Balaji Srinivasan (2:34)

They were going to water companies.

Benedict Evans (2:35)

Utilities.

Balaji Srinivasan (2:36)

Oh, utilities, yeah. Okay.

Benedict Evans (2:37)

They were going to connect everybody in the world. And then they did. And like now what? They were like Mark Andreessen's phrase. They were like the dog that caught the truck.

Balaji Srinivasan (2:43)

Right.

Benedict Evans (2:44)

And I went and worked in strategy and a bunch of media and telecom things. And yeah, I was analyzing, looking at smartphones because that was suddenly become the center of the industry and no one understood it now like it happened. I'm time to look for different questions.

Balaji Srinivasan (2:59)

Well, it's, it's funny because I think when we were overlapping, it was right in the middle of the. The smartphone dividend, the smartphone explosion. And just to, you know, we actually there's a few things. One is the smartphone dividend. That's a useful concept, right? Like that the rise of a billion smartphones meant that everything that went into them became cheaper. And that enabled VR headsets, that enabled drones. Right, all this stuff.

Benedict Evans (3:21)

Yeah. All the components that came out of it. Yeah. So smartphone sales are now from memory, like one and a quarter, one and a half billion units a year. And all supply chain from that, all of those components is then available off the shelf. If you want to buy 5,000 of them or 10,000 of them. All the WI fi chips and the batteries and the cameras and all the other bits. And before, if you wanted to put computer into something, you basically need to use PC components. So ATMs and so on are all basically PCs. Like elevators are basically PCs. And that has size and power and cost constraints. And then smartphones become the thing. And then all those components are available. And so that's what gets you drones and connected light bulbs and all the other bits and pieces around the edge of that.

Balaji Srinivasan (4:10)

One of the things that I think people don't appreciate is they think, for example, like the consumer. They think like the military has like special gear and it's got its own kind of supply chain. And often the military supply chain is often just a subset of the consumer supply chain because sell a billion units of this and Maybe you have 100,000 or a million units of a military thing.

Benedict Evans (4:29)

Actually, it's almost kind of the reverse now in that it used to be. So the way I think about this is like in the past, like before we were born, the intelligence agencies would get the cool new stuff first and then the Military would get it and then big corporations would get it and eventually consumers would get it like 30 years afterwards. So this is like the conversion.

Balaji Srinivasan (4:48)

Yes.

Benedict Evans (4:48)

It's like microwaves were invented for NASA and eventually consumers get them.

Balaji Srinivasan (4:52)

Yeah. Or like GPS was invented to guide missiles.

Benedict Evans (4:54)

Exactly.

Balaji Srinivasan (4:55)

And now it's used for tagging cat photos.

Benedict Evans (4:56)

And the shift is like a combination of the stuff getting cheap enough that it can be for consumers instead of you needing a billion dollars to have one, and then the scale of consumers once it gets cheap enough. And so now the way it works is the consumers get the new stuff and the military gets it 10 years later. Because that's how long it takes to.

Balaji Srinivasan (5:17)

The bureaucracy to assimilate that.

Benedict Evans (5:18)

A, the bureaucracy, B, to harden it and product it and turn it into what you need if it's going to get shot out or it's going to be cold or hot or warm or whatever it is.

Balaji Srinivasan (5:28)

Yeah, that's. It's funny. Does it really improve through that process? I know people think it does, but I'm not sure it does relative to the cost of not using the pretty good product versus whatever improvements come from the delay to harden it. I'm not sure if it actually, I don't know.

Benedict Evans (5:46)

But there's clearly, there's a sort of a process of you have to put it into a fighter jet. You don't replace the avionics in a fighter jet every six months.

Balaji Srinivasan (5:52)

Right.

Benedict Evans (5:53)

Well, but, you know, but yeah, you know, that's the kind of the core of it is the, the cutting edge of the innovation is for consumers and then that flows back through to everything else.

Balaji Srinivasan (6:01)

That's right. Well, you know, I was going to say maybe in China. You do. Maybe in China. Like, I think what happened with the consumer drones, they got good at quadcopters and that's led them to their new form. Have you seen Ehang? It's like the Chinese flying cars.

Benedict Evans (6:16)

Okay.

Balaji Srinivasan (6:16)

And you know, I played this clip like a year, year and a half ago and people said kind of, you know, the teal one, like, we want to fly in cars. We got 140 characters. And I was like, a lot of people didn't riff on that, but I was like, we want to fly in cars. We got them in Chinese characters. Okay.

Benedict Evans (6:31)

Yeah.

Balaji Srinivasan (6:32)

And the thing is when I put that up there, people are like, that's not a car. It's, you know, it's a, it's a copter. Right. That's not a car. It doesn't have wheels. But it solved the problem. Differently, Right. And actually I think was one of your lines, it's like unfair comparisons are often the best kind of comparisons. Right?

Benedict Evans (6:49)

Yeah. I remember seeing a bunch of flying cars when we were at Andreessen Horowitz, I think Mark Andreessen, he said it's like they're all like houseboats and the houseboat is a crap house and a crap boat.

Balaji Srinivasan (6:58)

Yes, that's right.

Benedict Evans (6:59)

And you know, think of it, thinking of it as a flying car is like the wrong term. It's better to think of it as like a much, a small, much better, much cheaper helicopter maybe, maybe.

Balaji Srinivasan (7:10)

But the point is that they. Now the other thing is it stands for short hops and like city to city where you fly over the traffic and. And they've got this. Uber was going to do this, by the way, before they decapitated Uber. Like the low altitude economy was something they were thinking about. And a lot of things get like cut off in the west and then they appear fully formed in China. Like consumer drones, for example. You know Chris Anderson, he was very early on drones and that got blocked by the faa and so consumer drones were hobbled in the us that's why DJI arose in China. So lots of things get blocked in the west and they arise in China because of that. Anyway, coming back up, so smartphones, I mean, I think you and Horace Didiompo who I was on his pod a while ago, I think you're two of the best. He's also like European or something like that.

Benedict Evans (7:59)

French. He was at Nokia. I mean there's an interesting kind of like information.

Balaji Srinivasan (8:03)

Do you know him?

Benedict Evans (8:03)

Yeah, he's a great guy. Part of it was, it was like. And there was a moment in time when there weren't many people doing mobile who really understood this and were industry analysts and were able to talk in public. So there were people inside Nokia or Goldman's or Bain or wherever who had all the data, but they couldn't publish the data and they weren't allowed to say stuff in public. Or if they were writing analysis, it was analysis for public markets, investors or something. And so there were very few people who were like, knew that you could go and take Apple's reports and make a chart of unit cells and make a chart of ASP and knew what ASP was, right. Or knew what ARPU was right. Now there's like an explosion of this. So there's huge numbers, particularly if you look at AI now there's like 10 people who do a really, really good 200 page deck of every possible AI chart.

Balaji Srinivasan (8:59)

Oh, is that right? Interesting. Yeah.

Benedict Evans (9:01)

And so that whole thing shifted, but at the time, yes, it was me and Horace and, like, Ben Bahar and the only strata, kind of. Jason. Exactly. There were, like, you know, a handful of people who understood this and could do the charts and were allowed to do the charts. And so that was sort of, you know, being at the right place, right time got me a lot of attention.

Balaji Srinivasan (9:23)

Yeah, it's interesting, I think, like, you and Ben Thompson of Stratecheri, and I'm not sure if Horace had a newsletter, but you guys were newsletters before Substack productized it. Sort of like Rogan was podcast before that became productized as a category. Um, and are you on Substack or were you on Ghost? No.

Benedict Evans (9:44)

Yeah, no, I'm.

Balaji Srinivasan (9:45)

You got your own custom.

Benedict Evans (9:46)

I'm still on my old Cobble together stack of mailchimp plus memberful plus Squarespace. Why.

Balaji Srinivasan (9:51)

Why don't you. You don't want to move to something. It's a pain to move.

Benedict Evans (9:54)

It's a heavy lift to move platform. And you sit and do the analysis and you're like, is this a good use of, like, a week of my time? Right. Maybe.

Balaji Srinivasan (10:03)

Maybe it might be. At this point, Substack's pretty good. But I mean, it's your.

Benedict Evans (10:05)

You know, you're. Well, there's. There's a separate substack thing, which is, do you want it to be on your. Your newsletter or your substack?

Balaji Srinivasan (10:13)

Yes, that's true. Yeah.

Benedict Evans (10:13)

Because it's a platform and you get the advantage of. I mean, this is something we can talk about. You know, it's Quiztix in this line of come for the tool, stay for the network. You go on Substack, they will get you new subscribers. Ghost won't get you subscribers. On the other hand, now they control who your readers are and you don't, which is always the thing of a network.

Balaji Srinivasan (10:31)

Well, I mean, you should still mail.

Benedict Evans (10:32)

Out to everybody you do, but then they're trying to get you to use their website and their algorithm to decide who reads what. That's true. So there's always these kind of questions, like, do you want to go with the people who will give you an audience? And in exchange for that, they're deciding that they'll give you the audience.

Balaji Srinivasan (10:48)

There's always a trade off with the distribution. Yeah, I think Ghost is another option.

Benedict Evans (10:52)

And I think both Ghost and Beehive are the two others that people use.

Balaji Srinivasan (10:57)

Ghost is like, you know, I saw Ghost rooms very early and I just thought it was so good for what it was. Like it was. I mean, not even for it. It's just a very polished thought through. For an open source product, it's unusually polished. John Nolan's very, very good. Um, it's funny, you know, like on that there's a bunch of things we could talk about, but the, the whole newsletter thing, it's sometimes there's things that are like newsletters or podcasts that are what I consider lowercase in technology before they become uppercase. Like for example, Odeo, you know, like what Twitter was. Twitter was a podcasting company before it became Twitter. And the time constant, they just got the time constant wrong. Where you needed, which is hard to predict that microblogging would take off first and then it required like AirPods and everything being online for a long time. And maybe, you know, Covid, before podcasts really exploded and the term was around in lowercase.

Benedict Evans (11:52)

You could even argue it's needed like 5G or 4G, something like that.

Balaji Srinivasan (11:55)

Yes.

Benedict Evans (11:56)

If you're listening to it on the car, in the car, then you need a half fast enough network.

Balaji Srinivasan (12:00)

Yes. Bandwidth is a constraint. Yes.

Benedict Evans (12:03)

And then the time works.

Balaji Srinivasan (12:04)

So what do you think is lowercase today that's going to become uppercase? Like what exists in tech that people are like, oh yeah, that exists, but it's going to go big. I have some ideas. I want to hear yours.

Benedict Evans (12:16)

Interesting question. I think there's probably the answer. If I was a consultant and trying to whiteboard this is, I would be looking around AI because that's a new platform and there's a lot. All the old white space got filled in and now you've got a whole bunch of new white space. So deterministically there should be a bunch of those things here. AI, which I'm sure we'll talk about, does feel very sort of mid-90s in that you're like mid-90s Internet in that. Like, well, is this a browser? How do you use it? What's it for? How would you get to it? How does this work? Where's the value capture going to be? I'm not sure that there's like maybe one answer is like, I'm too old and I'm not spending too much time looking for weird stuff around the edges. The last one of these that I spotted personally was Shein, which Shine?

Balaji Srinivasan (13:12)

Is it Shein or Shine?

Benedict Evans (13:13)

I'm told it's Shein.

Balaji Srinivasan (13:14)

Is that right?

Benedict Evans (13:14)

Speaking to people there, I haven't worked out her team. That was an interesting one. That it was. Maybe you could also say it was the last of the ones that you could spot because suddenly. Wait, what is this thing that's at the top of the iPhone app store chart all the time?

Balaji Srinivasan (13:28)

Oh, I see. Yeah.

Benedict Evans (13:29)

Suddenly that thing exploded and that's like probably the largest apparel, pure play apparel retailer on earth.

Balaji Srinivasan (13:36)

Yeah. And like Shein and Temu. Yeah, that's right. They're now getting hit with the tariff stuff and, you know.

Benedict Evans (13:42)

Yeah. Tariffs plus, plus a de minimis rule.

Balaji Srinivasan (13:46)

In the US that's only the US market. And that's not, you know, I don't know what fraction of their sales that is.

Benedict Evans (13:50)

Yeah, it's like a third of their sales or whole quarter of their sales or something. So that was like. That was. That was a thing that was interesting. I'm not sure. There's not like a new thing that I'm watching that I've noticed recently. I'm sure there will be. You know, I keep looking.

Balaji Srinivasan (14:03)

So I have a few. We were talking about the glasses. Right. Like, I think smart glasses are sort of like the most predictable thing after the iPhone.

Benedict Evans (14:11)

Oh, yeah. I'd put it in a different category. I was sort of thinking, like, what stuff that's being used now that people haven't quite noticed is being used yet.

Balaji Srinivasan (14:16)

I see. So.

Benedict Evans (14:17)

Well, I guess glass is definitely your next thing.

Balaji Srinivasan (14:20)

Sure. So I guess I would sort of bundle VR headsets, AR headsets, you know, like that with glasses and say that that's just. Glasses are sort of the next version for goggles. But. Okay, so that's one that, that, that we agree on. Except the question is, as you said, is it going to be watches or phones? How big does that get? Right. Yeah, I think, I think, you know, just like podcast grew to me and like a video podcast and so on. The robot dogs are interesting. They are fun to play with and they're getting way cheaper now.

Benedict Evans (14:55)

Right.

Balaji Srinivasan (14:55)

They went from the Boston Dynamics kind of things. So the home robot as a toy, I think is probably going to become more and more popular. Like a Christmas present kind of thing at first. Right. Because I see kids playing with them and they just love them just as a toy. And the, you know, it's kind of like the robot dog, the drone as like a starting to become a, like a Christmas present kind of thing. I think that that becomes a thing. And eventually, like we were talking about this at the Museum of the Future in the uae. They clad these so that it doesn't. It's not just like a skeleton of a robot. Dog, but it actually looks like an animal. And that completely changes your perception of. So I think that that'll be a thing. And with respect to AI and so let's do. I mean there's AI, there's bitcoin, there's China, there's drones, there's biotech. There's actually several different areas that I'm tracking. I'm tracking a bunch of these various singularities, whatever. They're not all really, actually singularities in the technical sense of going to infinity. Ramps, Curves. Curves, that's right, yeah. With AI, there's, you know, one way of thinking about it is like now we're two and a half years in, let's say, let's call the ChatGPT moment, right?

Benedict Evans (16:03)

Yeah.

Balaji Srinivasan (16:03)

And it's interesting because it, I think what people really overestimated was how much it's agentic intelligence versus amplified intelligence like that. Say you still have to prompt it. So prompting is like higher level programming, number one. You still have to verify the output. And that means you kind of need to know what it is you're looking for. For example, if it spits out a bunch of mathematical symbols in an area of math that you don't know, then you have to be Terence Tao to verify it. It might be gibberish, it might be real, who knows, right. And so the prompting and verifying are actually the bottlenecks in many areas. Now Karpathy and I, you know, the Andrej Karpathi, we were just having a discussion on this like a week or so ago. And the thing about verifying is if you're using the GPUs that we have built in and you're looking at images or video or front end code, right? Like a user interface, your eye can just instantly pick out and you can verify pretty quickly. So for that side of things, AI is quite good. Anything that's images, video, your ear can also pick out audio, right? And front end. But when it's back end stuff, right? When it's like database code, when it's like crypto, when it's mathematical equations that you don't have, like GPUs, you can't just like hit it with your eyes and quickly detect it, right? That whether it's, whether it's correct or not, you have to deep read it carefully, right? So it can generate reams of text, but then you have to verify it. Exactly, that's right. Maybe you have some thoughts on that.

