MGI Research

Forward Thinking on tech and the unpredictability of prediction with Benedict Evans

In this episode of the McKinsey Global Institute’s Forward Thinking podcast, Michael Chui talks with Benedict Evans, an independent technology analyst. Evans covers topics including the following:

  • The generational shifts in technology that occur every 15 years or so
  • What characterizes a “universal product”
  • Why Web3 generates the most polarizing views among senior tech people
  • Whether tech is becoming a regulated industry

An edited transcript of this episode follows. Subscribe to the series on Apple Podcasts, Amazon Music, Google Podcasts, Spotify, Stitcher, or wherever you get your podcasts.

Podcast transcript

Michael Chui (co-host): Janet, have you heard about this Danish proverb that says something like “It’s difficult to make predictions, especially about the future.”

Janet Bush (co-host): Yes, I have. Is this the kind of thing you think about when doing MGI research on technology topics like artificial intelligence or Internet of Things?

Michael Chui: Yeah, I think about it a lot. A lot of people have heard me say about our research that we lay out scenarios, or potential. But a lot of times people want a definitive answer about, “Will this technology be a big deal?” or “When will everybody use a technology?” But what really matters for making decisions is laying out the factors that could lead to one set of outcomes or another. And these are the kinds of questions that today’s guest has been grappling with for years.

Benedict Evans is an independent technology analyst with a huge following, and he talks about how he thinks about whether a technology might end up being just a niche product or something that everybody uses. We also discuss some hot-button tech topics of today, including crypto, Web3, and the impact of regulation on technology.

Janet Bush: I find some of those things impenetrable but also fascinating. I know I need to learn a lot more about them and I am very much looking forward to hearing the conversation.

Michael Chui: Benedict, welcome to the podcast.

Benedict Evans: Thank you for having me.

Michael Chui: I’d just love to start with a little bit about your story, your journey to where you are today. You have a newsletter, if I recall correctly, with about over 170,000 subscribers, for instance.

Benedict Evans: Something like that, yeah.

What’s that joke? You know, “Life is what happens when you’re making plans.” I read history at university, went into investment banking as a sell-side equity analyst in the dot-com bubble. So for six months, I thought, “Oh, this is going to be fun.” And then it kind of wasn’t. And that whole industry doesn’t really exist anymore, or not in anything like the same way.

Then I worked doing various strategy roles in media and telecoms companies. Worked for a boutique consultancy in London for a few years. And then went to Andreessen Horowitz at the beginning of 2014, which is a big Silicon Valley venture firm with, now, $20 billion or something under management. And was there until the end of 2019 working at software companies and internet companies and technology startups of various kinds.

And then at the beginning of 2020, moved back to London to do my own thing. So I do various things where I try and explain what’s going on, which is sometimes publishing, sometimes talking to people, sometimes presenting.

Michael Chui: I’m curious. You talked about studying history, and now you’re mostly known for your work analyzing technology. I’d also love to just get your perspective on—even the word history is backward-looking in time. But a lot of the work that you do is meant to have implications for how things look going forward. How do you bridge those ways of thinking?

Benedict Evans: Well, there’s always patterns in technology. I mean, there are patterns in these kinds of analyses, technology, of course the patterns happen quickly. It’s like studying fruit flies instead of studying elephants, I suppose. You get the generations pretty fast.

A few years ago I met the CEO of a giant utility company, a giant water company, and they said, “I think maybe innovation is one of our top five priorities this year,” which sounds insane until you think that this is a water company. That might be an entirely sensible way of thinking about stuff like this, but the other stuff is more important than innovation if you’re running a utility company. Holes in the ground and price regulation, that stuff.

Whereas in technology, in the last 50 years, we’ve had these generational shifts every 15 years or so, in mainframes and PCs and the web and smartphones. And there’s a lot of value to looking and seeing, what happened the last time this happened? How has this evolved, and how might it happen differently?

