BY Fast Company 9 MINUTE READ

Tim O’Reilly has been a conversation starter within the tech industry for more than three decades. The company he founded, O’Reilly Media, launched the first true commercial website in 1993, and remains a tech-industry staple that publishes tech books, offers online education, and holds virtual events.

O’Reilly saw firsthand the first wave of big dot-com companies swell, crest, and crash in the late 1990s. His company helped promote the rise of a new group of internet companies that would learn from the hard lessons of Web 1.0. In fact, O’Reilly popularised a collective term for that next wave of early-2000s companies: Web 2.0.

Twenty years later, another major reimagining of the web may be on the way, and its proponents promise a structural transformation more radical than anything seen before. “Web3” companies eschew the centralised power of Big Tech platforms, and propose to build apps that store data and transactions via blockchains. The Web3 movement seems to speak to the public’s growing unease with entrusting data and currency to powerful corporate or institutional platforms that may have financial and political interests of their own.

But the Web3 wave has a long way to go before proving it can produce technology with the functionality, reliability, security, and scale needed to disrupt the internet we have now. O’Reilly is one of a handful of influencers who have begun to raise doubts about its chances of doing that. After all, he’s seen this movie before—twice.

Fast Company: Matt Mullenweg, whose company Automattic runs WordPress, recently pointed out that the first Web 2.0 companies (including WordPress) didn’t set out to be big centralised platforms, and talked about ways to decentralize their models. In fact, much of the philosophical underpinnings of Web3 are things that were espoused at the outset of the commercial web and then again in Web 2.0. Why has the tech industry been unable to deliver the kind of content, commerce, and community implied by those early ideals?

Tim O’Reilly: I think there are two reasons, and one of them is that our economy is designed to favour centralization. Why is it that . . . big companies dominate every industry and basically we favour them? Our antitrust [approach] is seen as “Well, is there another giant to compete?” [O’Reilly is referring to the “consumer welfare standard,” which weighs mergers based on their effect on consumer prices.] It could be that the giant is competing with smaller, less efficient producers. We say, “Tough, that’s actually a good thing.” It’s great when the economy becomes more efficient. But the curse of efficiency is that it goes to the lowest common denominator. So if we want a diverse economy, we have to value something other than pure economic efficiency.

The other reason, I think, is this idea that Clay Christensen called the Law of Conservation of Attracting Profits. I have observed that the monopoly of IBM that was talked about when I entered the industry was replaced by the personal computer, which was putting a computer on every desktop in every home. That was decentralisation. Hardware became a commodity. Very quickly, Bill Gates figured out how to monopolise the software in the same way that IBM had monopolized hardware. Software became the way that you could aggregate power; we entered a new period of consolidation.

And then along came the internet and the web and open-source software. Now software is becoming a commodity. Software APIs are becoming standardised and everything is callable over the internet. It was a new wave of decentralisation, and that’s where I started thinking, Oh wait, this pattern is going to repeat. And I started saying, “Well, what’s next?”

So I guess I’ve just seen that pattern repeat so often that when everybody started talking about decentralisation and cryptocurrency, I was immediately looking for centralization. And you see it everywhere. There was centralization of Bitcoin mining, and it turns out the locus of centralization there was access to cheap power. You get somebody who figures out how to centralise power and extract outsize profits, and they get enormously rewarded with enormous market valuations. If you watch what we do instead of what we say, that’s the market we want.

And even in terms of this decentralised crypto, who did we reward? We rewarded the big, centralised crypto trading platforms. We rewarded the people who initially minted all the crypto currencies and ended up with the pyramid scheme where they were the ones at the top who built the NFT (nonfungible token) marketplaces. There are stories about somebody who sold an entity for a lot of money, but mostly the really big rewards have gone to the people who are able to centralise the new market.

One of the differences between the dawn of Web 2.0 versus the dawn of Web3 is there’s a lot more VC money being injected. I suppose this also may push models toward centralization because VCs look for scale.

Web 2.0 was not a version number, it was the second coming of the web after the dot-com bust. I don’t think we’re going to be able to call Web3 “Web3” until after the crypto bust. Because only then will we get to see what’s stuck around.

In 1999, you couldn’t tell whether or Whoopi Goldberg’s Flooz currency was going to be a big winner because they all had this huge market cap. That’s a key point I keep trying to remind people of. This is why I’ve been suspicious of Web3. All anybody ever talks about is valuations. And valuation has very little to do with actual economic impacts. There were all these companies that had enormous valuations (for the time—they’d be small valuations today) that weren’t real businesses. And as soon as the capital dried up they went away.

I think that we’re seeing just how much air can come out of the tires. There are going to be some things that stick around because they were real businesses. In two or three years is when I think we’ll be able to say, “This is what Web3 is”—if it’s anything.

When you say that valuations don’t equal market impact, I think what you’re saying is that you can look at these companies in a speculative way, or you can look at them in terms of whether they’ve built a better mousetrap that offers real value to a customer.

That’s exactly right. And whether people are actually adopting it. But again, it’s very hard to know because we get very little reporting. There’s been all kinds of breathless reporting about valuations. For example, if you look at the NFT marketplace, you can see that the number of people who are trading is in the hundreds of thousands. It’s not in the millions. And you think back on the web and how quickly Google grew to indexing a trillion web pages, which was a lot of content being put online.

There are some metrics being reported. When I hear that Ripple [which transfers funds across borders for financial institutions] had [a run rate of] $10 billion in cross-border transactions last year, I think, Well, why aren’t we reporting more on those kinds of stats all the time so we can understand if this thing is real or not? We’re not even having the discussion of it in terms that actually reflect any thinking at all about the impact of what I call the operating economy.