Benedict Evans (17:37)

Well, so it's funny, I was talking to John Bolthwaite the other day and he said, benedict, you think in slides. So that.

Balaji Srinivasan (17:43)

We do, too. We both think in slides.

Benedict Evans (17:44)

So I have a slide. Yes, and maybe there's a sort of. I'll talk about the slide, and this is an observation around it. I think a lot of discussion of LLMs is sort of hunting for the, like, what's the right. The right way to conceptualize this? It's like with machine learning, the right way to conceptualize it with this is pattern recognition. And we're still sort of hunting for the right way to conceptualize LLMs. The slide is that traditional software is deterministic and does things that are easy to explain to machines. In fact, automation, machine tools, sewing machines, typewriters, adding machines, things that are easy to explain to a computer. There may be things that are very hard for people to do, but they're easy to explain. So it's hard for you to drill a hole 100 times or to calculate a mortgage in your head, but it's easy for you to write down the logical steps to explain how you do this. So that's traditional software like databases, data processing, the whole 60s, 70s mainframe thing. Machine learning is stuff that's hard to explain to a computer. So it's hard to explain why that credit card transaction is weird or how.

Balaji Srinivasan (18:42)

To move your hands or something?

Benedict Evans (18:43)

Yeah, it's hard to explain why that's a picture of a dog and not a cat. You think it's easy until you try and do it. And then it's like you tried to make a mechanical horse. It always folds over until robotics comes along. So that was machine learning. I also think that as a kind of quiz for you, do you think machine learning is still AI, or is that now just software?

Balaji Srinivasan (19:00)

Well, so the way that I think.

Benedict Evans (19:02)

There'S a process that once it's been around for a while, it's not AI anymore.

Balaji Srinivasan (19:05)

It's funny. So I think within the field, technically, the division would be machine learning would be everything up to linear logistic regression and SVMs, all that kind of stuff. And then right at the point you start doing deep learning and you have large neural networks now you start getting into what people would call modern AI. So ML is almost like the boundary of understandability, you might say, where you can write clean equations and really understand what's going on. And to me, the most surprising and confusing, I still don't feel I know what the phenomenon is, but I still find it magical, is something called the double descent problem. Do you know what that is? Basically, normally when you're fitting to data you want to have the fewest possible parameters because you can overfit. Right. And so your error goes down and then your error starts going up on the holdout set. Because you train your model in machine learning and you want the minimum number of parameters to be able to explain the training data and predict the test data. And if you overfit, then you're no longer predicting out of sampled stuff. But double descent is when you do AI, you get actually a second wind when you start going to a very highly parameterized model and the error actually drops again. Right. And. Which is just a really weird phenomenon that there's papers on this and so on. And it's one of the most counterintuitive things about the whole thing that just having these gigantically parameterized models would generalize. Well, right, because it violates. That's the biggest difference. Go ahead. There's other things people might see as the biggest difference.

Benedict Evans (20:52)

I think one of the ways I think about the term AI is that people kind of use it like technology. The word technology.

Balaji Srinivasan (21:00)

Yeah, that's right.

Benedict Evans (21:01)

That anything new is technology. Anything your parents had is technology.

Balaji Srinivasan (21:05)

I'm a stickler for precision.

Benedict Evans (21:06)

So there's different ways that you can say, what do we mean by the word AI? I feel like AI has almost become like the word metaverse, where like, you don't know what somebody means when they say it. But to continue my slide. So the first point is there's deterministic software, which is stuff that's easy to explain. There's machine learning, which is stuff that was hard to explain, which basic machine learning salt is. And now an LLM is maybe stuff that's easy to explain to an intern. So it's something where if you had to go away and have a, like a kickoff meeting and spend half an hour working out how we're going to do this project, then an LLM probably can't do that. But if it's something that you could explain in 10 seconds or 20 seconds, then an LLM is going to be able to do that. And part of the problem is, are you able to explain it even to yourself? Could you explain it to another human being? Can you actually kind of shut your eyes and conceptualize how is it that I'm going to explain what it is that I want this thing to do?

Balaji Srinivasan (22:04)

It's. So what you're saying is very important because, you know, the. There's like several different angles I can, I can. I want to take off that. You know, in one sense, I had this tweet we're living in the age of the phrase, right? So the prompt for the AI, or the 140 character tweet, or actually in crypto, like 14 words, 13 words, 12 words can be your crypto reset phrase. Right? These are phrases of power in AI, in social and crypto, Right? Like this strings of characters that do a lot, you know, and they're spells. They're spells, right? And the thing about it is, the crisper you are as a manager. Like, you know, if you're, if you're. If you're a really good engineering manager, you're great at prompting AI because crucially, you don't just say, hey, code this. You say, hey, you know, try and use, you know, React for this. You can use REACT native for, you know, the iOS and Android interfaces. Use tailwind.

Benedict Evans (23:04)

Use it.

Balaji Srinivasan (23:05)

The more, in a sense, vocabulary terms you have, the better. You can prompt something with. And you have to use those vocabulary terms correctly. And what that meant is, for example, I realized with Dall? E, when that was first, you know, before the ChatGPT moment, I was like, wow, art history is now an applied subject. Knowing, like, Cezanne and Picasso and, and what, you know, these various kinds of obscure styles, suddenly you can be like, boom, style it like this, style it like this. And it'll do that. Right.

Benedict Evans (23:28)

You could say the same thing for music. Like, what exactly is it that's being done there? I knew there is a word for that. So. And you can.

Balaji Srinivasan (23:36)

That's right, exactly. So you can upload a track and you can say, what style is this? How do you caption this? Right. Have you ever seen, you know, like, the restaurants with the fancy menus and they don't say. They don't say tomatoes, they say, like heirloom.

Benedict Evans (23:52)

There's a word, there's things that are theoretically subjective.

Balaji Srinivasan (23:56)

Yeah.

Benedict Evans (23:56)

But there are, within the provisions, there is a particular term for doing that particular thing.

Balaji Srinivasan (24:01)

Exactly. It's like the difference between, like, red versus burgundy and, you know, crimson and what have you. They've got precise words which mean something, and then you can summon greater precision with those precise words. And so what a way I was thinking about what you're saying is that. And I've written about this, AI is like undocumented APIs. So normal API, every function is written out. And it's like, you can do this and you can do that. And So I got 20 functions and here's everything is there. Right. With AI, it can do lots of things that even the people who wrote it up. So it's much more mysterious as to what it can do. You just have to try things. Right.

Benedict Evans (24:37)

So the way I was thinking about this from a different angle was to think about GUIs.

Balaji Srinivasan (24:41)

Oh yeah.

Benedict Evans (24:42)

What a GUI is doing. There's several things that a GUI is doing. One of them is it's telling you all the features that the developers have created. And part of the reason that was a revolution is A, you knew what they were and you didn't need to memorize keyboard commands. But B, you can actually have more stuff because you're not constrained by the number of keyboard commands that you can write down. So you can have hundreds of functions instead of like, you can just put them all. You can just add more shit to the menus. Yes, but the other part of it is that the GUI is telling the user a whole bunch of accumulated decision and institutional knowledge at. About what the right things to do at this point would be.

Balaji Srinivasan (25:15)

Yes, that's right.

Benedict Evans (25:16)

And so if you're in a workflow, as opposed to just a blank screen, it's one thing if you're in like Photoshop or Excel, yeah, it can prompt.

Balaji Srinivasan (25:23)

You on the prompt.

Benedict Evans (25:23)

But if you're in a workflow in Salesforce, then there's a decision taken that says, I'm going to offer the user these five options here and not 750 options. And with a prompt you don't have any of that. So you've got to shut your eyes and think for a minute of like, well, what would I do here? And you don't have that help.

Balaji Srinivasan (25:44)

This is, you know, Karpathy has talked about this also, but I do think there's room for aios, right? Like in a sense, and we can talk about crypto in a second. But I think AI and crypto are both actually operating system level innovations. And for example, maybe someone just does it as an app or like a downloadable thing and just does as a layer on top of the Mac. But, but if you have the full context of all the actions that are happening on your Mac, you can suggest which apps to use. Suggest which apps to download. Suggest. Hey, you probably want to change these keyboard settings. And so like there's, you know, it's funny to put it this way, but Clippy is finally vindicated. Clippy, but for everything. Right? And because Clippy can now be really, really, really, really smart. Right? Like, you know, it was Anderson's line. It's like everything in tech works, it's just when. Right? And even the thing that's interesting about the Clippy thing is somebody also made a point, which is that you actually want to put faces on your AI avatars on your AI agents. So you could pick from Clippy or 10 other kinds of things. And the reason you want to do that, this is counterintuitive, but people like you and I can use ChatGPT and Claude and what have you because we're familiar with interfaces, but the reason they're actually intuitive to 100 million people is they're used to chatting with another human on the other side. So they're already modeling the chat box as being a human like response because They've been using WhatsApp or Facebook messenger or Instagram chat or something like that for a long time. Right. But when it's outside of that chat box environment and it's like suggesting on the screen, you kind of want a face to pop up so they can associate. Okay, this person is suggesting this because that's who they are and they kind of map that personality onto the AI agent. And, and so you could choose from different kinds of clippies that would give you prompts on what to do. Or. Or it just does it for you. That's another possibility. But I don't think people like it when it does it. They want, they want to be able to approve it before they do it.

Benedict Evans (27:48)

I think there's a sort of sense in here of how people conceptualize what this thing is and how it works. I remember John Protheroe at Google showing me a chart, a Google Trends chart of best versus Cheap, Best versus cheap.

Balaji Srinivasan (28:06)

Yeah.

Benedict Evans (28:06)

So the best does this and cheap does that.

Balaji Srinivasan (28:09)

And what are the axes crossing over time?

Benedict Evans (28:12)

So Google Trends. So what's the frequency of the word best?

Balaji Srinivasan (28:15)

So it starts with like cheap phones and it goes to best phones.

Benedict Evans (28:18)

Yes. And so the thesis was that this was shifting from the Internet as price comparison, where you'd already knew what you wanted. And it's at the top of the. And that's at the bottom of the funnel to the Internet as recommendation curation suggestion where you're looking for suggestions.

Balaji Srinivasan (28:33)

So interesting. So let me see if I can understand the psychology. So when you look at it.

Benedict Evans (28:37)

So it starts from, it starts from 2004.

Balaji Srinivasan (28:39)

Okay.

Benedict Evans (28:40)

In 2004 you go on the Internet and you already know what you want and you look for the cheap. What is the cheap X? And then you put in a scoop or you put in a product or something. Whereas over time that goes down and best goes up and best goes up and crosses it. It's a perfect X on the chart, unfortunately. And the thesis is you're going further up the funnel. You're looking more and more for. I want someone on the Internet to tell me the best X or Y. Where previously you'd have got that from the magazine or newspaper or something.

Balaji Srinivasan (29:11)

There's two or three. There's two things about that. The first is, you know, Andy Grove's thing about the paired metrics. So, Andy Grove, whenever anybody's optimizing, like sales, for example, they will usually start recruiting. They'll start by optimizing quantity. But there's, you know, you can sometimes optimize quantity and then quality drops off.

Benedict Evans (29:30)

Yeah, yeah, right.

Balaji Srinivasan (29:31)

So quantity is easy to measure. It's like just a number of people we hired or whatever, but quality is how good were they. Right. And so that's the second paired metric is usually a quality metric that. And so quantity is cheap. Right. And people start with cheap and then quality is best and they go to best. So that's another lens on this, a third lens. What I thought, my explanation, maybe it's different than what actually happened, was when people are just trying out a space, they just want the cheap version to try it out. And once they've committed to a space, and like, for example, the cheap digital camera, cheap, cheap drone or something like that, I want to try it out, right? And they want to try it out at low cost. Try before you buy. And then once they're committed to a space, then they're like, I want the best drone out there now, because I want to.

Benedict Evans (30:14)

Well, the analysis then would be cheap drone versus best drone.

Balaji Srinivasan (30:17)

Right.

Benedict Evans (30:18)

But I think the. There.

Balaji Srinivasan (30:20)

That's what I thought you were saying. Yeah, but you're saying cheaper best overall.

Benedict Evans (30:23)

Yes, yes, overall.

Balaji Srinivasan (30:24)

But I'd love to see a category by category. I wouldn't be surprised to see that happen. Category by category. But maybe not.

Benedict Evans (30:29)

Well, there's a different plot point there, which is sort of what I was talking about in our panel this morning, which is there's infinite product. So how do you know what to buy? And it used to be that you'd start with a magazine and then you go to the Internet to find the cheap place to buy it or you know, what you wanted. And now you go to the Internet to figure out, like, what's the right place to do this. So the Internet has become much more kind of a default. But actually the thing, the thought that prompted me to that was you can also go and play with Google Trends. And I did a chart, played with like, how, why, where, what? Like more kind of basic Questions. And you really need to be inside Google to do that analysis properly.

Balaji Srinivasan (31:06)

Exactly.

Benedict Evans (31:07)

But it's that sense of how much are people doing conversational queries into Google as opposed to typing keywords into Google and things that are not really a Google query, like what is a is not probably doesn't help Google, but that's still how people use it.

Balaji Srinivasan (31:22)

People are trained for years to not to remove all prepositions, remove all that stuff and just do keyword ease. And now we're trained the opposite, to write full and complete English sentences. Like prompting is the new searching, but it's a completely different behavior. Right, go ahead.

Benedict Evans (31:39)

Well, I was going to say this is one of the. There's a sort of tangential point to that. Like one of the, like the early, easy, obvious things that people have deployed with AI with LLMs on the Internet is different kinds of sort of natural language queries or different, not so much natural language, but like different kinds of query. So the canonical one people talk about is Walmart saying, now you can search for what should I buy to take on a Picnic, which isn't a database query. And for Google, for Walmart or for Amazon five years ago, that search just wouldn't work. Because unless there's a product that's like tagged with Picnic, it's not going to come up. Whereas now there's an LLM with a world model that has some sense of how you might answer that question.

Balaji Srinivasan (32:21)

Yes. Is it a world model? It is at least a web model.

Benedict Evans (32:24)

It's a different kind of query. Anyway, you're not doing a SQL query, you're doing something else.

Balaji Srinivasan (32:29)

That's right. And I think one of the things that's interesting is computers are. We knew they were very good at that first kind of deterministic computation, the SQL query, the calculation. That's what they're built for. Doing math, Right. And now they've gotten good at probabilistic kinds of things. Right. So this would be like system one and system two thinking. Right. Probabilistic is like the quick impression and then this is like the logical calculation. So it's actually good at. The thing that's harder for humans is the, you know, like long involved mathematical condition can do that errorlessly and now it can also do the other kind of thing. And so it does suggest that there would be some synthesis of that eventually where an AI can. I mean this is like AI tool use or what have you. Like, it detects that it needs to go to system one and it starts invoking Python for that and this is getting better, but it's surprisingly not amazing. Two and a half years in, right when it needs to go deterministic.

Benedict Evans (33:27)

Well, so I, I wrote. Last long thing I wrote about this was about looking at deep research, which OpenAI launched. And one of the kind of traps in looking at the new thing is to test it based on what was important to the old thing. So you know to look at the Apple II and say, does this match the uptime of a mainframe? No. So it's useless. Well, no, but that's not the right question.

Balaji Srinivasan (33:50)

Yes.

Benedict Evans (33:51)

Can you write and build an Excel model on an iPhone? No, but that's not the point. It still replace PCs and, and the reason I mentioned this is so, so, so, so Deep Research, Open Air launch this thing and it's what everybody was, $100 a month or whatever. So I. But they look at the marketing page and the marketing page shows it doing a research project about mobile, which as we said, I know a lot about. And it got the answers wrong. And it, that's verifying.