There’s a great line by Hugh Trevor-Roper, “History teaches us nothing, except that something will happen.” There’s a narrative that says that the Bolsheviks were extremely interested in looking at all the failed revolutions of the 19th century and trying to learn, well, why did this happen, and why did 1848 fail, and why did the [Paris] Commune fail, and so on.

We have to learn the lessons of this, and the important lesson that they took is, you have to be really careful of successful military generals, because they might turn into Bonaparte and make themselves emperor. So they’re really suspicious of Trotsky and don’t pay any attention to Stalin, who of course shoots them all.

It’s like the thing that columnists say, “All models are wrong, but they’re all useful.” You can over-index on learning these lessons. People looked to the iPhone and said, “Well, we know what happened with the Mac versus Windows, and Apple’s going to be destroyed.” And that was the wrong lesson, and you should’ve looked at another aspect of what was happening in the ’80s.

You have to be careful about what you’re looking at. But it’s tremendously useful just to think about, “Well, what patterns are there, and how have those patterns evolved, and what might happen now?”

One current example: everyone in the tech industry looks at cars and says, “This is going to play out like the iPhone versus all the old phone makers. It’s going to play out like the iPhone versus Nokia. And all these darn old car companies that thought they could just hire software people, they’re going to get crushed.” Maybe, but that’s not necessarily what’s going to happen. It may not play out like that.

This is the elemental question for Tesla. Tesla bulls think it’s a software company, and Tesla bears think, no, it’s a car company. So it might play out like that, it might not. You look at those lessons, and you draw perspectives. But then the hard part is working out, “Well, which one?”

Michael Chui: I would love to come back to how you think about what you do, and how others might think about understanding it as well. But maybe we can touch on a few actual examples. You’ve come up with a few. You’ve mentioned a few already.

These fruit flies, as you might say, which are evolving very quickly in technology. Let’s start with one that’s very current. We’re recording this in February of 2022. The word metaverse has come up many times in the dialogue. And how about I just raise the word and get your reflections on what you see there.

Benedict Evans: Different ways to talk, to think about this stuff. Immediately I’m going to go to historical analogy, because the metaverse thing reminds me a bit of the information superhighway. Just a little bit before my time. I was still a teenager when that was a thing.

Information superhighway is this early-’90s thing, which is basically people have worked out that maybe a lot of people are going to have computers.

There’s all this stuff that means a turning from being these boring boxes with green screens that do accounting. And there’s all this stuff happening, and they’re probably going to get connected to networks. And they’re probably going to have consumer information on them.

And so you get a whiteboard and you write words on the whiteboard. And you write, like, “graphical user interfaces.” Because that was an exciting new idea. And “multimedia” and “convergence,” and “broadband” and “fiber,” probably “virtual reality” as well, and “CD-ROMs” and “color screens.” And you’d write them all on the whiteboard, and then you’d call it an information superhighway.

Who’s going to build this? Bertelsmann and Viacom and the New York Times Company. And it didn’t work out like that, but we got most of the stuff that was on the whiteboard, just not like that.

And the same thing now with metaverse—you write a whole bunch of words on a whiteboard. You write “games” and “virtual reality” and “popular culture” and “self-expression,” and maybe “NFTs” and “Web3” as well, maybe. And “augmented reality” and “glasses” and “3-D” and “spatial internet.” You write all this stuff on the board. And then you call it metaverse.

Then you have these weird conversations, like, “So is a Facebook metaverse going to be compatible with a Google metaverse?” Which to me is like saying, “Is a Facebook mobile internet going to be compatible with the Google mobile internet?” I don’t understand what that question means.

Some of that will probably happen, but some of it won’t. Maybe, maybe not. There is a bit more of a binary question around metaverse, in that we don’t quite know whether, if you make it very, very tangible, the core is, “Are we all going to be using VR or AR or something? Are we going to be wearing some fundamental new kind of screen that is much more about placing stuff into the world or placing you in a world or much more 3-D or immersive or something?”

There’s one argument that says, “Look, another ten years of Moore’s law, and virtual reality will be this transformative experience that everybody uses.” And maybe it will. Maybe it’ll be pass-through, maybe you’ll put it on and you’ll look through it into the real world.

