Jan 2, 2019 · 46m · a16z
a16z Podcast | How the Internet Happened
0:00 / 0:00
▶ Watch on YouTube →
gold bands on the timeline = statements, start to end. Hover to read, click to jump. CC turns on captions
Host Chris Dixon interviews author Brian McCullough on the a16z podcast to discuss the history of the web from Netscape in 1993 to the iPhone in 2007. They analyze dot-com market cycles, pioneer business models, timing failures, and how infrastructure enables application innovation.
How this conversation actually went
Every chapter scored 0–10 on four independent dynamics. Hover any point for the reasoning behind the score. How this is scored →
The host as informed peer 7.6 Guest teaching 4.4 Guest disagreement 2.2 The host pushing back 3.2
speaking balance: gold is the host, purple is the guest (3 minute bins)
Sharpest disagreement ▶ 5:55 Direct rejection of host premise on early web skepticism
Brian directly interrupts Chris with 'I don't think that's it' to reject Chris's framing that early skeptics simply didn't believe in computer networking, asserting instead that they knew infrastructure wasn't ready.
Hardest push from the host ▶ 29:04 Host challenges market lull thesis using Master Switch modelChris refuses Brian's simple diagnosis of a tech market lull by citing Tim Wu's Master Switch thesis, pressing whether market stagnation is driven by incumbent consolidation rather than a lack of ideas.
Biggest teaching moment ▶ 43:54 Deconstructing the myth of Jeff Bezos's master planBrian educates Chris on startup history versus hagiography, explaining that Jeff Bezos started selling books not out of a fully formed grand vision, but as an empirical test to prove e-commerce viability to himself.
The host holds their own ▶ 38:37 Deploying Rampel's theorem on startup distribution racesChris showcases deep VC expertise by deploying Alex Rampel's distribution race framework and TiVo/Comcast case study to counter Brian's view that incumbents inevitably neutralize startups.
the scores for every segment, with the reasoning behind each
| Chapter | Topic | The host as informed peer | Guest teaching | Guest disagreement | The host pushing back | Why |
|---|---|---|---|---|---|---|
| The Book's Thesis and Target Audience | 6 | 3 | 1 | 1 | Chris demonstrates solid knowledge of dot-com history by referencing a 2005 Dot Bomb card game and quoting Marc Andreessen on ideas being too early. Brian explains his two-part thesis targeting non-tech audiences and young entrepreneurs. The dynamic is highly collaborative with mutual agreement. | |
| Dot-Com Crash Cycles and Post-Bubble Recovery | 7 | 5 | 5 | 4 | Chris lists early 2000s green shoots like Flickr and Delicious before framing the information superhighway versus decentralized web debate. Brian directly interrupts to reject Chris's framing ('I don't think that's it'), arguing people believed in networking but tech wasn't good enough yet. Chris rephrases his premise to maintain his thesis. | |
| Good Enough Tech and Permissionless Innovation | 8 | 4 | 2 | 3 | Chris introduces a strong framework on the necessity of developers, hobbyists, permissionless architecture, and DNS in driving bottom-up adoption. Brian elaborates on Berners-Lee, HTML simplicity, and draws a parallel to modern VR limitations. The exchange shows peer-level technical depth. | |
| eBay's Legacy in Trust and Network Models | 6 | 6 | 2 | 3 | Brian outlines three major eBay contributions including trust, self-organizing reputation systems, and asset-light platforms. Chris probes whether eBay invented or popularized reputation systems and adds context about modern public misunderstanding of Uber's P&L assets. Brian educates Chris on Pierre Omidyar's motivations and newspaper executive confusion. | |
| Napster and the Convenience of Digital Media | 7 | 5 | 1 | 2 | Brian shares insights from Napster founder interviews regarding instant gratification versus piracy. Chris synthesizes Napster's historical arc into a clear 'three-act play' and connects it to Steve Jobs and post-Napster media savviness with YouTube and Netflix. The tone is constructive and analytical. | |