Balaji Srinivasan (34:18)

See, you knew you could tell that it was wrong.

Benedict Evans (34:20)

It looked exactly. So this is the thing. And it got stuff wrong in several levels. People can remembering now what I wrote like two months ago. And so there was a specific. It was make a table which shows mobile smartphone adoption in a bunch of countries and then the operating system market share. And then this is like an intern teaching moment because first of all, what does adoption mean? Does that mean unit sales share installed base app store sales? Like what metrics specific are you asking me for? Then it had given a source for the number it had come up with, which was Statista. And Statista is an aggregator that steals other people's data and republishes it. And when you jump through a bunch of registration hoops, you discover that the actual source was I think Kantar. Kantar. It's an ad agent, it's part of GroupM. I thought it was part of one of their. It's consumer survey data. So it's a crop company. So it was actual proper consumer survey data. But the two things. So then when you go to the Canton chart page, you discover that Deep Research had got the numbers the opposite. So it had flip percentages.

Balaji Srinivasan (35:28)

I see.

Benedict Evans (35:29)

And then it had also said right.

Balaji Srinivasan (35:31)

Because it didn't ab action.

Benedict Evans (35:34)

It had copied them from the website wrong.

Balaji Srinivasan (35:36)

I see.

Benedict Evans (35:37)

And then the other source it gave was statcounter. And statcounter was just using the same wrong data, which is a traffic measure. So that's not Going to tell you what options because high end phones get used more and iPhones get used more. And there's a bunch of things in here where you'd like. This is what I'd expect from an intern. I would go back and say, no, this is what I mean by adoption and this is a good data source and that isn't. And it's like a great first version. The problem is a, I had to copy the number out wrong, which is not what I would expect from an intern, or at least not a good intern. But secondly, I'd have to be a mobile analyst to know any of these things.

Balaji Srinivasan (36:14)

And that's a verifying thing that I was getting at.

Benedict Evans (36:16)

This is the kind of, the core of it is I all these people were looking at deep research and saying this is fantastic for researching things you don't know anything about.

Balaji Srinivasan (36:23)

And I was like, no, no it's not.

Benedict Evans (36:24)

Yeah, it's fantastic if you need a bunch of material about something you know a lot about.

Balaji Srinivasan (36:29)

Exactly. So that's thing is, that's why I think AI in its current incarnation is better thought of as amplified intelligence. Because the better, the more you know about a field, the better you are at prompting because you've got better vocabulary and the better you are at verifying because you know more facts about it and you have more cross cutting checks and that is less true for the visual area. But just identifying that is a very important limitation where you have a completely different system you can use for the visual stuff, which is just your eyes, right. You don't have to use the, you know, we have just different hardware for quickly seeing, you know, this way, the hands or something like that. Whereas if that was.

Benedict Evans (37:04)

It's a monkey brain.

Balaji Srinivasan (37:05)

It's a monkey brain, exactly right. So that's now an interesting question. This is, you know, Karpathi and I were discussing this is is there some way to turn some or a subset of of the non visual things into visual cues where you could see it was wrong immediately. So I'll give you a small simple example. Let's say it generated an audio file, right? You know, like a spectrogram of an audio file. Right. You could maybe immediately see if there's some artifact there. Right. That's a trivial example.

Benedict Evans (37:31)

So I think this, it's a fascinating concept. I would wonder whether that's the right split.

Balaji Srinivasan (37:39)

Okay, it's at least one split I found useful for now. But what are you thinking?

Benedict Evans (37:42)

Well, so the split I was thinking was that the natural language generation to make text is perfect so the text is always grammatically correct.

Balaji Srinivasan (37:52)

That is true. Yes.

Benedict Evans (37:52)

But the model underneath, like the facts presented by the nut in the text, might be wrong.

Balaji Srinivasan (37:57)

Yes.

Benedict Evans (37:58)

And that's sort of deceptive to us because we see the text is correct and it looks confident.

Balaji Srinivasan (38:03)

Yeah, that's right.

Benedict Evans (38:03)

Whereas in an image, like you ask it for a picture of somebody and everything's perfect except the person's got six hands. Mm. I'm not sure conceptually, what is it that that's flattened? Is that you're seeing two things in one layer, or is it that. Do you see what I mean?

Balaji Srinivasan (38:25)

I see what you mean. I think there is.

Benedict Evans (38:27)

Or is it that it's a different level of. Well, maybe then maybe there's a different point here, which is if you ask for an image of a car.

Balaji Srinivasan (38:35)

Yeah.

Benedict Evans (38:36)

And the car. I actually did this ask for a fantasy 1960s French sports car. It will look French. It will look like a sports car. It will have four wheels. It might have two steering wheels.

Balaji Srinivasan (38:47)

Yes, that's right.

Benedict Evans (38:47)

The two steering wheels is the equivalent of a grammatical mistake or spelling mistake in the text generator.

Balaji Srinivasan (38:53)

Yes.

Benedict Evans (38:53)

Because, however, it may also be that the balance of the car is all wrong and it would flip over if it tried to go around a corner. But you'd have to be an automotive expert to know that. Yes. So I'm saying there's like, there's different levels of error.

Balaji Srinivasan (39:07)

That's right. What you're saying is the two steering wheels is like a spelling error. But spelling errors are very rare for AI, Whereas the two steering wheels is a common error. Right. And I think that has to do with just the way diffusion models work versus how transformers work. That'd be like, one high level answer I'd give where it's doing, like, kind of. It's more local with the diffusion model, and you can be locally correct with the steering wheel, but globally incorrect. Whereas locally correct with spelling is usually correct. That's like, maybe once.

Benedict Evans (39:42)

Yeah, I'd say useful.

Balaji Srinivasan (39:43)

That's one answer. The second is that there's only a small space. I think, like, for example, we are optimized to recognize faces, so we can detect very subtle differences in faces. But if I gave you, like, five different sheets of, like, static noise. Right. Even if there are very clear patterns, like math, mathematically, like, these are all, like, Fourier transforms of the same object. And this is the one at. They just look like total noise to you. A computer would be like, these 12 are the same, and this one is the odd one out.

Benedict Evans (40:13)

Right.

Balaji Srinivasan (40:14)

So in a Sense our eyes are optimized for a very low dimensional set of things which are the things that occur in the real world. Like those are the things we can pick out. Right.

Benedict Evans (40:23)

Which is also that our eyes are like dogs are better at motion than us.

Balaji Srinivasan (40:26)

Yeah, exactly.

Benedict Evans (40:27)

Even eyes are different depending on the species.

Balaji Srinivasan (40:29)

That's right. So because of that we actually have a. Like. Because they can't detect patterns in static. That's like too high dimensional space. I think text is kind of like that because it can describe one of the most, I mean, surprising things to me about how AI has evolved. We were talking about this question before is I was surprised you could get so much mileage out of pure text. The reason is so much what?

Benedict Evans (40:54)

Sorry?

Balaji Srinivasan (40:55)

So much mileage out of pure text.

Benedict Evans (40:57)

Right.

Balaji Srinivasan (40:59)

And the reason I was surprised by that is, you know, you'd think you.

Benedict Evans (41:04)

Meaning like reasoning and all stuff that looks like reasoning.

Balaji Srinivasan (41:06)

Reasoning and also spatial manipulation, like, like picking, like, like having cameras, having eyes, seeing the world, reasoning about it like a baby and so and so forth. It is amazing how much of that world humans have assigned machine readable labels to with text. And the way that, you know, it's just, it's just very surprising how well that worked. Like language, what I'm trying to say is in a few, in like 40 words, you can describe. It's like code, you can describe many, many, many different kinds of things in like 40 words. Right. And. And it's just more general. It's one of those things where if you're. Sometimes when you're really close to a space, you're actually more surprised by a breakthrough than if you're farther away. And I should say like, you know, even looking, seeing all the style transfer stuff in the mid 2010s and seeing Imagenet and seeing the benchmarks and so on and so forth, I was surprised that it got beyond, you know, what a Markov chain is. Well, if you saw the stuff before GPT3, right. It was like semi coherent, but it didn't look like it was converging on something, you know, it just looked like, you know, it repeat itself many times and what have you. And the fact that it broke through to what it did just based on language was so counterintuitive. And it's. I think it's because it's such a high dimensional thing. It captures so many different aspects of the world. Like anything you can perceive in the world, there's a word for it, there's many words for it. And then we also have billions of people who've been typing those words for two decades. Right. So in a sense like the entire Internet, the video games and social media were like this boom boot bootstrapper for, for AI. Anyway, so on the, on the other hand, AI is very bad at spatial stuff. You know, this thing called arc. Francois Chollet has this benchmark.

Benedict Evans (42:59)

Oh yeah, that, yeah.

Balaji Srinivasan (43:01)

And so he is, he has benchmark that actually got beaten by the recent, you know, you know, chatgpt release and he's got like a new one. And it's, it's almost like a tetrisy kind of thing that's got some degree of logic and spatial type stuff that AI finds it hard but humans still find it eas. It's kind of like maybe the next generation Captcha. And it's, it's visual more than it is verbal. Right. So for every reason, is it something.

Benedict Evans (43:30)

That would be hard to explain in words?

Balaji Srinivasan (43:34)

Yes, I think kind of, it's, it's about like this is here and it's almost like Minesweeper, you know, Minesweeper, where you click and it expands and so, and so forth. I, I think AI because it started with words, it doesn't do well with the spatial side of things. Now on the other hand, what the Chinese are working on in particular is physical robotics. Obviously Elon's working on it and so on and so forth, but China's way ahead on the, on the physical supply chain. So like physical AI is robots and those definitely have cameras and XYZ and spatial and rotation and so and so forth. So there's some, some eventual fusion. You know, like the, the self driving cars are gathered hundreds of millions, billions of miles at this point. So there's some fusion of the web which is words and the world which is, you know, spatial. That will get you like a completely, you know, maybe a fusion set where it can reason about the world as it is. It knows how tall Everest is because someone, some robot has hiked it. You know, like Google, Street View. You might eventually imagine a bunch of humanoids walking the world just like that.

Benedict Evans (44:40)

You know, I wrote a thing years ago about Street View and Yahoo. And the sort of thing I was kind of poking away at is that basically every big Internet system is a mechanical turk.

Balaji Srinivasan (44:52)

Ah.

Benedict Evans (44:52)

And the question is, where do you put the people?

Balaji Srinivasan (44:54)

Where there's humans. Yes. Yeah.

Benedict Evans (44:55)

And with Google Search, the people are everybody, A, everybody making a link on a webpage and B, everybody using Google.

Balaji Srinivasan (45:04)

That's true.

Benedict Evans (45:04)

Whereas with Yahoo, they tried to like have a bunch of people in an.

Balaji Srinivasan (45:08)

Office yeah, doing it in the middle.

Benedict Evans (45:09)

Making a hierarchical list of all the websites on the Internet, which became impossible. And with Street View you just pay a bunch of people to drive down every street in the world, which is actually not impossible. It's just expensive.

Balaji Srinivasan (45:18)

It's just expensive. It's really, it's an interesting computation. It's not obvious that it would be feasible to. It's funny, you know, the Yahoo thing. Yahoo, you know, I think got started in like the early mid-90s, right. I think 94ish, 93, something like that. And the thing about it is it like Yahoo had to kind of get to its limit before it was obvious that you needed something like Google because like webpages had to be suffused with at the time they put on page spam and so on and so forth. You had to kind of top out. You had to get enough web pages that the hierarchical model broke down. You had to get enough economic value that people were really incentivized to game the system and so on and so forth. Before, you know, maybe Yahoo could have self disrupted, but before something like Google was there, Yahoo almost built out enough of the web economy to make something like Google necessary. You know. Anyway, so one thing I wanted to talk about, I'm going to go through various other areas but what is AI disrupted? What is AI going to disrupt? Right. So what is it already disrupted? So search, it's taken points off of Google Share, you know, like stack overflow, you know, their queries are down. Image search, because now image search is image generation, obviously video, obviously many different kinds of specialty apps will, you know, things that are, for example, like various sales tools that make templated emails and things like that, those all, you know, change. I, I, I'm not sure. Salesforce, I mean Salesforce is, you know, certainly they're using AI, but the entire Salesforce model, like spamming people with email, I'm not sure that's going to last in the age of AI because you can spam so many of them now.

Benedict Evans (47:11)

Right.

Balaji Srinivasan (47:13)

So, so those are some of the, you know, obviously robotics, obviously protein folding and whatnot. What is it going to disrupt that people haven't thought about yet? And I, I can, I can give some ideas as well.

Benedict Evans (47:23)

I mean one answer is we don't know. It's is like trying to say that, ask that question about the Internet in 1994.

Balaji Srinivasan (47:29)

Sure.

Benedict Evans (47:29)

And the joke is always that newspapers thought the Internet would be great because they'd save on printing. And they did at first, at first.

Balaji Srinivasan (47:34)

Probably was good for them.

Benedict Evans (47:35)

Yes, they did.

Balaji Srinivasan (47:36)

Yes.

Benedict Evans (47:36)

I did a slide in the last presentation I did, because it struck me that people would always say, well, you know, Uber didn't sell software to taxi companies and Airbnb didn't sell software to hotels. They redefined what those things were. So I went and did a chart of, well, what happened to taxis versus what happened to hotels. And actually, rather unsurprisingly, what happened. Taxi medallion is Uber demolishes taxes mostly. Airbnb is mostly additive to hotels.

Balaji Srinivasan (48:03)

Why is that? I think Airbnb is a different kind of experience in a hotel.

Benedict Evans (48:06)

It's not a substitutional experience in the same way.

Balaji Srinivasan (48:08)

It's complimentary.

Benedict Evans (48:09)

Half of business travel, half of hotels are business. There's another whole bunch of conferences. There's a bunch that's about like, I mean, just, you know, okay, so two examples like, my fiance works for a country, goes to fly to Milwaukee. She arrives in Town at 10 o' clock at night, she needs a gym. She's got a client meeting the next morning and then she's got her flying back to New York. She doesn't want to go and stay in some random stranger's hotel, which she's got no idea what it's going to be like. She wants, you know, a very specific brand promise from random strangers.

Balaji Srinivasan (48:38)

Airbnb. She wants to stay in a hotel.

Benedict Evans (48:39)

No, she will stay in a hotel. She's not going to stay in an Airbnb. The other side of this is, I think there's a more general point and same thing, I arrived in Singapore at 2 o' clock this morning. I'm not going to go and work out whether this Airbnb is any good. I can stay in a hotel, Sure. I think there's a more general point, which is that everything is probably disruptive to someone at some point in the value chain, but it kind of depends on the industry quite how much and in what sense. So, like, the iPhone demolished the existing cellular industry didn't really have any effect on telcos. Telcos kind of hoped that they were going to do all these services, but that was never going to happen. But telco mobile operators today are basically the same companies that they were 20 years ago with basically the same share, because their business was not in anything that the iPhone changed, except that they're providing massively more data than they were in the past. The business is basically owning sites and owning spectrum and connecting them up and selling that to consumers. The same thing with like online travel booking completely demolished the travel agent industry didn't really change the airline business. Airlines had to do a bunch of stuff around loyalty and pricing and maybe pricing became much more transparent and so on. But the end of the day their business is owning or leasing aeroplanes and there's a front end change, buying fuel and owning landing slots and maintaining the aircraft. And so now of course, the counterargument would be you could have looked at taxis and say, well, clearly that's not going to get changed by the Internet, except maybe you'll be able to book a taxi more efficiently until Uber comes along and changes it. But the point is you can't. There's this sort of very naive view that says, oh well, the software will just destroy everything thing. Right, Right. And it, the answer is. Well, it kind of depends.

Balaji Srinivasan (50:25)

It's paths. That's true.