Equally, maybe somebody will solve AR optics, and you’ll be able to wear a pair of glasses that can place something in the world that looks like it’s there and that works in daylight, outdoors, in a café with a broad field of view, and looks like a pair of glasses, not like a headset.

It’s just engineering and time, basically. Give it five to ten years and everyone will do this. Which is sort of what took us from black-and-white WAP phones to iPhones, basically ten years of Moore’s law. In 2006, someone was going to do that if it wasn’t Apple. Something like that was clearly going to happen.

The counterargument is to say, this is a bit more like games consoles, in that it doesn’t matter how much Moore’s law you give to a games console, it’s still basically a niche device. And 200 million people have a games console, which is smaller than Snapchat, whereas five billion people have a smartphone.

It doesn’t matter how amazingly cool a games console is, it’s still basically not a universal device. And you could extend this and say, “It doesn’t matter how cool drones are, we all got one for Christmas a couple of years ago.” By four days later, guess what? It’s in the closet. Like, OK, I’ve seen the roof of my house now. Now what do I do?”

You buy a 3-D printer, many or few, maybe you got a 3-D printer. “OK, I made a little Lego Eiffel Tower. It’s kind of cool. Now what do I do with it?” And that’s kind of that. So there’s this technology determinism which also applies a bit to Web3, which is to say, look, smartphones, PCs, the internet, cars, aircraft—it starts out looking like a toy and it gets better.

The question is, OK, yes, it probably gets better, although there may be reasons why it doesn’t, which is another interesting conversation. Not everything gets better. Sometimes stuff is just stuff. It can’t get better. We don’t have flying cars yet.

But if it gets better, OK, what does that mean? Is that everybody? Is that a universal product? Is that, or is this, some kind of niche device? I just remember my math teacher at school blew my mind by saying, “There’s all sorts of different sizes of infinity.” And there can be all sorts of different sizes of success. Maybe it works and no one cares.

This is the metaverse question. If everybody had VR and it worked, and we were using it all day every day, well, what would that be? And you can make your list of stuff, of what that would look like—games, pop culture, self-expression, identity, spatial internet. Maybe, yeah. Maybe not.

Michael Chui: Why is it not old wine in new bottles, in the sense that people have talked about VR before? Is it because of the addition of all of the other—it’s undefined, right?—but is it the addition of all the other things that you mentioned?

Benedict Evans: We tried to do VR in the early ’90s, late ’80s, early ’90s. And at that point, the technology, the hardware was good enough to show that this would be cool if we had vastly better hardware. But it was not good enough to actually do anything. You’re basically still wearing a CRT, and you’ve got a $50,000 graphics workstation to power the thing. And it’s still kind of crap.

So we basically put it aside for 30 years and came back. And then, getting on for ten years ago now, people realized, “Hang on a second, with the screens, the flat-panel screens and the smartphone chips we have now, this would kind of work now.”

So the gyroscopes and the screens and the cameras and so on, that lets you do the sensing. And here we are ten years later, and the Oculus Quest 2 is a great product. It’s still sort of a smartphone game kind of a product, rather than a PlayStation 5 game kind of a product.

But if you put on an Oculus Quest 2, you don’t say, “Well, it’ll get better, it’s a prototype. You’d have to make allowances for it.” It’s a great product. Question is, what do you do with it? And then the opinion divides. There are some people who would say, “Another five years of effort and Moore’s law, put our shoulder to the wheel, it’ll pop out into a great product.”

Which is, as I said, sort of where we were with smartphones in the mid-2000s. Clearly the iPhone, the BlackBerry, and Windows phones were not the endpoint, but we were clearly on the way there. And to extend the analogy, VR in the ’90s was like general magic in the mid-’90s. It was just “great idea.” It was just, “We can’t build that yet.” Counterargument is, “No, you need another 30 years of Moore’s law before everybody will use this.” And even then, maybe nobody will use it. Maybe we’ll just go straight to brain implants.