| Google's Search Miracle and Monetization Engine | 9 | 3 | 1 | 2 | Chris displays deep industry expertise by detailing Google's AdSense/AdWords evolution, GoTo/Overture's pure auction model, quality score mechanisms, portal business pressures at Excite, and Google corporate server box history. Brian shares his Google talk anecdote while agreeing with Chris's breakdown. | |
| Speculative Froth of the Dot-Com Bubble | 8 | 5 | 2 | 3 | Chris contrasts early high-quality 1995 dot-coms with late-bubble 1999 cynical VC flipping strategies, explaining the game theory behind greater-fool investing. Brian highlights demographic drivers like baby boomer peak earning years and lists failed precursors like MySpace cloud storage and Webvan. Both contribute rich historical perspective. | |
| Evaluating the Current Tech State and Market Lull | 8 | 4 | 3 | 6 | Chris challenges Brian's assertion that the current tech market is in a simple lull by proposing Tim Wu's 'Master Switch' media consolidation framework and questioning if software dynamics differ. Brian maintains his lull diagnosis based on picked low-hanging fruit and missing wow factors post-2012. | |
| Infrastructure versus Application Innovation Cycles | 9 | 4 | 4 | 5 | Chris articulates his infrastructure-versus-application innovation cycle model, explaining how the 2007 iPhone unlocked 10 years of mobile app development. Brian dissents slightly, arguing entrepreneurs are stuck copying the 2004 scale playbook rather than waiting on new infrastructure. Chris extends his perspective to enterprise SaaS upgrades. | |
| Incumbent Advantage and Startup Distribution Dynamics | 9 | 5 | 3 | 6 | Brian questions whether dominant incumbents block new startup platforms by instantly copying features or acquiring competitors. Chris counters using Alex Rampel's theorem on startup distribution races versus incumbent feature attachment, citing TiVo, Netflix, VR, and crypto as examples. | |
| The Role of Hobbyists and Long Time Horizons | 8 | 3 | 1 | 2 | Brian notes that major paradigm shifts originate with passionate hobbyists like the Homebrew Computer Club. Chris expands this into a full economic thesis comparing corporate 2-to-3 year ROI cycles with 10-year hobbyist and DARPA horizons. | |
| Surprising Historical Realities and Conclusion | 6 | 6 | 1 | 1 | Brian educates Chris on historical realities vs myth-making, explaining that Jeff Bezos ran books as a personal test before committing to everything-store scale, and Sean Parker had to convince Mark Zuckerberg to prioritize Facebook over Wirehog. Chris adds context on Wirehog before wrapping up the episode. |
Statements from this episode (25)
McCullough: Dot-com failures like cloud storage were just executed too early
“When you go into the dot com era stuff in those chapters, you're seeing all the things like grocery delivery, like cloud computing, cloud storage, and things like that, that were always good ideas, but seemed like the most boneheaded wrong ideas at the time be…”
Dixon: There are no bad tech ideas, only ideas launched too early
“Which I think the lesson is, I think Marc Andreessen said this, like there are no bad ideas, they're just ideas that are too early.”
Dixon: Post-dot-com consensus viewed the internet as commercially unprofitable
“The assumption was, it's one of these things like the airline industry or something that'll just never make money, right? It was a cool thing, cool tool, we're using it, but man, was that a bad idea from a business perspective, and that was the sentiment”
McCullough: Pre-IPO Google kept its massive profitability secret from the market
“Everyone had been conditioned to think that while the internet was fun, it was probably a niche, and you were never going to be able to build a truly great company on top of it. And then all of a sudden, because they kept it secret for so long, I remember peop…”
McCullough: Early internet believers did not think the 1990s web was ready
“I don't think anybody, even the people that eventually became internet true believers, thought that it was good enough.”