Benedict Evans (50:26)

Yeah. And like one of the ways that I sort of think about this is that like the tech industry kind of comes and changes everything in industry and resets how it works and then leaves and goes off and works. And so you know the joke about how consultants are seagulls.

Balaji Srinivasan (50:40)

Yeah.

Benedict Evans (50:40)

They fly everywhere, make lots of noise and fly out.

Balaji Srinivasan (50:43)

Right.

Benedict Evans (50:43)

And so if you think about what happened to books or music, no one in the tech industry cares about music anymore. Right. Like recorded.

Balaji Srinivasan (50:50)

Yeah. Spotify does, but Spotify does. It's not the main event.

Benedict Evans (50:53)

Yeah. Recorded music is like $20 billion a year. It's like a rounding error in the scale of the tech industry. Streaming means it has no strategic leverage for Apple or Google.

Balaji Srinivasan (51:01)

Suno is interesting though. So the AI clearance.

Benedict Evans (51:05)

But for the last 20 years, 20 years ago, the Internet completely screwed the music industry and since then it left and doesn't care. Same thing in books. Like all the conversations around books right now, some of which are about Amazon, are book industry conversations. I think there's something similar happening now with video generation and Hollywood. Like everybody in Hollywood got over the panic and now everyone is sitting and looking at this and thinking, okay, well this saves a bunch of second unit stuff.

Balaji Srinivasan (51:30)

So one way of thinking about they're.

Benedict Evans (51:32)

All like, all the questions for what does this mean? Are questions for people in la.

Balaji Srinivasan (51:36)

So one way of thinking about it is conversation is proportional to derivative rather than absolute value. So let's say you have a, a sigmoid that's going like, like this and then it flattens out, right?

Benedict Evans (51:47)

Yeah.

Balaji Srinivasan (51:48)

So when it's like a nullity or ubiquity, you know, when it, when it, when it doesn't exist or when it's everywhere, when 0% or 100% it's just not notable. It's not worth talking about. Right.

Benedict Evans (52:00)

Yeah.

Balaji Srinivasan (52:00)

People use Uber or Dropbox a lot more today than when they were talking about Dropbox and Uber a lot. Right. So the conversation is maximum at the time of maximum growth and then it's just much less. Because now it's like not notable. It's just a feature of the environment. Right.

Benedict Evans (52:18)

So you can do Google N grams that show exactly this.

Balaji Srinivasan (52:22)

I think that'd be a great. That'd be a great craft.

Benedict Evans (52:24)

So you can do them for like steel or. And some of these.

Balaji Srinivasan (52:28)

Oh yeah, railroads. Yeah, steel.

Benedict Evans (52:31)

Because it starts in 1800. And of course some of them, you look at it, you go, oh, I'm actually seeing a chart of World War II do. In some ways where you see steel suddenly does that, or shipping suddenly does that.

Balaji Srinivasan (52:42)

And that's not obvious. Right. Because conversations or like attention is focused on change rather than absolute value.

Benedict Evans (52:48)

Well, I always used to do a. I always used to be fascinated by elevators. I get these kind of autism spectrum fascinations about things. And there's a chart I did of the number of people employed in the US as elevator attendants, which is a perfect bell curve.

Balaji Srinivasan (53:03)

Oh, and shift. Yeah.

Benedict Evans (53:04)

It's all curves up and down. And this is because first half of the 20th century you didn't have any elevators. You deploy a lot of elevators.

Balaji Srinivasan (53:10)

Right.

Benedict Evans (53:10)

Second half of the 20th century, they become automatic. You have a button and you can go and find all this advertising.

Balaji Srinivasan (53:16)

Why were they at the beginning? Was it just like switch short operators?

Benedict Evans (53:18)

That was how it worked.

Balaji Srinivasan (53:19)

Was it technical enough?

Benedict Evans (53:20)

There was no. But there was the. Well, if you think about what it actually takes to have an automatic elevator system in a building, you've got to have all the dispatching.

Balaji Srinivasan (53:29)

Uh huh.

Benedict Evans (53:30)

Oh, you've got to have the dispatching and the queuing. I see there's an interim stage where you have an elevator attendant who would just stand in the elevator and you would say, I want Buffalo 5, please. And they'd press the button for 5.

Balaji Srinivasan (53:41)

Right.

Benedict Evans (53:41)

But if you get in, you know. And originally elevator.

Balaji Srinivasan (53:44)

Oh yeah. What was it originally? Before the buttons?

Benedict Evans (53:45)

There is a lever that's an accelerator and a brake.

Balaji Srinivasan (53:48)

Oh, so it was like a car.

Benedict Evans (53:50)

Almost exactly. It's a streetcar. Elevators.

Balaji Srinivasan (53:52)

It was a vertical streetcar. I didn't know that. That's interesting.

Benedict Evans (53:55)

So there's a fantastic book I have called Cultural History of Elevators, which is all about how weird this was.

Balaji Srinivasan (54:01)

So it was a vertical train.

Benedict Evans (54:02)

Yes, it's a vertical train.

Balaji Srinivasan (54:04)

Wow. And that's how People thought about it.

Benedict Evans (54:07)

Yeah. And so an elevator train, you can kill people. And there's this wonderful story I tell everybody, which is that you press the buzzer to summon the elevator, but it's literally you're just ringing a bell and a light goes on in the elevator car. And. And there's this story from the War Department.

Balaji Srinivasan (54:27)

It's like hailing a taxi.

Benedict Evans (54:28)

Yeah. There's a story from a War Department or ringing for a servant. There's a story from the war department in D.C. which is that you would buzz more based on how senior you were. So imagine you're like a lieutenant and you get into the elevator on the second floor and you want to go to the 10th floor, but on the way the buzz rang, it rings four times.

Balaji Srinivasan (54:46)

That's the general. And you have to go down to.

Benedict Evans (54:49)

Stop on the sixth floor and go down to the first floor, and then a major gets in. So now you're. So theoretically, this call of tenant could spend an entire day in the elevator going up and down.

Balaji Srinivasan (54:58)

So interesting.

Benedict Evans (54:59)

And we don't see any of this now, which is your point about conversation.

Balaji Srinivasan (55:02)

Yeah.

Benedict Evans (55:02)

You don't get into an elevator now and say, it's an. It's an electronic elevator.

Balaji Srinivasan (55:06)

Right.

Benedict Evans (55:06)

It's automatic. Right. It's just an elevator.

Balaji Srinivasan (55:09)

It simply said something like, there's a phrase which is, civilization advances as you can do more things without thinking about them. Like the quote, just work.

Benedict Evans (55:18)

Right. The classic one is light people always.

Balaji Srinivasan (55:21)

Yeah, electricity.

Benedict Evans (55:22)

Light gets cheap.

Balaji Srinivasan (55:23)

Yes, that's right. I think, you know, the age of Internet now, sometimes what happens is these things get really ubiquitous and they're out of the conversation. And then there's this. Now that you can treat them as like at 100% adoption, then the new thing arises. For example, all of the craziness of the last 10 years is in part a function of the fact that. That social media got such ubiquity in the early 2000 and tens, such that it was no longer the novelty was, oh, I'm on social media, I'm using it. How do I use this Twitter app or whatever? Everybody knows what Twitter is. Everybody knows how to use it. They know what a like is. Whatever, whatever. And then you start getting.

Benedict Evans (56:05)

Then you get the second order effects.

Balaji Srinivasan (56:06)

The second order effects. That's right. So it's almost like. It's like installing a device driver and then you can install the next one and the next one. But it's like the device driver is the percentage of the population that's adopted something. And once it gets to 100% or 90% something, then you can like, I'll give you an example. Like during the pandemic, there's just the assumption that everybody had a mobile phone. Right. And they could QR code scan this and the other. In Asia, that was a really big thing. Right. That's how you'd show your health.

Benedict Evans (56:34)

That made QR codes work in, in the west as well.

Balaji Srinivasan (56:36)

Yeah, that's right. But basically, obviously, 10 years ago, you know, 10 years beforehand, they wouldn't be able to do that. They would have to have some other paper system or something like that. In 2010, you couldn't assume everybody on the planet had a smartphone. It was, it was getting big, but it wasn't yet there. It's certainly 15 years ago.

Benedict Evans (56:50)

Yeah.

Balaji Srinivasan (56:50)

Nobody would have it. Right. So that was something where the ubiquity of something, maybe sometimes the next step is, comes from that ubiquity or, or ubiquity of two or three things at the same time.

Benedict Evans (57:05)

Yeah. I mean, you could think about TV and radio, all forms of mass media in the past. And you know, the growth of pop music requires recorded music and requires radio. And you know, the greater mass democracy kind of goes hand in hand with literacy. Newspapers. Right. That you need newspapers before you can have. Other stuff has to happen. Right. For that to come. And then of course you have backlash sort of thing. There's something interesting in looking at stuff like the arts and crafts movement in the late 19th century, because here's a bunch of people who say we hate all this mass manufactured stuff that's handcrafted things. It's funny that that's not a statement that would make any sense in 1800.

Balaji Srinivasan (57:44)

Well, well, it's, it's funny because there's this, what you're talking about, like people were farmers that are artisans and are like, oh my God, this automation is disrupt. We hate it so much. We want to go back to the old ways. And, and now it's funny is those manufacturing jobs that all these workers were so mad about in the late 1800s and early 1900s, all the strikes, all communism and so on, those are now the things that are looked back on romantically by a lot of MAGA types where they're like, oh, that was such a great job. I wish I had that. I hate this information job kind of thing. I hate this, you know, these, these, these desk jobs and so on and so forth. So it's interesting because there's a romanticization sometimes of the past thing even as millions of people are Exiting that for the next thing. Now this is a little more complicated obviously by the fact that China has a lot of those quote manufacturing jobs, but yet a lot of them are being automated in China as well with the robots. So it's funny, the thing that people were so mad about that they were getting seemingly pushed into which was manufacturing, out of farming, into manufacturing are things that at least some fraction of this generation wants to go back to or they think they have. Do you know, I think that's interesting.

Benedict Evans (58:50)

Some of those things. I mean the, the Luddites are those one of these sort of misunderstood movements because a lot of what the Luddites are about is self employed, high status artisans losing that status and being pushed into low status commodity jobs.

Balaji Srinivasan (59:07)

So this is going to be the big thing with I think have you seen the elephant graph? So the elephant graph and some people dispute the graph, but I think it's probably gesturing at something. That's right. It shows percentiles or deciles of the world in terms of income and it shows over the last 20 something years, I think from 91 to 2008 or something like that where the growth went like whose incomes rose and basically most of the world. So the lower 10% in Africa didn't gain that much but like maybe from the 10 to 20th percent through the 70 to 80% had huge growth. Then it drops off in the 8th 90% to almost zero and then picks up again at the very top. Right. What that means is the like global, you know, elite in every country did great and so did China, India, Vietnam, Eastern Europe. All these countries are no longer socialist, communist, etc. Right. But the western middle class didn't. And that is a big part of I think the societal instability now when we're looking at it is, you know, in America they have, you know, obviously red versus blue. But one way of thinking about it is starting in, you know, certainly in 2008 there's a ramp where China flips US manufacturing. And so China puts all this pressure on red America and that leads to Trump and trade war. And you've seen that, that graph of print media disruption. Disruption, right, that's the Internet suddenly rising after 2008 to flip blue America and it takes all the ad revenue away and it's all not shattering. It's also Craigslist, it's classified ads, a bunch of other things. So the Internet disrupts blue America and that leads to wokeness in the 2010s, I think and also Tech Lash, right, which is the anti tech movement. So we look at it as red and blue, but there's also China and the Internet over here where the Internet is disrupting blue and China's disrupting red. So the thing I think that's coming next is, is AI disrupts blue America and robots disrupt red America. And so Chinese robots and Internet AI and so that artisan movement kind of thing is going to accelerate where people are going to be mad about that happening. I think on balance there's going to be a lot more productivity in the rest of the world. But it's possible, for example, that a job that's at let's say 200k or something like that in the US and there's somebody in, in India or Mongolia or Vietnam or something who's at $2,000 a year that, that equilibrates at like 20k, right. For like somebody supervising medical results or something like that. Right. Where the licensure is no longer as important. The Western licensure, the Western state doesn't, can't really protect it as much because it's all on the Internet. And that's a boon for everybody who's a customer of that. Like healthcare costs go down around the world, you've got a great doctor on tap at any time. Most people benefit from it. But those people who lost, you know, relative status, relative money and that get super angry. And I think the burning of the waymos and like the extreme anti AI sentiment that I see among some people is, is kind of a precursor to that. Let me know your thoughts.

Benedict Evans (62:14)

So I think this is a general observation that like, like when Europeans live in Europe, probably the, something similar in Asia, when Europeans live in Europe we all feel different. So like Germans are very different to Italians and different to British people, different French people and so on. And when Europeans live in America they all feel European and America is buried is in a different place to the aggregate of Europe. And the US has its own sort of political culture and political questions. Ah yes, that are different to the questions in France or Germany or Britain. I do think some of what's happened, and I don't wouldn't call myself political analyst, but I think some of what's happened is that certainly in the US to some extent the UK there were coalitions, particularly there were, on the progressive side or the left side, there was a coalition of, of urban upper middle class, highly educated people with a certain set of social attitudes. And working collar, blue collar, working class, blue collar people has been totally busted in a different part of the country. Right. Often with rather different Social and political attitudes.

Balaji Srinivasan (63:32)

Yes.

Benedict Evans (63:33)

And the same thing I think in the US and the Republican Party on the right there was a coalition of.

Balaji Srinivasan (63:43)

Sort of Wall Street Journal reading capitalists.

Benedict Evans (63:45)

Yeah. Like Mitt Romney and military guys that, that has split apart completely and all of those, you know, center right, economically conservative, socially liberal people who are Republicans kind of don't have a political party anymore. Right. And equally people who were sort of.

Balaji Srinivasan (64:08)

All slightly Bloomberg central.

Benedict Evans (64:10)

Bloomberg centralists. Centralists kind of don't have a political party anymore. And there's a lot of those coalitions have kind of broken apart. Now what you have in a bunch of European countries is partly because of proportional representation is it's viable to have half a dozen different parties. And the US and the UK because of the first past the post system, you don't have multiple. It's never been viable to have five different political parties at different points in the spectrum. In the same way the US has, the UK has got this kind of weird hangover center Liberal Party which is. No one's ever been quite clear what it was for sort of in the middle quote unquote called Liberal Party. There's an interesting sort of sideline there, which is the Liberal party in the UK in the 19th century was one of the two parties of government and it was socially liberal and economically conservative. But in the 19th century, what we now call economically conservative in the 19th century meant pro free trade and against regulation.

Balaji Srinivasan (65:03)

Right.

Benedict Evans (65:03)

Whereas now economically conservative is the other way round.

Balaji Srinivasan (65:06)

Yes.

Benedict Evans (65:07)

So all of those labels kind of shift and move and change and mean different things over time.

Balaji Srinivasan (65:11)

It's interesting mag, I would say mag is arguably against, certainly against free trade, but they're also against regulation. So it's like half right but, but it's finish what you're saying. I agree with you. Of course the labels do change.

Benedict Evans (65:23)

Yeah, the labels change. The coalitions broke apart. Break apart. I think there's always this tension in looking at progressive ideas and saying because if you look at the last hundred years, the social progress progressive ideas have always one like nobody today says like being gay should be illegal. Like so, you know, a little bit like what we were saying about AI a while ago today you, you, you could deterministically say that what is woke today in 30 years time will be what every far right conservative agrees with.

Balaji Srinivasan (66:00)

Like yeah, people have said that kind.