Michael Chui: How would we figure out which it’s going to be? If it’s going to be a game console or 3-D TVs—where it’s, as you described it, even a 200 million shipment, niche product—versus, as you described it, a universal product.

Benedict Evans: I don’t think these are questions you can know. Yes, this will get faster. What’s more difficult is to say, “OK, if it got faster and everybody could have it, what would that mean? Would everybody want that?” People always crave these kind of axioms. It’s like the Henry Ford line, “If I asked people what they wanted, they’d have said ‘faster horses.’”

Go back and read Brave New World. Everybody has a personal VTOL airplane. Everybody flies around. That’s in the ’20s and ’30s, when you could see that planes were going to be a big deal. You could see they were going to get faster and reliable, and everyone could fly.

But you didn’t really understand what that would mean. People still thought of airliners. People thought airships would be a thing. And didn’t understand that what would happen is the aircraft will get big and fast, and there’ll be a 500-seat airplane, but no one will have their own airplane. It’s a kind of classic sci-fi problem.

My grandfather wrote science fiction and was quite a successful, famous science fiction writer in the ’40s, ’50s, ’60s. And I have a cover of one of his novels, which is about the terraforming of a planet. An essential plot point is that the planet gets forgotten halfway through the process, because they move the terraforming office to a different planet.

An index of paper card files is knocked over, and the card file for that planet flies under a table and gets lost. So now, we’ve got faster-than-light interstellar travel, but we’re storing stuff on card files. And navigating this starship. You have interstellar travel, but you have to queue up to buy a paper ticket. It’s always hard to predict the axes of this stuff.

Michael Chui: Indeed. Speaking of predicting axes that are difficult, you’ve been involved in a lot of thinking and discussions about crypto, Web3. For folks who are listening who don’t know what all these things are, what are people talking about, and what conclusions have you come to?

Benedict Evans: I’ll give you three bullet points. In fact, I’ll give you three sets of three bullet points.

The first is that the first wave of crypto is—well, to go back a second, there’s this great quote of Voltaire’s where he says that “the Holy Roman Empire is neither holy, nor Roman, nor an empire.” Which was true. It wasn’t any of those things. Cryptocurrency is not cryptography, and it’s not a currency, or not only a currency.

It’s not secret, which is what people understand about crypto. It’s not secret, it’s not only a currency. Then you get this phrase “distributed ledger,” which implies it’s a database, and it’s not really a database. If you say “Bitcoin,” then you’re ignoring all the other stuff. It’s really hard to work out what you should call this thing.

Michael Chui: Just to pause, why is it not a database?

Benedict Evans: Because, well, let me come back to that. I’ll explain why.

The first wave of crypto, for want of a better term, is to say, “Well, this is a way of moving money around. It’s an asset, it’s a digital asset, it’s a bearer currency of some kind.” If we agree that a bitcoin is worth something, then you can move them around on this public database. But it’s a public database, which is important. It’s not just a database, it’s a public trustless distributed database.

You don’t need to trust that a particular third party will do the right thing. It’s a mechanistic system, so that no one has a choice. It’s sort of digital gold, or it’s digital bearer bonds or something like that. And that’s crypto one.

Crypto two is people say, “Well, what if you could put software on a blockchain? What if you could put basic scripting on a blockchain?” That software, because it’s running on this trustless environment where no one can change anything without permission, that’s basically a contract.

So the software has become a contract, which gets you this phrase “a smart contract.” You could lend money and be paid. And the payment would be mechanistic rather than based on people doing what they’d agreed.

That gets you DeFi [decentralized finance]. So this is crypto two, it’s distributed finance. You could build a lending application on this.

Crypto three is people say, “Well, hang on a second, if we can program this, why do we have to program money? You could make Instagram on this, you could make Twitter on this, you could build Spotify on this, you could build actual software on this thing.”

At that point, you’re saying, “What is a blockchain?” It’s a distributed, decentralized, community-based computing system that has trust incentives, contracts, structure, ownership, portability built in.