I like to contrast the betting economy and the operating economy. The operating economy is the one that delivers goods and services and people pay for them. And the betting economy is where you guess at what the value might be, or what other people think it might be. I mean, it’s the Keynesian beauty contest. The market is like a beauty contest where you’re not trying to pick the most beautiful contestant but the one who other people will think is the most beautiful, and they’re trying to guess what you’re going to think.


I hear people beginning to ask if we really need some radically new infrastructure to rebuild the way the internet works in a way that’s more egalitarian and fair. What do you think about that?

I think there’s way more interesting decentralisation stories than cryptocurrency. And the biggest one to me is the potential for the decentralisation of big data and AI. Google used to have this secret sauce. It only came from Google. And now that’s been encapsulated in these models that know a lot of stuff. BERT [Google’s breakthrough natural language model] was open sourced. GPT-3 (OpenAI’s even larger natural language model) isn’t open source, but it’s a big enough model that you can call on it [via] an API. So effectively you’re making a call to a centralized player when you do that. There may be sufficient advances in some next-generation model that’s so much better that it gives them a continued centralized edge. But what I’m seeing is that the stuff that they already did put out as open source, and that can be duplicated, is good enough that you can do vertical market training on top of it, because it’s a base.

There’s machine learning everywhere, in everything, and it’s having a real impact. If I’m right, it’s actually coming into fruition in a way that is going to be somewhat destabilising to the power of the big data companies. You look at how they’re enabling very, very powerful search for individual brands in a decentralised way. And then you go and there’s a decentralised shopping cart with Shopify and decentralised shipping or warehousing services. And all of a sudden you’re going to have competition for Amazon that’s getting more and more real. It’s a true enabler of the marketplace.

We’ve talked about the decentralisation of AI resources for 15 minutes and we haven’t mentioned the word blockchain at all. This is all happening independent of that.

Exactly, and so I think the patterns of centralization and decentralisation are everywhere. And the idea that it’s just about this blockchain technology is something we could easily do away with. My big takeaway on all the Web3 stuff is, there’s something really interesting there, but I don’t think we’re going to know until all the underbrush that’s growing up because of all of that indiscriminate funding. Then we’ll be able to see what’s left. It’s not like we shouldn’t be engaging with these technologies, but we should not believe the hype.

You have talked about how the decentralised Web3 frameworks will need to develop robust interfaces to the real world, including to legal systems and the operating economy. But it seems like some of the people who are building on these decentralised structures are actually hostile to the centralised systems we use now. And perhaps vice versa. How is that going to work?

Personally I’ve never had a whole lot of sympathy for those points of view. My involvement with open-source software really came out of looking at the world that was out there. There was Richard Stallman saying proprietary software is the enemy. What we have to do is create an alternative, and that was originally going to be the GNU system, and then it became Linux. And there were people who were going to make the GIMP, which was going to be the free version of Adobe [Photoshop]. And meanwhile, there are a bunch of other people who are like “No, we’re just making cool shit and we’re making it available for anybody.”

And I look at what came out of that movement—we got the World Wide Web, which is put in the public domain. We got the original TCP/IP implementation that Bill Joy wrote made part of Berkeley Unix so that anybody could do the internet stack. We had the DNS, or Domain Name System. We had sendmail, which is routing all the email. We had Apache, which has become the dominant web server. And none of this shit was like “Oh, we’re going to replace the proprietary bad stuff.” It was like we’re going to make this cool new stuff. And that’s the stuff that really ended up mattering, I think much more than Linux.

In a similar way, I look at the Chris Schroeder stuff about the uptake of blockchain technologies in Africa and I’m not close enough to really be able to validate it. But I think there could be some empowerment to people who are currently outside the system. But I’d be very, very surprised if we end up with a decentralised financial system as a result of blockchain. There are just too many points where it can be centralised, and too many points where it will need to be centralised. Where are the backstops? People have ended up in a position of power in networks for good reasons. We have a big complex world of radically decentralised stuff. And then there are things that are more enablers of decentralisation.

But you recognize that there are aspects of consumer or commercial tech that are too centralised right now.

There are things that we centralised unnecessarily. And I don’t see the alternation of decentralisation and centralization as bad. I do see, in general, that during the centralization phase bad behaviour happens. A good example of that is your laptop. This is the end game of the personal computer revolution and today your laptop is becoming less and less your laptop and more and more Apple’s laptop or Microsoft’s laptop. iTunes is a great example. You used to be able to rip your CDs and you had your music [library]. And now Apple says no, you don’t: You have to sign into your Apple account to use your music. You have to sign into your Apple account to use your PC.

And so here’s the centralization overreach, because everybody wants to be a subscription service. Everybody wants to have lock-in.

Amazon merchants [say] “Actually, I’d rather not be with you because you take too much of the pie. You tilt the playing field. People search for my product and you show them some third-party product that makes you more money.” That’s what I call algorithmic rents. It’s like the control over the algorithm means.

So you get to the end game, and that’s when everybody goes, “Screw this, we want something new.” The centralization overreach is one of the things that leads to a new wave of decentralisation.


Fast Company Senior Writer Mark Sullivan covers emerging technology, politics, artificial intelligence, large tech companies, and misinformation. An award-winning San Francisco-based journalist, Sullivan’s work has appeared in Wired, Al Jazeera, CNN, ABC News, CNET, and many others.