Benedict Evans (66:02)

Of thing theoretically in 50 years, you know, maybe, maybe not. But there's also. You also have these kind of overreaches around this.

Balaji Srinivasan (66:07)

Right.

Benedict Evans (66:07)

It does strike me that one of the differences between the US and UK politics is that what happened in the last, in my lifetime is that the right, for want of a better term, won the economic argument that state ownership and government control of the economy is bad.

Balaji Srinivasan (66:23)

Right.

Benedict Evans (66:25)

And the left won the social arguments that like gay marriage is okay, well it's, and, and so on. And in what happened in the UK was that the right embraced that and the Conservative Party is the party that brought in gay marriage in the UK Whereas in the left in the US it's kind of the other way around. The Republicans kind of.

Balaji Srinivasan (66:46)

Tony Blair sure brought in kind of.

Benedict Evans (66:48)

Yeah. And he brought in high level economics. Whereas what happened in the US is that the Republican party in the US never kind of accepted that it had lost the social arguments.

Balaji Srinivasan (66:58)

Well, it's interesting, I think from 1950, like the moment of 1950, you do have something where because communism fell basically because Nazism is defeated, the world moved socially to the left. And then when communist, as communism was defeated, it moved economically to the right. And so thus, for example, like the immigrant billionaire or gay billionaire is like in a sense can be white wing. Well, they're far, well they're far to the left of 1950 socially and they're far to the right in an economic right, in a sense of 1950 economically. Because 1950, yes, the Soviet Union had 100% taxes because it's communism, but the US had 90% marginal tax rates and you really couldn't get rich mid century in the US you could be a corporation man, you could work for NASA or gm, General Motors, General Mills, General Electric, but you were sort of funneled channeled into like these gigantic things. You had more freedom in the US than other places, but you're still very stealthified. It was too capital intensive to be an entrepreneur and so on. And then gradually with, you know, I think the transition was the mirror moment where that's begun a decentralization arc and history is running in reverse since that moment. But, and so I think a lot of things are happening this century that are like a reversal of things in the past. You know, I think it would be.

Benedict Evans (68:13)

Interesting, and I have no opinion about this at all, but it would be interesting to ask what is behind the growth in billionaires? Oh, is this an unlocking of a new kind? Is this a wave of company creation?

Balaji Srinivasan (68:26)

So I, I, I have, you see what I mean? Yeah, I do have thesis on this.

Benedict Evans (68:29)

Which is to your point is why are there new billionaires? Is that because there were a bunch of new companies and there are first generation Owners and where did those come from? And certainly some of them came from Google and you know, winner, global, winner takes all effects. And some of them didn't, I don't know, I mean I, I, I, I, I'm, I'm not sure how much value I can, I can kind of add to that conversation. There's a bunch of people of economic statistical questions where does not the time.

Balaji Srinivasan (68:55)

I can give some thoughts on that which is that has a U curve, right? Where for example, like who is the richest guy in the Soviet Union? Like didn't exist Communism, you know, basically Stalin, you know, didn't need money because you could just requisition anything.

Benedict Evans (69:11)

Well the Soviet Union is kind of a bad example of creating billionaires cut the country up and gave it to 20 people.

Balaji Srinivasan (69:16)

Well, no, well it's Right, but that's starting in the 90s, right. Then it wasn't, that was Russia then, then. Right, but basically the number of like independently wealthy men who could do things in the US for example, a lot of the, the, the great fortunes, the robber barons and, and captains of industry were forced into foundations. That's why you have the Ford Foundation, Carnegie Foundation, Mellon Foundation, Rockville foundation. Because in the 1930s Roosevelt didn't want any other powers besides him. So he, you know, went after Andrew Mellon, all these people, Ida Tarbell went after Rockefeller and, and those fortunes were corralled and basically controlled by the state in these foundations. In the Soviet Union, in communist China, they were just seized. Right? So basically, let me give the normal way of talking about this is inequality is rising and that's terrible. Right. Another way of thinking about it is what is the state, right? The state is in a sense it's like all the people who are its citizens and they kind of crowdfund the state. Right. And the question is, is are they, do they have a choice in doing that? Can they opt out of that? Like what set are they part of? You know, for example, if they're on the Franco German border, can they call themselves part of the German side or the French side? You know, how about the Polish, the Polish German border? But that kind of thing. And how much does the state take and how powerful is it? And mid century because of mass media and mass production, the states were more centralized they've ever been in history. I could show a bunch of graphs on that. That's not just, that's a quantitative of thing. So you had these giga states, you had fewer sovereign units on the planet than at any time before or since. Like only like 50 UN countries today, there's like 196. So things have decentralized since then. If you go backwards in time, you go to like Germany under Bismarck, you've got all these principalities. You go to France before the revolution, you have all these things. Italy before Garibaldi, you have all of these little city states and so on, right? So you go backwards in time and forwards in time. It's decentralized. And the same thing happens where you got lots of fortunes, you've got lots of, of, you know, individual potentates and what have you. Right. So in a sense, like the world is sort of returning to what it used to be, with a big exception being China. I think China is the, like the, the 20th century centralized state that will keep scaling into the, into this century. So anyway, the reason I just say that is I, I think there is something real going on, which is that the state is just take. Capturing less of the wealth of its individuals. People are sort of breaking away on the borders of it and then being able to do their own thing, one thing. And so it's like Elon, not NASA, it's like Travis, not, not, not taxi medallions and so on. And there's a good to that where there's a lot more room for individual initiative. But there's a bad to that as well, which is then people don't feel as bought in on the collective project and they're not like included in it. It's some guy's thing, it's not their thing. It's not like America lands on the moon or it's Elon, okay, fine, you know, and they don't feel as bought in. Right. So it's a complicated kind of thing. I think we're gonna have to renegotiate all that stuff in the future.

Benedict Evans (72:02)

I think there's a layer to this outside, again, outside sort of U.S. politics, which is that partly because the U.S. partly the nature of the U.S. economy, partly because the U.S. is a big domestic market, partly because the successful Internet companies are in the US and have global winner takes all effects. People outside the US for the first time think, well, we've got all these giant US companies that are running stuff in our country. And that was kind of true for General Motors or Coca Cola, but it's much more direct, but not really.

Balaji Srinivasan (72:36)

Yeah, yeah, right.

Benedict Evans (72:36)

You know, General Motors sold cars, but you had a lot of your own car companies as well. And IBM didn't decide how you built roads or anything. And there's certainly a sort of a, you know, you go to European events now and there's people saying, well, do we need our own Google? And at one level those are like dumb questions, but they're dumb questions about, like a real issue, which is you have this other layer of stuff that you're using, which didn't used to be globalized and used to be subject to local democratic control. And now, now. Well, it's not quite clear how that works.

Balaji Srinivasan (73:05)

Yeah, yeah. So actually it's very important. I mean, what you're hitting on there is, I think, one of the core questions, and I'll actually ask it in reverse, which is, are those American companies? Basically, is the Internet American right Now, on one level, you'd say that's a weird question.

Benedict Evans (73:20)

Of course, there's two parts to that, is, are they American? But also is they're not. They're not in our country. If you're Swedish or Italian, it's not a Swedish company.

Balaji Srinivasan (73:28)

That's right. That's right. So. So, like, you know, my view is the Internet is to America, but America was to Britain. It is like the version 3.0. And because the early Americans actually considered themselves as, you know, British. Right. All the folkways and stuff came from Britain.

Benedict Evans (73:47)

And the American War of Independence is essentially a civil war.

Balaji Srinivasan (73:50)

Yeah, exactly. That's right. So they had a people and they had a. They had a land, but they didn't have a government. Right. Because the government was in London. Right. And when they had all three, they became Americans. They had a sense of self. And I think with the Internet, we have actually a lot of tribes that actually have a people and a government, but not land. And the reason they have a government is they have a blockchain, they have a social network, they have with moderators or forums. And now increasingly they have an AI agent or a central oracle or something like that, where it almost takes a role of a God, which they all ask questions to. So you think of every large enough online community that has its own social network, whether it's a discord or a forum or something like that, its own cryptocurrency, which has its, you know, smart contracts and currency, and its own AI, which is sort of like its oracle or search of all the community's knowledge, right? And that's like a digital community that actually has a fair amount of strength. And then because you know where your communications happening, they're happening online, whereas your transactions are online, more and more of your wealth is stored online like crypto's at trillions of dollars. Now, it wasn't, wasn't that 15 years ago it was at zero, basically. And so the significance of these cloud communities I think is underappreciated and eventually they're going to be able to have enough money to crowdfund territory. And so the. Because the tension between your primary identity is online, your social network's online, your currency's online, your information is online and then not being grouped offline, that'll resolve in my view, in terms of the descent of the clouds that land.

Benedict Evans (75:21)

So it's interesting, I mean I probably take a sort of more prosaic view of this, but listening to you talk I am reminded of like distant memories of being at university and looking at social history and you know, there are a lot of social history is about the kind of. The joining into groups.

Balaji Srinivasan (75:38)

Yes.

Benedict Evans (75:38)

And so the joining. Why are you joining and how. What is the sort of form of.

Balaji Srinivasan (75:43)

Sort of self expression?

Benedict Evans (75:44)

Yes. Why do people want to fund monasteries? Why do people form lay brotherhoods around the church? Why do people like There's a whole 19th century British thing of like all sorts of social joining. Why do people want to join militias and you know, why do they want to form all these kind of former guild. Why do they want to form all of these kind of different social groups and social clubs and ways of getting together and what are they trying to achieve? And some of it is about, you know, self defense, you know, or putting, not in a kind of military sense, but about, you know, forming your group to protect your group's interests. Some of it is about a establishing status, some of it is about self expression and self actualization, you know, kind of classic maslow hierarchy stuff. But it's not new to have lots of communities. What is new is that they're not necessarily kind of physically like co located.

Balaji Srinivasan (76:38)

Yes.

Benedict Evans (76:38)

They're not necessarily centered around, I mean things like women's suffrage, you know, they're not necessarily centered around a movement or some specific objective.

Balaji Srinivasan (76:46)

I think they will be. I think they will be.

Benedict Evans (76:48)

Well, there may be, but we've had those in the past, you know, the Cornwall League or women's suffrage. Suffrage, all of those things.

Balaji Srinivasan (76:52)

Yes.

Benedict Evans (76:52)

Veganism, slavery, anti slavery movements and so on. So those senses of, you know, social organization and joining and grouping and clubs in different forms in different aspects of society for different reasons is kind of a recurrent pattern of human society.

Balaji Srinivasan (77:09)

Yes.

Benedict Evans (77:10)

And now it gets expressed, which is the sort of thing we always talk about is, you know, the Internet is human behavior and it expresses and channels it in new ways. And that's everything from you know, people being horrible on Twitter or doing terrible things on the Internet, rude to people, forming groups, clubs and societies on discord or Reddit or whatever it is.

Balaji Srinivasan (77:28)

That's right. You know, by the way, I have, I have an explanation which you might find funny as to. I used to wonder, why are people so crazy on Twitter? Why are they so crazy on social media? Because, you know, like starting fights and stuff. Just as a sidebar, I was able to explain it in the following way. You know. You know the Unabomber in the early 90s? Yeah. So he blew up all these people.

Benedict Evans (77:48)

Yep.

Balaji Srinivasan (77:49)

But you know, the reason he did that was to get an op ed in the Washington Post. Right. So he killed all those people for the distribution. He killed all those people just to get his message out there. So you realize there's people like that, then it actually makes it more understandable how many crazy people there are on social media. If, if someone is willing to kill all these people to get, you know, just his message out there, a lot of other people would be willing to be very nasty on social media to get their message out.

Benedict Evans (78:10)

Yeah. I always thought a lot of it was about context collapse, which is sort of actually.

Balaji Srinivasan (78:14)

Yeah.

Benedict Evans (78:14)

Buggy doesn't mean anything.

Balaji Srinivasan (78:15)

Yes.

Benedict Evans (78:16)

I felt like some of it was, you don't know who that person is and you haven't understood what they've seen, said and what else they think. And you presume they think X. It's like, it's lossy compression. You kind of compress three paragraphs. There's no subclause. There's no nuance. You can't say, of course I'm not a Nazi. And.

Balaji Srinivasan (78:32)

Right.

Benedict Evans (78:33)

Some of it is also which.

Balaji Srinivasan (78:34)

And in fact, they can't take that for granted because you're not.

Benedict Evans (78:37)

Because maybe you all.

Balaji Srinivasan (78:38)

Yeah, well, yeah. Or, or, or basically they're like, you know, they have no context on you. They can't read 5,000 posts. They don't know where to trust you. And so, and so some of it is also.

Benedict Evans (78:46)

So it just says Morgan Housel, I think.

Balaji Srinivasan (78:49)

Yeah. Morgan Housel. Yeah.

Benedict Evans (78:50)

He wrote a book, quoted me. And that gets endlessly re quoted where I said something like, like the Internet means that basically you're confronted with people who disagree with you.

Balaji Srinivasan (78:58)

Yes.

Benedict Evans (78:59)

And you're.

Balaji Srinivasan (78:59)

All the time.

Benedict Evans (79:00)

And you didn't realize there were all these people who, like, the particular thing I always found was weird was there were people who were like very, very far left. There are people who are communists and they're like, you'll say Something that isn't communist. And they'll be, like, amazed. They were like, oh. The thing was always. I always thought was weird is like, I can. I think it's weird that you're a communist because at this stage, you have to be an idiot to be a congress.

Balaji Srinivasan (79:21)

Right.

Benedict Evans (79:21)

But it's even more weird that you don't know that most people aren't.

Balaji Srinivasan (79:25)

Yeah, yeah, yeah, yeah. They're, like, shocked by it.

Benedict Evans (79:27)

They're, like, amazed that anyone doesn't agree with their tiny minority opinion.

Balaji Srinivasan (79:31)

Yes, that's right.

Benedict Evans (79:31)

And a lot of Twitter was that. It was like, you're amazed that I don't think everybody should own a car. You're amazed that I don't agree with. I'm like, I don't necessarily share your opinion on every possible matter.

Balaji Srinivasan (79:43)

That's right. And I think the way that's going to reconcile is you're going to get a lot more, I think, smaller. I mean, in a sense, Twitter doesn't exist anymore. Right. X.

Benedict Evans (79:53)

Well, Twitter fragmented.

Balaji Srinivasan (79:55)

Exactly. It's a tower Babel moment. Right. So Twitter no longer exists. There's X and there's truth and there's Gab and Blue sky on the left and Mastodon and Threads and then the crypto ones like Farcaster, Lens, Noster.

Benedict Evans (80:08)

A lot of stuff went to things that didn't look like that. So stuff went. Went to LinkedIn, TikTok, or it went to TikTok, or it went to Instagram. And people make fun of LinkedIn, like, there isn't a bunch of bullshit on. On Twitter. But, you know, there's. I realized that an awful lot of corporate people were sitting quietly using LinkedIn when it didn't feel that they could use Twitter.

Balaji Srinivasan (80:27)

Yeah. Because basically, the funny thing is, it's interesting, something about LinkedIn means people are artificially polite. And something about X or Twitter, especially Twitter, I think. I think even more than X in some ways meant that they were artificially negative. Hostile. Right. And the funny thing about it is artificially hostile reads to people as more sincere. Like, that's to say of the two, there's something about the artificially polite. Like. Like, for example, a good review is not a rave review. A good review is, I love, you know, Ben's book. It was great. But he could improve X, Y and Y and Z. That's like the best review you'll get.

Benedict Evans (81:06)

Yeah.

Balaji Srinivasan (81:07)

You know, I mean, usually. Whereas a hater will be like, just. Just complete crap on you. Right. So the. The negative is generally much more negative than the positive is positive. And so when you see a LinkedIn style post, it's often like super positive and it feels fake immediately. But people don't apply the same filter. They think negative is real, but they don't think negative could also be fake. I always like a mental.