Explore our Forward Thinking podcast series

My third list of three bullet points, this is another wave of open source at this point. Because original open source back in the ’90s—some radical, very religious idea. Anyone can write the software. The software is written as a distributed, open community project rather than inside a company.

Lots of people think this is absolutely insane. Turns out it takes over the tech industry. But what actually happens, the second phase, is actually all of that software is deployed inside big companies and inside Google. You can’t actually see the code. I mean, technically you could download it, but you can’t see what Google or Facebook are doing with it. Crypto, blockchain is an open-source computer, but the code is open as it’s running. It’s not just that you could get the code, it’s that you can see the code as it’s running.

So you have this open-source, distributed, collaborative, trustless computing platform, which is a data store, is a way you could write software, is a way you can communicate, that has incentives and payment and ownership built into it.

That’s my three, and that’s why you would call it Web3, because the original web was Yahoo and the New York Times dot com. Company runs website, writes stuff for website. You have a website of your own, you have a personal website, you can try that.

Web2, which is—“[What Is] Web 2.0” is this very influential essay by Tim O’Reilly. He said, “Look, all the stuff that’s happening in the 2000s is, the users are creating content.” So that’s Yelp, Tripadvisor, Flickr, Delicious, Google, YouTube, Instagram, Myspace, all of these things.

Instagram came later, but all of these things where the users create the content—maybe it’s just content, but it might be reviews. It might be votes. But it’s still controlled by a company. And so then Web3 is, well, if you were to build Yelp or Tripadvisor or Twitter on a blockchain, then the users would control it as well. And instead of all the value going to the shareholders in the company, the value would go to the users.

There’s first of all is what was Web1, Web2, Web3. Secondly, what was crypto one, two, and three. Then what’s open-source one, two, and three. This gets you this kind of very intoxicating, very messianic, very religious vision that you could completely remake the whole internet on this entirely new platform. It’s just as religious as, of course, the early internet was and early open source was.

Early open-source people thought they were going to destroy Microsoft and no one would ever buy software again. And early internet people thought that this was the end of government, and there’d be no war again, because everyone would understand each other.

Of course, the crazy ideology falls away, but open source runs the whole tech industry. The challenge is [to] take all that ideology and say, “OK, how is it that we take this from this very early, very rough, very interesting technology and make it into something that you can actually build production software on?” We’re five years away from building all of that.

The second challenge is, as you do that and you create all of that infrastructure and all of those abstraction layers on top, how do you do that without removing the whole point that it’s decentralized? Because it would be very easy to make all this stuff work really well if you just moved it into your own data center and controlled it. But then what would be the point? You just described SAP.

The challenge is, how do you make Twitter on a blockchain without you having to know all of these acronyms? But how do you remove the acronyms without doing that by just making another centralized piece of software, with one company or one group in control? And that’s the puzzle I wrestle with. We’ve got this explosion of complexity and sophistication.

Somebody I know in tech said, “All the dumbest people and all the cleverest people I know are all in on Web3.” Because you have this wave of noise, of bullshit, of argument, of people saying “It’s all a Ponzi scheme,” on the one side. On the other, people are saying, “This is going to end government and gold. And I need this in case the government steals my money and I need to move to Singapore. And it’s going to end trade. All trade unions will be built on the blockchain.”

There’s an enormous amount of noise that you have to try and pull apart and work out. But behind all of that noise, there’s this very, very powerful, interesting technology that lets you build internet applications, lets you build software in a different way. But right now, we don’t have that. We just have a lot of people building stuff that will let us have that in five years.

Michael Chui: It’s extremely interesting. A key tenet of the theology behind Web3 is the value of decentralization—is that fair as a way to think about why people are messianic about it?

Benedict Evans: Yeah.

Michael Chui: Who cares or wants that? I saw you post something about “any consumer product that requires you to write lines of code is going to be challenging in its adoption.” I’m paraphrasing.

