned, iterated. So a mere media company couldn't have done that. it was an organizing center for a new ecosystem. And in AI, that really doesn't exist. There is no publication that's playing that participant observer role. The nearest is Sam Altman's blog. where this week he made a big post about the near future of OpenAI. But even that isn't because he's a single company. So there's no tech crunch in this AI era, which, by the way, probably is an opportunity, but it has to be an insider participant.
Yeah, but then it isn't by definition the way you're presenting it. I don't quite know how it could work. I mean, isn't the AI revolution by definition against OpenAI? single publications?
No, I don't think so. It's against single publications as sources of knowledge, but it's not against... TechCrunch wasn't in that narrow sense of publication. TechCrunch was the platform on which Web 2.0 came together very different to say Tim O'Reilly's Web2 conference, which was literally just an event. TechCrunch was an organizing center. TechCrunch Disrupt was where every VC, every company came and talked to each other. And so in that sense, it was way more than a publication or an events company.
Yeah, as Arrington said in that original article, in that original note um we're also launching a new business weblog called tech crunch hearing about new companies every day and i realized that there is no place to go to see profile new services so that's what he was doing it will be interesting to see keith we're in this early age uh may last another 15 years but um the web as we know it It's finished, at least in the long term, and it's going to be replaced by AI. We will be reporting on this on our weekly That Was The Week show over the years. So we will be doing it for the next 15 years, Keith, and by that time, there will be something else. So keep well. I don't think we're going to do a show next week because I'm in London, but we will see you all in two weeks. So keep well, keep reporting, and we will talk in a couple of weeks.