Benedict Evans (81:28)

You know, there was a thing that went viral a while ago of some surgeon who got a, they got a review and it was like, he saved my life. He's the most wonderful surgeon in history. It's amazing. It's wonderful. Four out of five stars.

Balaji Srinivasan (81:40)

Yeah, yeah, yeah, yeah, yeah, yeah, exactly.

Benedict Evans (81:42)

Okay.

Balaji Srinivasan (81:42)

Exactly. Wow.

Benedict Evans (81:43)

What did I have to do to get five stars?

Balaji Srinivasan (81:44)

Yeah, exactly. That's right. Like, you know, I forgot to give the mint chocolate under the pillow or something. Yeah, yeah. Okay. So like, you know, let's do, let's change gears. Let's talk about just survey of tech. Just things, you know, you can tell me you've been thinking about this. You have anything about this?

Benedict Evans (81:58)

This.

Balaji Srinivasan (81:58)

So we talked about like gadgets. So we talked about, you know, the glasses. We talked about.

Benedict Evans (82:06)

We haven't actually. Did we talk about glasses on the podcast or on the car? In the car.

Balaji Srinivasan (82:09)

We talked about glasses a little bit on the pod, but basically. Well, tell me, tell me your thoughts on glasses. Oh, so AR VR glasses.

Benedict Evans (82:15)

Yeah.

Balaji Srinivasan (82:16)

XR glasses. Yeah.

Benedict Evans (82:17)

So I've made this point a while a bunch on online as far as I can see. Like you have the VR experience, you think it's amazing. Amazing. It's not clear to me that this, my base case of VR is that it may end up like games consoles. In that you see a games console, it's amazing. Most people don't buy it. There's a portion of people that don't understand that games is actually quite a small industry in terms of number of people. It's a lot of money, but a lot. There's like 2 or 300 million people play games, console games. And so it may be that VR, you have the experience, it's amazing. You put it down, you walk away. Most people don't buy it, no matter how good the hardware. I think it's much easier to see something like what I'm wearing now being a universal device at the level of a smartphone. Clearly we don't have the optics for that yet. We may have.

Balaji Srinivasan (83:03)

It's improving every year though. It is, yeah, it is.

Benedict Evans (83:06)

The question is, is that next five years time, is that two years time? Is that 10 years? It's not clear yet.

Balaji Srinivasan (83:15)

Yeah. There's a few people I know who just like they almost subscribe to this space in the sense of they're constantly just getting the latest glasses, usually out of China, and they're just trying them out. Right. Or getting prototypes. There's various prototypes people are making and this is something that I feel there's some value in tracking because it's almost being ignored by the world right now.

Benedict Evans (83:38)

It is. Because it's like, it's, it's, it's one.

Balaji Srinivasan (83:40)

Of the hit that Gartner hype cycle.

Benedict Evans (83:41)

Thing that's bumping along the bottom and hasn't quite happened.

Balaji Srinivasan (83:44)

Yeah. Or it's a trough after the hype of Metaverse.

Benedict Evans (83:46)

There's a subset of that which is. Okay, clearly you want a wide field of view. Do you need to have something that looks like it's 3D, like it's really there? So do I need to have glasses that could put something on the table in front of us that looked like it was there? And that's radically harder than having a really good heads up display that could put a popular. That could put an eye. That could put like an iPad display hovering in front of me.

Balaji Srinivasan (84:09)

I think it helps a lot with things like repair hair. Like, you know, for example, you open the hood of a car and, well, that.

Benedict Evans (84:16)

But that's still a hud. That's still like a hovering label over the thing. Versus, does it need to work in broad daylight? Does it need to have black? Does it need to be able to occlude a bright white table like this?

Balaji Srinivasan (84:27)

Right.

Benedict Evans (84:27)

Maybe, maybe not. I think there's a range of outcomes there where you. Maybe it ends up like a watch that it's a. To be. Clearly, to begin with, it'll be a smartphone accessory just to have the computer. But does it end up like a watch where there's hundreds of millions of people who have it, but the smartphone is the main device? Right. Or does end up. No, actually a couple of billion people are wearing this.

Balaji Srinivasan (84:46)

Let me ask you another question. Does. Does a watch top out? And because the thing is, wearables are another thing that has huge traction and it's kind of like there's a lot, a lot, a lot. We could fill this table, this whole room now with Iot health stuff, right? Because there's watches, there's rings like the OURA rings, there's, you know, wristbands. I suppose it depends.

Benedict Evans (85:14)

It depends on the question. Is a nickel white like Mark Zuckerberg bought Oculus, Sorry, Mark Zuckerberg bought Oculus because he thinks this is the next smartphone. He didn't buy it to be a games device or to have 100 million people using it. He bought it because he thinks this is an Xbox smartphone.

Balaji Srinivasan (85:28)

Yeah. Because also he had been hit by the platform so hard. Yeah. He wants to own the platform for sure. It makes sense.

Benedict Evans (85:33)

So there's. My base case is that VR might be. Might crap out at 50 or 100 million people and I struggle to see it being 5 billion. I can see glasses being a couple of hundred quite easily. Once it worked, the optics are there. I can imagine it being 5 billion. I think that's harder.

Balaji Srinivasan (85:52)

But AR. Yeah, AR. AR slashR is probably bigger than VR. But as we were talking about, VR is very, you know, the thing new thing they're doing with for controlling military drones.

Benedict Evans (86:02)

Like there's loads of vertical stuff where. Absolutely that's going to nail it. Definitely. No question.

Balaji Srinivasan (86:07)

That's right. So all the telepresence.

Benedict Evans (86:08)

Yeah. And you know the guy up the telephone pole, the guy in the oil wearing Glo says yes, absolutely, that will be a. That is a thing already.

Balaji Srinivasan (86:15)

That's right. And I think, have you seen this movie, it's called Surrogates. It's actually, you know, pretty good sci fi movie from like almost 10, 15 years ago. And essentially like people are like, they stay at home and they pilot a good looking version of themselves as a surrogate walking around outside. So you can take more risks and so on because if that thing gets in a car crash or whatever, nobody cares. And then they could just do another surrogate and run around set. Right. So I do think what are the use cases for like a proper, the VR. The VR control of a remote thing. So it starts with I think drones. And have you ever done a VR headset with a drone? It's an experience. You should definitely try it. It's a wow moment because it really does feel like you're flying. Right. Which is very cool and an interesting experience. So I think it starts with drones but I think it eventually gets to something where you got gloves and maybe an omnidirectional treadmill or something like that. There's various kinds of things like that. And you are able to control a humanoid anywhere. Right. So you control a humanoid and you can, I don't know, clamber up a telephone pole and fix something. You can. And you're training the AI as you're doing this. Right. You could have a maintenance worker with skill in the art and we're not there yet. It'll be years before we're there. But eventually you have all these humanoids around where you can just go into this, like, animate the suit and start doing things, you know, so that's a pretty important use case for VR, like physical telepresence. You have to nail a bunch of technologies for that. But I could go through the gloves, I could go through the haptics. A lot of those things are moving forward.

Benedict Evans (87:48)

Right.

Balaji Srinivasan (87:50)

And, you know, a lot of people are pouring money into this. That's something I give a lot of credit to Zuck for. He's just, you know, he's just continuing this, you know, like, I don't know how many tens of billions of dollars have been put.

Benedict Evans (88:00)

He's probably. Probably put the thick end of a hundred billion into that.

Balaji Srinivasan (88:04)

Something along those lines.

Benedict Evans (88:05)

Like 75 to 100.

Balaji Srinivasan (88:06)

Yeah. I mean, they are actually selling a fair number of units now. It just hasn't come close to keeping up with the spend.

Benedict Evans (88:12)

Yeah, the sales are just bouncing along. It's like it's not good enough to break out of VR enthusiasts. And it's funny, you go back to what you said about Twitter, there's almost like a test, which is if you say that something probably isn't working yet, and you get a bunch of people shouting at you on social media, then that proves your right. Because if it was working, they wouldn't care.

Balaji Srinivasan (88:34)

Yeah, yeah, that's right. Well, so if you went on social.

Benedict Evans (88:37)

Media and said, Nobody uses TikTok, then people would just say, this guy's an idiot. Go on social media and say, and there aren't actually any consumer use cases for drones. You'll get like, the 10 people who love their drones.

Balaji Srinivasan (88:47)

Okay. There's the one exception, which I will argue with you on, which is crypto.

Podcast Host (88:51)

Yes.

Balaji Srinivasan (88:51)

Right. So that is something where people will say, there's no use for crypto. You will say, there's no. But there.

Benedict Evans (88:55)

There's just a huge number of idiots. It's on every side.

Balaji Srinivasan (88:57)

That's also true. That's right. That's right. Okay, so we did.

Benedict Evans (89:00)

So I don't say there's no use case for crypto. I have the most unpopular position possible, which, as I say, it's kind of useful, but not completely useful, which means I get both sides screaming at me.

Balaji Srinivasan (89:08)

Yeah, that's fine. That's perfect position. So actually, what is Ben Evans on crypto, then? I'll tell you biology on crypto.

Benedict Evans (89:16)

There's several answers to that question. One of them is, and this is sort of more an observation, which I hope you won't tell me I'm wrong, is like like there's a bunch of clever people working away building like all the tourists left, like the whole NFT thing was all nonsense and that all there, all the tourists left. The tourists and the grifters basically all moved on to AI.

Balaji Srinivasan (89:35)

Yeah, a lot of them, yes.

Benedict Evans (89:36)

And all the kind of people trying to build content brands saying this is all wonderful or this is all bullshit. They all moved off to AI and there's a bunch of people sitting and doing like abstruse. Very clever, very technical stuff. There's a bunch of stuff working or being built that may work around a finance. The finance industry around finance, rails around stablecoins, various kinds of financial instruments, most of which is storing money or speculating in money or moving money around.

Balaji Srinivasan (90:04)

Yeah.

Benedict Evans (90:05)

There is a thesis that you could build Instagram on this, that this is sort of an open source computer in which you could write software that consumers would use. And I have a bunch of questions about how that would work, whether that would work, whether you would need to abstract the open. Sort the crypto stuff away so that the consumers didn't see it. And if you did that, then why would they care?

Balaji Srinivasan (90:32)

Totally.

Benedict Evans (90:34)

But none of that's kind of there yet. Like there aren't billion scale consumer apps built on blockchain yet. So there's a sort of watch this space around that. And then there's the finance side, which I think is sort of theoretically very interesting. But I struggle to get very interested in it, just personally not what I'm interested in. And I struggle to see ways that I could add value in talking about it. So I kind of pay attention to it. And every now and then I point out, like my newsletter on Sunday I pointed to the Shopify and stripe announcement, the stablecoins.

Balaji Srinivasan (91:12)

Yeah.

Benedict Evans (91:12)

And said like there's stuff happening here. Yeah. And you should pay attention to this. And there's people still interested in trying to build things. So if you've just written this off as all, you're kind of wrong.

Balaji Srinivasan (91:23)

Right.

Benedict Evans (91:24)

But as a writer and an analyst, I haven't moved it on to something that I feel I should write about.

Balaji Srinivasan (91:30)

Totally. So, okay, so that's very helpful. It's always helpful for me to kind of triangulate on an area, you know, the. So here is my, my basic view. You may have heard me say this 12 years ago. I think this is still true. Crypto is good for transactions that are very large, very small, very fast, very international, very automated, very complex, or that need to be very transparent. And the reason for that is like for example, a Starbucks Swipe like of a credit card is none of those things. It's not very large or very small. It's like a mezzanine transaction. It doesn't need to be very automated because you can just talk to the, you know, a cashier and see your receipt. It's not international. Both you and them are in the same room at the same time. It doesn't need to be transparent. You don't need a receipt on the blockchain for everybody to see and so on and so forth, right? So the reason people think about the coffee transaction when they think about crypto is it's one of the most common transactions people do. They pay for their coffee every day, right? So it's like, I don't know, 10% of your transactions, 20% are maybe coffee, because there's very few things you buy every day. Coffee is one of those things people, people buy every day. So where crypto really shines is the, the alternative forms of traffic. Actually, let me take your mobile example, right? The Internet can do telephony, but that was actually the thing that was best served by the existing system, right? We still have like local telephone calls, right? You can still use a telephone network to place telephone calls. Where the Internet shine was. And telephone calls were sort of of like mezzanine amounts of information, right? Especially local was like between people in the same country. Wasn't very international. Where the Internet shine was for example, moving really large files like Dropbox or very small files like tweets, right? Being very international, like across borders, being very automated. So it wasn't a human on both sides of the call, Right. It's shown for being very transparent. You're broadcasting the webpage, everybody. It's not a phone call, just between two people and so on and so forth, forth, right? So that I think is a good analogy where like yes, now, today, eventually the Internet took over long distance telephony because that was Skype and then WhatsApp and what have you. But even still today, telephony is well captured by the current system, right? And like the existing phone line still exists. That I think is a useful analogy for crypto, where crypto, for example, if you have, if you're a power user of money, right? If I want to receive or send a wire to a startup in Japan, usdc, I can do that in seconds and then I can refresh the page. They can refresh the page and they can see it's cleared, right? Yeah, that is a real use case. That's international wire transfers from anybody to anybody. With. And by the way, the bank account setup also is instant. Right. So think about what we've done. We've taken it from days to get a US and Japanese bank account set up to seconds. We've taken it from paying money to do that for the transfer itself to free. Free. We've taken it from taking multiple days for a wire transfer to clear to seconds. And we also, by the way, the uptime, it's not nine to five banking hours. You can do it, you know, 24, seven and you can do it on any device. Right. That's a lot of improvements. Just for the, the important use case of international wire transfers. Right. Then you also have the digital gold use case.

Benedict Evans (94:51)

Yeah, it's also, I mean digital gold I think is also something that there's a kind of country mapping here because some of what you're talking about is a much bigger problem in say in the US.

Balaji Srinivasan (95:01)

Yes.

Benedict Evans (95:01)

Than it is in countries with less. With different banking systems.

Balaji Srinivasan (95:04)

Yes. Some of it is also sepa. You guys have SEPA in Europe and it's not terrible.

Benedict Evans (95:07)

You send the money, it arrives for free. Also this is a point about PayPal.

Balaji Srinivasan (95:11)

But that SEPA works within Europe though SEPA would not work for wire transfer to Brazil, for example. So you saw the same issue.

Benedict Evans (95:17)

I think there's another point which is like I remember reading, reading about people in Argentina.

Balaji Srinivasan (95:24)

Yes.

Benedict Evans (95:24)

Literally keeping their money in bricks.

Balaji Srinivasan (95:27)

Exactly. That's right. So Argentina, Nigeria, Lebanon, where you, you.

Benedict Evans (95:31)

Actually can't trust your government.

Balaji Srinivasan (95:33)

Yes.

Benedict Evans (95:33)

And there are kind of places, there's.

Balaji Srinivasan (95:34)

A lot of places like that, unfortunately. Yes.

Benedict Evans (95:36)

There's also a bunch of places where nobody's worried about that they sent you for 100 years.

Balaji Srinivasan (95:41)

Exactly. That's right.

Benedict Evans (95:42)

So.