Benedict Evans: There’s a strand of religion within tech of course that says, “No, no, no, consumers have to learn.” And so people hated GUIs and they hated the web, and those people hated the iPhone. “No, no, no, you should have to learn.”

Well, great, you can do that, but that’s not actually what’s going to happen. I think the challenge here is that this is an architecture that allows you to build different kinds of product.

There’s two slightly different parts to this. One of them is, open source didn’t exactly let you build software you couldn’t build before. It was just it turned out to be a better way of getting the software written. And then Apple could go and take the software and use it themselves, and they contribute to it. But you didn’t need to know that it was open source. You weren’t using it because it was open source. It was just a better model for producing open source.

It’s a bit like capitalism. There’s a moral question around capitalism, and then there’s an efficiency question around capitalism. It turns out a free market is just a better way to organize economic activity than central planning. It might be immoral, but functionally it works better. So that’s one side: open source just tended to be a better way of making software. And this gets also to questions of, should you have an open ecosystem or a closed ecosystem?

The other side of that is, does this actually let you make a different kind of software that you couldn’t make before? And of course, this is inherent to the web or the internet. Go back to what I was saying about the information superhighway.

The breakthrough of the web was effectively that you get this phrase “permissionless innovation.” That anybody could just make stuff and distribute it instantly to everybody on Earth. And you don’t need to get a carriage deal with every telco in every country. You don’t need to get a distribution deal. You don’t need to go and talk to a publishing company. What the internet did, you could say, is massively expanded the scope of the free market.

Because by analogy, you could say that telcos and publishing companies and so on were central planning. That they were this massive bottleneck on what you could do.

Michael Chui: I’m curious. Let me just step back a moment. We started the conversation a little bit about how you do what you do. You’ve described yourself as trying to work out which questions to ask. How do you work out which questions to ask about tech?

Benedict Evans: I’m trying to think about this. There’s an observation I made, a sort of joke that the moment you understand something in tech is generally the point that you should stop paying attention to it, because it’s become boring and it’s become well understood. This is kind of my fruit fly point.

Cars were a big deal for a long time. PCs were a big deal for 15 years, and then smartphones now. The bizarre thing about all the Apple App Store court cases is, we had all these arguments in 2010 or 2011—2011, probably. I could find the stuff I wrote as an analyst in 2011 and just republish it. It’s all the same questions.

But you just have to keep moving, and you have to say, “I’m not going to be the VoIP guy anymore,” because VoIP happened. It is not interesting in any way anymore. “I’m not going to be the smartphone person anymore,” because it happened.

We said everyone’s going to get connected, and we did. Now what? You have to push to the next question. I think the other side is, for me personally, I’m not a research group, I’m one person. I have to make a decision, are there places in that field where I’m going to be able to ask and answer the kinds of questions that I’m good at asking? Is it worth me investing three months or four months of my time to get to the point that I know what everybody knows?

I have to ask myself, if I do that, am I going to be able to say anything that people in this field don’t know? And is anyone else going to care? I could make a decision, how deep should I dig into games? How deep should I dig into ad tech?

Michael Chui: If I could sidebar for a moment. Last year, [in] your big annual presentation, there’s a section that says, “Tech becomes a regulated industry.” Is that still your view? Has it changed? Or are you saying certain parts of tech are becoming regulated?

Benedict Evans: The analogy that I use here often is that when we say—two steps to this. First of all, every company is subject to general legislation. If you kill somebody in an office, you don’t get arrested by the consultancy regulator. You just get arrested by the police. But then particular industries have very complex, specific issues that require specific regulation, like railways or food production or chemical refining or banking or cars. There’s specific rules about how you make a car, specific rules about how you can structure a bank.

I think in a general sense, technology is going to be a quote-unquote “regulated industry” in that sense. The problem is, if you then dig into, well, what does it mean when we say “regulate cars?” and think about that for a minute, well, hang on. We regulate, we tell General Motors to put airbags and seat belts in the car. But that’s got nothing to do with building light rail or how our tax policy accounts for building single-family homes versus apartment buildings near railway stations.