Dixon: The decentralized web was an underdog to corporate networks in 1993
“And then there was sort of this really kind of unlikely alternative, which was this, at the time, pretty crappy experience in Mosaic and, you know, homebrew websites and Yahoo, and you talk about this, you know, in the book, but like, and I think, I don't know…”
Dixon: Permissionless web architecture put university hobbyists on par with Disney
“Him and then also the academics and other people who made DNS, for example, like the fact that it was permissionless, the fact that a hobbyist in a university could be on sort of level playing field with Disney on the web, right?”
McCullough: High developer complexity prevents VR from reaching mainstream consumer adoption
“The obvious analogy to think of today is like, VR is still, it's not only can you not get mainstream people to adopt it, but it's still relatively difficult, and you have to be super sophisticated to create decent VR experiences.”
McCullough: eBay proved self-organized online reputation systems could scale effectively
“They created, or they proved that a non-hierarchical, sort of self-organized reputation system could function online, maybe not perfectly, but well enough to achieve scale to be good enough most of the time.”
Dixon: Avoiding physical inventory makes platform business models highly profitable
“That's precisely why they're so valuable because that's, it turns out that's a very profitable thing. You don't have to deal with all of the inventory and everything else.”
Dixon: Napster's core value proposition was convenience, not music piracy
“So it turns out, we now know that their value proposition actually was, as they claimed. A great deal of it was about convenience and not about stealing.”
McCullough: Spotify-style music streaming could have existed as early as 1999
“You could have had a Spotify in 1999 or 2000 at the very least.”
Dixon: GoTo invented the pay-per-click auction, the greatest business model ever
“Company called Overture, GoTo originally, and then Overture, For sort of completely, a little bit orthogonal reasons, invents the greatest business model of all time, which happens to tuck perfectly into Google's model.”
McCullough: Google manufactured physical enterprise search appliances until around 2020
“I think they were still manufacturing those as recently as two years ago, by the way.”
McCullough: Baby Boomers' peak earnings fueled the 1990s dot-com bubble
“As I point out in the book, like this is, historically, these are the baby boomers reaching their peak earning years, and they're gonna, that money had to go somewhere.”
McCullough: Dot-com era on-demand delivery startups failed because they lacked smartphones
“No one knew at the time you needed smartphones to get the on-demand delivery piece of the puzzle to work, but it's not like people hunkered down and waited.”
McCullough: The technology industry is currently experiencing an innovation lull
“I actually think we're in a lull right now, if I'm being honest.”
McCullough: Reaching a billion users today requires qualitatively superior products
“If you're going to reach a billion users now, you have to offer something qualitatively different than what other people are doing.”
Dixon: The current tech era is defined by modernizing outdated corporate software
“Like the software, the work feels like 10 years ago. So I, one interpretation I think is what's happening right now is we're just upgrading all that stuff, and that's how, that's quote unquote SaaS and all these other kinds of things going on, and it's, there'…”
Dixon: Snapchat's growth was severely capped by Facebook cloning its features
“I think it definitely had, I mean, I think it had, like look, look at Snapchat's a great example, right? That was a, I mean, you know they really invented sort of a new media type, right? I mean, that type of, you know, whatever, the vertical short video with …”
Dixon: Competition is a race between startup distribution and incumbent innovation
“My partner, Alex Rampel, I'm going to butcher the phrase, but it's something like startups try to find distribution for their innovation faster than the incumbents attach innovation to their distribution or something, something like that.”
Dixon: Most corporations are structurally unable to invest on 10-year horizons
“Most corporations are designed today, they just simply aren't designed to operate at that timescale.”
McCullough: Market cleansings are necessary to eliminate incumbents and old models
“It's so beneficial to have a cleansing, a calling of the herd sometimes. To, Blow up the old models and blow up literally the old incumbents”
McCullough: Jeff Bezos used books as an experiment to test e-commerce viability
“He did not necessarily believe himself that commerce was better in a virtual environment. But once he proved it, so like run it, starting with books and running Amazon was almost like this test that he was running to himself. And then because we see as soon as…”