Balaji Srinivasan (95:42)

So the more middle class stable and so on, you are basically crypto is for the power user of money and the powerless. Right. The person who's reinventing what a bank account even is and the person who's just trying to hang on to a bank account. So it's like a U shaped coalition. Right. Similar to the people who actually benefited most from the global economy. Remember I said it was like the elephant graph. Right. You had the basically 10th to 80th percentile of the world who grew and you had the top 1% who grew and the, the western middle class didn't. Right. That coalition is actually also the crypto coalition. It's like the people who are just, you know, Internet as, as Tim Ferriss put it. James. God, what's, what's his name? Jason Bourns of The Internet. Right. Like just Internet hackers who are just trying to move money. Like for example. I'll give a concrete example. Brian Armstrong, you know my friend, CEO of Coinbase.

Benedict Evans (96:37)

Yeah.

Balaji Srinivasan (96:38)

One of the reasons he got into crypto, he had a few different life experiences that led him there. One was actually lived in Argentina for a while. So he saw like what a failed state would be like. And the second though was actually being an Airbnb, an engineer. See, the thing is, Airbnb, even still today has the problem of transactions that are very large, very international. Right. And, and also very one time low trust. Right. Because you've got like somebody from Denmark staying with someone from Japan and it's, it's a one time transaction of maybe on the order of a thousand dollars. There's actually a fair amount of money.

Benedict Evans (97:12)

Yep.

Balaji Srinivasan (97:12)

And like the wire system is simply not set up for that frequency of use between unrelated parties. And so there's a lot of friction on something like that. And to a surprising extent, Airbnb had a lot of forex risk, like, you know, because they had to hold currencies and all these different things. And the thing you thought was a solved problem, like just moving money from country to another, it's like, well, Airbnb has to do its accounting in USD, but it's got income in, you know, if they, if they're an American company and they've got somebody transferring money from Denmark to Japan, there, there's three currencies in that transaction just right there.

Benedict Evans (97:50)

Right.

Balaji Srinivasan (97:51)

So there's at least three currency pairs which fluctuate and you've got at least two or three banking systems and all the delays and fees you start to see, people are like, wow, this sucks so much. We need an Internet first banking system.

Benedict Evans (98:04)

Right.

Balaji Srinivasan (98:04)

We need something which is payments as packets.

Benedict Evans (98:06)

Yeah. Right.

Balaji Srinivasan (98:07)

So that was the second thing that motivated Brian to do it. Right. There's other things as well. Right. But, so where, where would I put crypto today? Right. I'd say there's at least three applications. There's more, but I'd say at least three that are at the trillion or multi hundred billion range. And those are a digital gold. Right. Just whether you believe in gold or not, like that's, that's there. People, people do believe. Even if you just consider an insurance.

Benedict Evans (98:31)

Policy, it's the thing that people are doing.

Balaji Srinivasan (98:32)

It's organic.

Benedict Evans (98:33)

That's right.

Balaji Srinivasan (98:35)

B is, it's like even if you didn't believe in luxury cars, that's a market. Right. So there's a market for It Right. Okay. B is international wire transfers. I think stablecoins are now there at this point there are now 1 2% is $250 billion trillions stable coins have passed Visa. They pass MasterCard.

Benedict Evans (98:52)

Right.

Balaji Srinivasan (98:53)

And then third is actually crowdfunding. Right. So if you look at the largest crowdfundings of all time, most of them are crypto. Crypto. And the reason is that capital formation online, like if you think about something like Kickstarter or what have you, it's actually more geographically limited and more limited by the credit card rails than you might think. For example, it's not that easy for somebody in Brazil and Japan and India to put 5,000 bucks into your Kickstarter.

Benedict Evans (99:18)

Right.

Balaji Srinivasan (99:19)

The credit card rails may not accept it might be fraud hit. Go ahead.

Benedict Evans (99:22)

Yeah, I was just going to say I wonder with some of the. There's a certain amount of swapping paper for paper in some of that.

Balaji Srinivasan (99:28)

Go ahead.

Benedict Evans (99:29)

Well, in the sense of here is a new crypto project.

Balaji Srinivasan (99:33)

Yes.

Benedict Evans (99:33)

A bunch of people who've speculated made a bunch of crypto money for sure put their paper games in bitcoin into this new crypto project.

Balaji Srinivasan (99:42)

Yes, that's right. That's right. But, but, but I'd say you're. You're right. A bunch of it is like that.

Benedict Evans (99:47)

Which is what a lot of NFTs was.

Balaji Srinivasan (99:48)

Yes, that's right. But, but even if we, even if you just totally write what was funded off just the mechanic of crowdfunding.

Benedict Evans (99:55)

Funding. Yeah.

Balaji Srinivasan (99:56)

Shows that that mechanic for capital formation, what they spent it on. I would agree with you. Many of those projects didn't go somewhere. Some of them went really far. Like Ethereum was a really. That paid for all the rest in a sense. If all the ones went to zero, that was so successful. But, but just the mechanic of capital formation where you have. So, so that gets me to number four. Right. If you look at now you may, you may start disbelieving. So at least those three markets, gold wire transfers, crowd funding, those are very large markets. Those are $100 billion trillion dollar markets. So then you go to like other cases. Now if I just look at trade volume right. Crypto today is actually the number four stock exchange in the world in terms of volume. Number one, nice. Number two, nasdaq. Number three, Shenzhen. Number four, crypto. And it's rising fast. The thing that has held it back for almost 15 years is the doing the obvious things was pathology meaning it like literally yesterday or like, like a day or two ago we finally fully legalized very clearly legalized. Putting a dollar on chain. Right now that we can put a dollar on chain very clearly. Such to the point that Amazon and Walmart are like, okay, congressional legislation is perfectly good. Let's go time. Right now we can finally put an equity on chain and we can put a fund interest on chain. We could put every paper kind of thing on, on chain. That is a very big deal. Right. That means that crowdfunding thing I talked about says that an Internet company can issue Internet equity and anybody in the world can be part of that cap table. Whether you choose to accept them or not is another thing. But the capital formation mechanism, it's now possible for somebody in Japan or Brazil or Mexico to invest in your company once you have Internet equities, Internet capital markets, that is now within sight. Now that we have the stablecoin thing, boom, done, done. There's nothing now. It's just a mechanical thing to get the legal system going to make the on chain equities work. And there's already work on that. So that is a big deal. Right. Because the US doesn't want to be the center of global financial empire anymore. Right. It's very conflicted about this, but with the tariffs and the trade war and tourist visas, work visas, student visa bans and so on, it is very conflict about whether it even wants foreign money coming, coming in to America. Right. And they've got remittances, taxes coming up, like one for 5%. So U.S. financial markets, I don't think are going to be there in the same way by 2035. I think Chinese markets are rising, Chinese stocks are rising. That's going to be one thing that's there. But I think the Internet capital markets will take over from American capital markets. And that's a very, very big application. Let me go through a few more. Is this interesting so far?

Benedict Evans (102:37)

It's interesting. I mean, I, I think about.

Balaji Srinivasan (102:40)

I mean, we've got numbers now. Yeah, go.

Benedict Evans (102:42)

There was a thing that I was. So I'm getting away from the microphone. As we were chatting about in the car this morning, I have a sort of a mental Venn diagram of like, stuff I feel I can add something to.

Balaji Srinivasan (102:54)

Sure.

Benedict Evans (102:55)

Stuff that I feel I understand and stuff where there's an audience.

Balaji Srinivasan (102:58)

Yes.

Benedict Evans (102:59)

And the challenge I always had in writing over at crypto is like a kind of a practical question as an analysis analyst is all AI or all kind of crypto questions. It felt like they were either very, very technical conversations about. It was kind of like writing about limits. And I always think that like crypto reminds me a lot of Open source.

Balaji Srinivasan (103:20)

Yes.

Benedict Evans (103:20)

And you are either.

Balaji Srinivasan (103:21)

It is open source.

Benedict Evans (103:22)

I know, but just in the sense of the general movement around.

Balaji Srinivasan (103:24)

Yes, yes, yes, yes.

Benedict Evans (103:25)

It was sort of. It reminded me a bit of like either I write something about like the new kernel memory management thing in Linux where I don't understand it and the people who do aren't interested in what I'm going to say and one else cares. Yes, it gets better.

Balaji Srinivasan (103:40)

Right.

Benedict Evans (103:40)

Or it was like imagine what will happen when it's like talking about open source in the early 90s. Imagine what is going to happen when software is free and there's not. There was. I've struggled and it's actually, it's a thing I've also had writing about AI because I want to kind of. It's not specific about what you think about this. What I'm most good at, I think, or the stuff that I write that people seem to like most is kind of talking about the product strategy of how is this going to work, who's going to win, who's not going to win? How is a corporation or consumer going to buy this? What would you do with it? Right. And I struggled for a while to write about LLMs on that point because it was either like, what are the 30 new papers this year? Or like this is going to transform humanity. Right. And it was kind of hard to find anything in the middle.

Balaji Srinivasan (104:31)

It was in the weeds or super macro with the mesos, or super kind.

Benedict Evans (104:35)

Of messianic, but not much about like product strategy in the middle. And I have the same challenge in writing about crypto in that it's either very, very technical, okay, I've got something for you. Or it's imagine in 30 years or it's about finance where I don't.

Balaji Srinivasan (104:50)

You don't care that much about it?

Benedict Evans (104:51)

Yeah, it's not just that I don't care. It's like I would have to spend six months to get to the point that I know what all the acronyms for moving money between banks are, my opinion about them. So I've never like seen. Well, is that. Is it in a completely different analogy? It's also like talking about chips. Should I get to the point that I understand what's going on in chips? Is that a good use of my time? Would I be able to say anything of value there? And so far I've kind of felt no. There's a bunch of people who know way more about that. Like the semi analyst guys have got it.

Balaji Srinivasan (105:23)

So let me Actually empathize with you in a certain way, which is. I was actually a very late user of social media.

Benedict Evans (105:31)

Right.

Balaji Srinivasan (105:32)

I only got on Twitter in, like, December 20, 2013. Okay. Which is like. Like a decade into.

Benedict Evans (105:38)

Hello, Boomer.

Balaji Srinivasan (105:39)

Huh? Hello, Boomer. Exactly. That's right. No, I mean, the thing is, I got onto Facebook very early because it just was, like, moving around universities or what have you at the time, but I didn't really use it. And the reason is that until 2013, I essentially believed that there was absolutely. I was just a very private person. You know, I was just like. You know, it's weird because I now post a lot or whatever have you. I was just a very private person. And I never. I didn't give any public talks until late 2013 and so on. And I just thought social media was a complete waste of time and that all that mattered was genomics and math and, you know, like. Like hard. Like what people call hard tech now. Like, I was doing genomics and robotics, and I'm, you know, proud of that work. I think it was important stuff. And I didn't see the utility in tweeting my breakfast, and I didn't see the utility and just, you know, petting each other's fur, which is a lot of what people do on Facebook or whatever, you know. Right. So I didn't see the value in any of that. And it was only once. All of that was what bootstrapped the space. All of the fur petting got hundreds of millions of people on there, all of the breakfast, tweeting and so on, until, you know, what actually made it useful and interesting to me was I saw somebody tweeting a summary of a genomics conference at Cold Spring harbor that I didn't have the time to attend, and they gave a much better account of it than any layman would have. It's like, you know, like someone tweeting a mobile thing, and you're like, oh, that's. Those are really great details. And you're skilled in the art, right?

Benedict Evans (107:13)

Yeah.

Balaji Srinivasan (107:13)

And. And then I was like, oh, wow, I can get, like, really detailed information here. Okay, now this is valuable to me as a reader.

Benedict Evans (107:19)

Right.

Balaji Srinivasan (107:19)

What's my point? My point is, I think the parameter that you want to track when you're looking at crypto, crypto is block space. Have you heard that parameter before?

Podcast Host (107:27)

Okay.

Balaji Srinivasan (107:28)

That is the most important parameter in crypto that people outside crypto don't realize governs crypto. Block space is to crypto what bandwidth is to the web. So if you Think about the early Internet or the early web, I should be more precise. In the 90s, like it was very bandwidth constraints. It was 28, 8, 57, 6 modems. And so that's why like Google was 10 to links and I think Amazon even had many images at all. And in fact you remember six degrees. It was a social. Right? So that was a text based social network. It didn't take off because without images people didn't really.

Benedict Evans (107:59)

Yeah, you got nothing to share.

Balaji Srinivasan (108:00)

You got nothing to share, exactly right. ICQ was a chat app that did work. AOL Instant messenger worked because that was just text that could be sent on that low bandwidth thing. It was only in the 2000s that you started to get more graphical things when bandwidth increased, like Facebook. The reason it took off at Harvard, everybody had a T1 connection being at Harvard and they finally had digital cameras know you so you could have photos. And as digital cameras propagated out, so did Facebook. Right. And you go further and further. And like, you know, the Internet only or Internet Explorer only got disrupted by Firefox in like the late 2000s. Right. It was only really by the early 2010s that you had the full JavaScript stack of like jQuery and then only later for React and what have you. So this concept that we have today of like a mobile web app where you can download JavaScript and run an app in the browser on a phone was a vision in the 90s, but it took a long time together because bandwidth had to increase for that. Right. So what's the analogy here? Block space, basically block space is the amount of storage that you have on a blockchain. Like think of a blockchain as like an armored car for data. Right? Because this is data that people want to corrupt, right? In a sense, if, if it's a file on disk, it's important to you. If it's a file online, it's important to others. And if it's a file on chain, it's really important to others. And it's so important that they might try to screw with it. And so Bitcoin came up with like an armored car for data where you could guard the minus one or plus one of who had what Bitcoin. And over time that block space increased so that you could do some basic smart contracts on Ethereum. And now it's increased enough that you can blast millions of stablecoin transactions a day on like BAS and Solana and and so on and so forth. And so, so you should conceptualize it as oh why hasn't this happened happened yet? And instead think of, okay, these applications are gated by the amount of block space. And so they're coming online similar to the amount of bandwidth. You had like text only apps, then you had images, then you had videos. And like Netflix only did streaming video in like the early 2010s. Right. I mean, we think about all that as recent. That's a way of thinking about it.

Benedict Evans (110:07)

I don't have a problem with the idea that you couldn't build Instagram on this because the infrastructure isn't fast.

Balaji Srinivasan (110:12)

Blockspace wasn't done yet.

Benedict Evans (110:15)

I think there's a bunch of like interesting conceptual questions around. Well, what would happen when we got there?

Balaji Srinivasan (110:22)

Yeah. So, so here's a few things.

Benedict Evans (110:23)

Well, there will be also kind of. You're sort of speculating five years in advance.

Balaji Srinivasan (110:28)

Yeah. So my, my view is, I'm not sure if it'll be exactly Instagram, you know.

Benedict Evans (110:34)

Well, it would, I'm, I was like, no, I think we can be sure it wouldn't be exactly Instagram.

Balaji Srinivasan (110:38)

Right, right, right, right.

Benedict Evans (110:39)

But just kind of conceptually, what is the app you could build? Consumer apps, applications you could use. I mean, this is the phrasing I remember you using years ago that one should think of. One should think of a blockchain as a distributed virtual machine.

Balaji Srinivasan (110:52)

Yes.

Benedict Evans (110:53)

And it's another layer of abstraction. That's right. And every layer of abstraction is always slower and crapper than running on the bare metal, except that it allows you to do a bunch of stuff that you can't do if you run on the bare method.

Balaji Srinivasan (111:06)

That's exactly right. That's exactly right. And thing is, blockchains are in a sense one of the frontiers of operating systems research. Like in the same way, like there's an operating system like Windows, there's a browser, which is itself an operating system because you can run apps in it. It's got a full programming language. Like that's how Chrome layer.

Benedict Evans (111:22)

Were you at a 16z when Martin Casado was there? I can't remember if you.

Balaji Srinivasan (111:25)

We overlapped just a bit and we invested a bunch of things together. Yeah.