It’s got nothing to do with teenage boys getting drunk and driving too fast. It’s got nothing to do with, should we build more cycle lanes or not, and how should we fund that? If you actually think about how the world of policy has reacted to cars in the last 50 years, that’s, like, 30 different questions.

Some of those are antitrust, some of those are regulation of actual car companies, some of them are laws about how we drive that have got nothing to do with car companies. Some of them are about tax or urban planning, criminal law. There’s lots of different things in there, and they’re mostly complicated and full of trade-offs.

The difference, perhaps, is that we grew up with cars. It took 75 years to put seat belts in cars. And so if a politician says, “We are going to introduce legislation that makes big Detroit produce cars that cannot crash and gasoline that can’t burn, and we know that technology exists, and we’re going to make them do it, otherwise we’ll send them to prison or fine them,” we would all understand, you’re a moron. You’re asking them to invent new physics. That’s not on the list of options.

Whereas in technology, tech policy is just as hard as transport policy or education policy or health policy or anything else. But we don’t have that innate understanding of what’s the difference between “they don’t want to do that” and “that would be a terrible idea and this is why” and “no, you’re just asking for new maths.”

That’s the challenge that I think we have in looking at the regulation for technology. Now obviously, this could be an hourlong conversation, because there are deep, profound, intellectual questions about how we think about free speech online.

Because online is not the bar or living room or private conversation, nor a newspaper. And the world outside America does not have a First Amendment and is entirely happy to pass laws about speech. But what speech, and what does that mean?

There’s all sorts of interesting questions in competition law. How do you handle mergers in an industry that happens this quickly, that changes this quickly? You can’t spend five years doing the study. But also you can’t just guess in advance how the industry’s going to evolve. How do you handle that? How do you reconcile for privacy versus competition?

The privacy regulator says, “Make it really hard to export data,” and the competition regulator says, “Make it easy to export data.” And the engineers at Instagram or YouTube are, like, “Well, we’re engineers. We can do either, but you’re going to have to choose.” All of that is to say, “Yes, regulation, but that’s not one thing. That’s, like, 50 things.”

Michael Chui: We could go on, as you said, for a long time on any one of these topics and very many more. But if we could, if you would allow me to do a lightning round of quick questions, quick answers. You can feel free to pass if any of them are—you choose to.

Benedict Evans: Sure.

Michael Chui: All right, here we go. What is your favorite source of data that informs your thinking about tech?

Benedict Evans: Oh, I don’t think I have a good answer to that. I mean, I’m on Twitter a lot, probably too much. And then I am relentlessly curious. I think the real answer is Google, that I sit and I think, “I wonder what the answer to that question is.” And you will discover that the answer might come up in 30 seconds and there’s a perfect chart, or it might take you an hour, or you might discover that that data just isn’t there at all. And you can never tell in advance. But the joy of the internet, if I want to know what, I don’t know, distribution of GDP in China between large and small cities has been over the last 30 years, there they are. There’s a great 200-page McKinsey study or Goldman Sachs study or World Bank study. Like, “Wow, OK, there it is.”

Michael Chui: What is the most overhyped technology trend?

Benedict Evans: I would have to say Web3. But it’s also potentially underhyped. I can’t think of a technology in my career this polarizing inside tech that had that many deeply respected senior, very intelligent people in tech who think “This is nonsense,” and people who think, “No, this is the whole future of how the internet’s going to get built.”

Michael Chui: You might have just answered it, but just in case you have a different answer, what’s the most overlooked technology trend?

Benedict Evans: The, well, not the answer but an answer, I’ve been intrigued by this phrase “digital transformation,” because it sounds like a slogan you would see on a billboard in an airport for an enterprise software company or an outsourcing company or something.

What it actually means is that in the ’70s and ’80s big companies bought mainframes, and in the ’80s and ’90s they bought PCs and Oracle and SAP and client server and Windows. And now they’re going to the cloud. And that’s a 20-year—those are generational changes, and it happened, 20 years to make that transition.

IBM’s mainframe base is still growing, believe it or not. And when you do that, you change how the whole company works. Because when you went from adding machines to mainframes, you didn’t just change the infrastructure, you changed how the company worked.

When you went to Oracle, you changed how the whole company worked and what you could do and how you could run that business. And that gets you just-in-time supply chains and everything else that comes with that. And as we go to the cloud, you have an order of magnitude more software in a company, because it’s so much easier to have new software.

That brings machine learning and data, all kinds of new workflows and things. And it’s incredibly boring if you’re trying to write for a newspaper or write about this stuff. It’s a chart of enterprise software. Like, we’ve just fallen asleep. But it just changes how the economy works. There’s this whole thing for writing books about shipping containers. They’re really, really boring, but they change your world. And it’s kind of the same with SQL [structured query language]. It’s really, really boring, but it changes your world. And I think digital transformation or SaaS [software as a service] or cloud, again, it’s really, really boring. But it’s kind of a big deal as well.

Michael Chui: What worries you most about the evolution of technology going forward?

Benedict Evans: I don’t know. I’m very sanguine about this kind of stuff. I think there are always moral panics. There are always problems. The core problem with the internet is that we’re connected to everybody. And unfortunately that meant we connected all the bad people and all the problems and all of our worst instincts. But in general, despite being a sort of Burkean conservative, I do tend to believe in progress. The world does tend to get better, as long as we don’t screw it up.

Michael Chui: What excites you most about the evolution of technology?

Benedict Evans: I think the most exciting thing is seeing some new company or some new product that solves some problem that it never would have occurred to you existed. And you look at it, and you’re, like, “Oh, wow, that’s fantastically cool.” This is such a great solution to some problem that you’d never realized or never seen.

Michael Chui: What’s the last time that happened to you?

Benedict Evans: The one I talk about a fair bit, because I use it as a case study, is a company called Frame.io, which I saw a while ago, but it’s just front of mind.

Basically, if you’re professionals working on a piece of video, then there’s one person [to] edit it and 50 people who need to see it. And so that’s email threads and private Vimeo links and a Google Sheet with lots of comments and timecodes and version tracking and FedExed hard discs. And Frame.io takes it into a website. It’s basically Google Docs for video. You can’t edit it, actually edit it, but everything else.

So you can just scribble on that frame and say, “This needs to get fixed.” And you look at this and you think, “That’s just a giant”—it’s like this, if you’ve ever seen the people who created VisiCalc, the first spreadsheet, and they’re showing it to people.

People are looking at it and saying, “Your 30-second demo is a week of my life. Take my money now. I don’t care what it costs, take my money.” It’s like—magic paper, they call it. And over and over again, people do that. They take something that’s weeks of people’s lives, and it’s just, that’s just a button that’s just fixed that. And people go, “Wow.”

Michael Chui: Which company is the most interesting to track to understand how tech will evolve?

Benedict Evans: Tricky. I’m not sure I have a good answer to that. Because Apple points to one way that tech evolves, Google to a very different way that technology evolves. Facebook serves popular culture and consumer behavior. Amazon is an organizational miracle. It’s a machine that makes more Amazon. I think there’s a lot of people who are just sort of climbing the mountain from different sides.

Michael Chui: What would you be doing if you weren’t doing what you’re doing today?

Benedict Evans: If I wasn’t doing what I do for a living, I don’t know. I might have been an academic. I might have run a bookshop. I might have gone and done something else. I tend to think I’m probably institutionally challenged. So I doubt I’d be somewhere inside a big company. And I’m too easily bored and too impatient, but so I would be looking for places where I could try and answer questions.

Michael Chui: And what one piece of advice do you have for listeners of this podcast?

Benedict Evans: I don’t know. Curiosity perhaps. Somebody once asked the Duke of Westminster what advice he’d give to a young entrepreneur. And he said, “Have an ancestor who was a good friend of William the Conqueror.”

Michael Chui: Benedict Evans, thank you for taking some time to talk with us today.

Benedict Evans: Thank you.

Explore a career with us