Benedict Evans (111:29)

Well, Martin had this great observation. You remember when YC said that like for corporations of their companies, 90% of the code was written with AI. And he responded to this by saying, yes, but if you write an iPhone app, 90% of your code is written by Apple.

Balaji Srinivasan (111:43)

Yes.

Benedict Evans (111:44)

So there are all those levels of abstraction.

Balaji Srinivasan (111:46)

Prompting is just a higher level of programming. That's right.

Benedict Evans (111:48)

Yeah, yeah, exactly. And so there's a the. I suppose the, you know, another way of answering your question is like, the finance stuff is there. I can see it. I get it. I'm not sure I can add any value to that. It's interesting and I would tell people it's kind of interesting, but you pay.

Balaji Srinivasan (112:03)

Attention to this, I think you'll be a leader. Go ahead, sir.

Benedict Evans (112:06)

The building more generalized consumer applications on it is conceptually more interesting to me as something that I could make money telling other people about, except that it isn't happening yet and it probably will at a certain point. The curve will curve up, the block space will expand, the stuff will get faster and cheaper and can store more stuff. And people will be able to build stuff on this deterministically. It won't be exactly Instagram. I think that's just kind of a useful, useful mental model for thinking that.

Balaji Srinivasan (112:35)

You could build a consumer app like.

Benedict Evans (112:36)

That, you could build consumer network apps like that on this. At that point, then I think you have a bunch of kind of new interesting questions like, well, is it a good idea to have a social network where all the users have a vote? What would that look like? What problems does that look like?

Balaji Srinivasan (112:50)

Right, right, right, right, right.

Benedict Evans (112:51)

Yes.

Balaji Srinivasan (112:52)

Well, daos are that already.

Benedict Evans (112:54)

Yeah, exactly. Which struck me the other day that all the arguments against that are basically all the argument and saying, no, you need a CEO in charge. Basically all the same arguments as saying, no, you need a. You don't want mass democracy, you need, you need a king and you can.

Balaji Srinivasan (113:05)

Have some balance, like representative democracy. Right. So you have the vote and they vote for somebody for a term.

Benedict Evans (113:10)

Constitutions, which again, like, look at Africa to see how Latin America to see how mixed constitutions work out.

Balaji Srinivasan (113:15)

Well, I'm saying representative, actually, where you have a leader, but they've got a fixed term and there's both of them, for example.

Benedict Evans (113:21)

All of that stuff is fascinating. I. It's like we don't have it yet. And I also. And no one's going to pay me to go to a conference and give a presentation. So it's kind of tough for me to write about.

Balaji Srinivasan (113:32)

Yeah, totally too. I, I will say, all I'll just say is to put on your radar, if you go to like snapshot.org or votagora, there are actually very large treasuries where all that voting stuff is happening on chain cryptographic voting. And so, and so that's, that's growing like stablecoins kind of people ignored stablecoins for a while, just kept compounding. So the the onchain voting stuff is there. But what I will say is that I think just like I was like a late adopter of social media since I just, it had to get to a certain level of significance before I cared about it for the kinds of things I care about. Just I think the kinds of people interested in crypto are either A, they're engineers and they just like the developer as their power users. B, their financiers. Right. Or in some sense financiers or day traders, whatever it is, both the high and the lower. And then C, in the part we didn't say is just like they're political. Right. It's like a political motivation. So I kind of mean like being a Protestant or a Catholic. They have a certain worldview.

Benedict Evans (114:22)

He's also very op.

Balaji Srinivasan (114:24)

Yeah, that's right, exactly. So, so like I have that, you know, we Both like enterprise SaaS type stuff, product type stuff, that kind of discussion. And. But I also like a bunch of other things and you like, you like art museums and, and things like that, which I'm like, okay, that's cool, you know, go have fun. Right. And so we have, we have our own Venn diagram kind of thing.

Benedict Evans (114:41)

Right.

Balaji Srinivasan (114:41)

So okay, so switching gears, I think you'll be more interested in crypto as block space increases and once crypto wallets. Let me actually give you an example of something which it's use for, useful for right now where the block space increased enough, you know, open, open router that, that allows you to try a bunch of different AI models and it just uses crypto to pay for all of it.

Benedict Evans (115:01)

Okay.

Balaji Srinivasan (115:01)

So this way you don't have to have 500 different accounts at 500 different. Because there's so many different AI models, you don't necessarily set up accounts and all that stuff. Right. So it just takes all that account setup process and you just have one account, you pay crypto and it settles it with all these other guys.

Benedict Evans (115:14)

Right. There is a kind of completely tangential thing that just occurs to me as you were speaking is, you know, Ella Marina as this distributed voting system. The thing I always thought would be interesting would be to flip that and say, can you pass a double blind test?

Balaji Srinivasan (115:30)

Yeah, yeah.

Benedict Evans (115:30)

If you take a model that's on the top 20, La Marina and give me a bunch of responses. How many people would pass a double blind test to know which is which? Well, the thing is there's probably some kinds of question you would tell very easily. An awful lot. I bet most people probably wouldn't.

Balaji Srinivasan (115:45)

So the Most fundamental one would be like, what is the private key to this? Or like, basically what is the private key to this wallet? That's something that, depending on how it's set up. We were talking about this in the car. But basically another major use case for crypto is AI makes everything fake. Crypto makes it real again. Because AI can fake all kinds of stuff and give you this very convincing thing on the deep research thing where it said 40% of the phones or whatever you're saying, but it cannot fake the private key. So it cannot show a non zero bitcoin balance or non zero ethereum balance balance without actually having the cryptographic solution there.

Benedict Evans (116:19)

Yeah, but it could probably just tell you that the balance is zero because it might be.

Balaji Srinivasan (116:23)

Yeah, sure, sure, make it up. But what I mean by that is like for example, all kinds of, let me give you, you know, captchas, right? We're websites. So AI can bust a lot of captchas. Now. It can get through. It can. Am I a robot? It can figure it out, get through. But if you had to log in with a crypto wallet that had $1 in it, or $10 or $100, AI can't fake that. It cannot fake the possession of that crypto cryptography. Right. Like to give you one, here's one motivating example for why crypto will get. Maybe this argument will convince you, maybe not, but it's fine, you know, Google login, you agree? Is it billions of users?

Benedict Evans (116:56)

Right.

Balaji Srinivasan (116:57)

But Google login, when you log into a website, you only can log in basically with your email address and the permissions to your Google account. There's something very obvious that somehow even Google with all of its strength has not been able to implement, which is an international bank balance, a spendable balance.

Benedict Evans (117:12)

Right.

Balaji Srinivasan (117:13)

Google login could not have, for rare reason, a spendable balance across different countries. They've solved that for Google itself, where everybody can pay Google and subscribe to Google with a zillion credit cards in all these different countries. But somehow they couldn't make it work. So you could log into a third party site with a spendable balance. Crypto did solve that. Just that alone means that every Google and Facebook login will eventually be either augmented or replaced by buy a crypto login.

Benedict Evans (117:39)

So I'm going to pick up something you said, which I, you mentioned, which I mentioned in the car around what's fake and what's real?

Balaji Srinivasan (117:47)

Yeah.

Benedict Evans (117:49)

So if you're buying an apartment and. Well, so going back a step, like I think most of what Most people follow on Instagram is no longer their friends. It's interest graph.

Balaji Srinivasan (117:58)

Yes, that's right.

Benedict Evans (118:00)

And so do you care if that phone photo is a photo of a real thing or not?

Balaji Srinivasan (118:07)

Sometimes you really do and sometimes you really don't.

Benedict Evans (118:09)

Exactly.

Balaji Srinivasan (118:09)

Yes.

Benedict Evans (118:10)

And I think that's kind of interesting. It's a sort of not so much generative search as generative content.

Balaji Srinivasan (118:15)

Exactly.

Benedict Evans (118:15)

If you're decorating your apartment and you want a mood board and you can specify some styles and you can say, I like this and this and this and this, and it gives you more, and you look and you say, I want more like that or more like this. It doesn't necessarily matter at all if those images are real. It does if, like, maybe you want to buy that table and that table doesn't exist. This just looks like those kinds of tables or it looks like those kinds of chairs or whatever. But if what you're looking for is, no, I want to be more like this or more like that, and you keep going until you get a mood board of exactly what you want. Doesn't may not matter at all whether those images are real.

Balaji Srinivasan (118:53)

That's right. So if it's Pinterest on the one hand, then just inspiration or what have you, but if it is, if it's.

Benedict Evans (118:59)

Shoppable, then maybe it does. Unless you. It's. Then there's an extreme case here, which is just send that dress to Shein. Shein will make it for you. That's right.

Balaji Srinivasan (119:06)

Or let's say, you know, there's some photo of a fire somewhere. Right. And quite a lot of times people will post photos of fires and it's from like some, like a concrete example, the Brazilian fires from a few years ago. There's like a fake photo, like from that, that Macron tweeted out because he. He was told it was a photo of the Brazilian fires, but someone was able to show that it was actually like a. Like a. I think it was like a Reuters image or something, but from a photographer who had died years ago.

Benedict Evans (119:33)

Yeah, it wasn't that image. Well, this is the funny thing about people complaining about deep fakes. It's like, we don't. The problem isn't the picture. The problem is the label.

Balaji Srinivasan (119:40)

This. The label. Exactly. That's right. So. So the thing is that with, with crypto, you can do what I call chain of custody, blockchain of custody, where you can have a camera and by. This is also important in scientific work as well. There's this huge replication crisis with all these Labs and data and you fight the data. Yeah, exactly. Or, or, or something. Right. So you could have, you know, there's something called pre registration of studies where like if you're doing a study you have to describe in some places who you're doing it on, what you're doing. It's like monitored to make sure that people report the results whether they're positive or negative. Right. So let's say it's a, you know, it's a study or it's a camera. You can have like a, either crypto software or hardware in there such that when the frames of images are recorded they're instantly hashed and put on chain either directly or as a digest of some kind.

Benedict Evans (120:30)

Kind.

Balaji Srinivasan (120:30)

Right. That basically is like tamper proofing such that before the data is even like collected or analyzed, this Internet connected thing is doing something. Now it's possible maybe to hack the firmware and mess with that, but it would be pretty hard to. Depending on how you do this, it can be pretty hard to do that.

Benedict Evans (120:46)

You also have this on, you know, Google and so on trying to watermark generated images. The challenge is if the image isn't watermarked, that won't stop people believing it.

Balaji Srinivasan (120:57)

True. That's right. But I think over time time this type of stuff where it'll gain traction at first are crypto oracles for prediction markets. Because if you're making a financial decision, I don't know if you've seen that stuff. Alex Tabrock has talked about this. When people have money on the line, their partisanship reduces and they actually get a different chip in their head where they're like, is this true or not? They're trying to dispassionately figure it out. Right. They're not just cheering my tribe, your tribe, whatever. And is this true? True chip basically means, okay, I'm going to double click into this, I'm going to verify this, I'm going to look at this. And that's where like oracles come in. They're like feeds of data that have some degree of verification and right now they're like mostly price data, but people use it for weather data, they use it for this, that and the other. Right. All these different feeds of information that people trade on and over time I think those feeds, once you can guard price data, weather data, you know, health data, etc. Eventually you can guard any kind of data data. And then now you've got like a chain of custody for data like the scientific data. Rough, you know, rough off it anyway. Why don't we, we should, we should wrap. But this is actually awesome conversation. Anything, you know, what's your latest stuff? What should people go and check out?

Benedict Evans (122:06)

Anything? Well, I've been publishing a newsletter every week since 2013 and I always welcome more subscribers to that.

Balaji Srinivasan (122:13)

You should write. Where is it? Is there gonna be a Bendick book?

Benedict Evans (122:15)

Google Benedict Evans had good Sean book is interesting. I've had publishers approach me every now and then about doing a book. I have to work out what it would actually be and why it would be worth reading.

Balaji Srinivasan (122:25)

Honestly, if you just, I don't know, maybe a history of tech, like, because all your slide decks are very good. Right. And there's one of the things I learned from, you know, my friend Neville. Like the navalmanak, right?

Benedict Evans (122:36)

Yeah.

Balaji Srinivasan (122:36)

That sold a million copies. Why did it sell a million copies? I was surprised, but he was surprised by that. It was Eric Jorgensen went and cured, curated novels, old content and turned into a book. And I was really surprised. I was like, wait a second, isn't that all available on Twitter for free already? Didn't people already see it? They did. However, if you say what is the one work that represents like the best of novel's thought over years, you know, just to see his latest tweets is not the entry point for that. You want to kind of collect all of them, sort them, filter them, organize them, thematically, style them, and so on and so forth. And I think you could have a pretty, pretty good book if you do that. Let me know.

Benedict Evans (123:14)

Well, that's one thing on the list. And yes, the other thing is I, I used to do an annual presentation. I've now shifted my cadence. So I did a, a new AI presentation last month that I published, which I, I was just in town to present. And then I will do another one in the autumn, the fall for American listeners. Great on sort of E commerce, advertising, marketing, brand, like all the other stuff that's being transformed by AI right now. And in general, what do I do? I try and work out what's going on and how to explain it and how I can explain it. And then I go and do presentations and speak at events and talk to companies. And I do slides for money, basically.

Balaji Srinivasan (123:52)

Well, that is similar. I do a lot of slides too. I do a lot of speaking. So you know, I've mentioned the cloud communities thing and materializing those cloud communities. So that's what I'm working on@ns.com like network school. So if people are interested in this kind of stuff, we talk about that. There's so subscribe to Bendick's newsletter at. Is that bendickevans.com ben-evans ben-evans.com okay, great. And then if you want to check out network school, come to ns.com.

Benedict Evans (124:17)

Sure. There you go. So benedictebbams.com is another Benedict Evans.

Balaji Srinivasan (124:20)

No.

Benedict Evans (124:21)

Who is a photographer.

Balaji Srinivasan (124:22)

Really.

Benedict Evans (124:22)

And so my profile picture is taken by him because I used to get his email. This is obviously. This is a blockchain use case. There's a contact form on my website.

Balaji Srinivasan (124:30)

And he won't sell it to you.

Benedict Evans (124:31)

And I redesigned it. I don't need to ask. I redesigned my website recently so it's clearer who I am, but it was quite, quite generic. And people would go to the contact form and they would say, hey, Benedict, we really liked your work photographing Harvey Keitel. Would you like to go to Mexico next week and take pictures of Robert De Niro and I would look at it.

Balaji Srinivasan (124:49)

That's so funny.

Benedict Evans (124:49)

Forward.

Balaji Srinivasan (124:50)

Well, you know what's funny? You know what's funny? There's actually probably as maybe even more biology stream of Austin's. There are, but because there's like 12 people last I checked in like the SFA area alone with my. My first and last name, you know, so just. I feel your pain.

Podcast Host (125:06)

Okay.

Balaji Srinivasan (125:07)

Well, this is great. Really great seeing you in a while. And we should do some more.

Benedict Evans (125:10)

Yeah, great.

Balaji Srinivasan (125:10)

Thank you, sir.

Podcast Host (125:15)

Thanks for listening to this episode of the A16Z podcast. If you like this episode, be sure to, like, comment, subscribe, leave us a rating or review and share it with your friends and family. For more episodes, go to YouTube, Apple Podcasts, and Spotify. Follow us on X16Z and subscribe to our substack@A16Z substack.com thanks again for listening and I'll see you in the next episode. This information is for educational purposes only and is not a recommendation to buy, hold, or sell any investment or financial product. This podcast has been produced by a third party and may include paid promotional advertisements, other company references, and individuals unaffiliated with AC. Such advertisements, companies and individuals are not endorsed by AH Capital Management, LLC, A16Z or any of its affiliates. Information is from sources deemed reliable on the date of publication, but A16Z does not guarantee its accuracy.