Answering The Internet Problem | Matt Dalio

I want to walk you through sort of a series of evolutions, which is probably about 2 years of evolution, as to how we found the answer for internet. So micro epiphanies, each of which felt incredibly large at the time, each of which is sort of just a little dot on a trajectory. So the first thing I want to do is I want to ask, how would you guys solve internet? I would build an internet between the computers in there. So a mesh network between them? Yeah, between those. OK. What are some other ideas of how you solve internet? Maybe like a contract with the carrier somehow, so you provide internet in return of something. Contact with carriers, great. Put a connection with your smartphone, great. An open free basics, something like what Facebook's trying to build, but an open one, not open free basics. Cool. So a network of companies who do that. Cool. So I'm going to connect the dots between a couple of the things. Did you have one? There's the Google balloon. Google balloons. Hack the telephone lines. Telephone lines. Hack the telephone lines. OK. Connect to the airline. OK. Cool. So I'm going to connect the dots across a couple of these things, because a lot of these actually really— 2 ways I can do it. One is to paint the picture of what it is. The other one is paint the picture of how we got there. Do you guys have any preference about which one? How you guys got there. Sorry? How we got there? OK, how we got there. I'll try to make it quicker. So we had a series of really simple insights. The first was I once stumbled across how large Wikipedia is. You can take a segment of— so 3% of Wikipedia entries account for 80% of Wikipedia searches. That's about 1 gigabyte's worth of data if you include all the images. So it's amazing. Let's just put Wikipedia inside. We'll just make a Wikipedia app. And then I went to Maker Faire, which is a cool thing. If any of you have never been there, go to Maker Faire. It's awesome. I think it's coming up soon. And the Wikipedia table had a book this thick on 3D rendering. And I realized, OK, it's literally all Wikipedia articles on every aspect of 3D rendering. It's basically a textbook on the topic. What I realized was, wait a second, you can repackage Wikipedia into little sub-applications. So you want the history application? Great, let's just take the stuff that's already in the device and repackage it into this thing. And so fast forward, we realized, well, OK, why wouldn't you also be able to take other content? So there's Hesperian Health Guides, which has really great health content. So now all of a sudden, you can have an application that has every ailment, every medicine, every major first aid issue, no internet required. And as that started to evolve, what we realized was that applications are really quite simple. They're 2 things. They're a database where you store the information, and they're templates where you view the information. And there's only so many ways of viewing content, and there's only so many ways of storing content. So if you were able to build an engine that allowed you to easily ingest and store content and then have templates where you say, great, health application, recipes application, construction application, farming application, and on and on and on. You can basically create a whole ecosystem of applications that doesn't require internet. The next insight we had was you can refresh it. Now why is refreshing it such an important insight? Well, because along with that insight came another insight, which is that Carriers— so here's the core problem, actually, of internet connectivity. I'm going to define it even more precisely. That will help make the solution actually make more sense. Broadband is not everywhere. Stringing a wire to every home on the planet is just something that's not going to happen. But 3G, 2G, 4G are— I believe the latest number, it's almost 90% of the planet is is covered with some form of wireless signal. So let's use that, right? Seems obvious. Here's the problem. It costs a lot of money. And a PC user uses a ton of data. So I was just in someone's home in Brazil, and I say, how much data plan do you have? I'm curious, actually. Who has over a gigabyte of data on your mobile plan here? I do. Okay, who has over 2 gigabytes of data here? Who has over 4? Not including unlimited because they don't make those anymore. The grandfathered people don't count. So who thinks their data plan is expensive? Okay. So to give you an idea of their— 400 megabytes is what this person spends. on their phone. They would use it for about 1 week out of the 4 weeks in the month. That's all they could afford, and so they'd use Facebook and WhatsApp for a week, and then it's up. 400 megabytes. The average PC user uses 2 gigabytes of data every day. They use 60 gigabytes of data per month. What that means is if you want to be a normal PC user using the internet through a browser, You are done with her data plan by breakfast on day 1. Your coffee's over and you're done. So that's what we have to solve for. We have to solve for how do you create a computing internet, an experience that works really well, that gets me what it is that I want on, 400 megabytes of data. Turns out that's actually doable. How do you do that? Well, the first thing is that most of the things that we're searching for every day are actually the same things. It's either one of a category of things— statistically, Wikipedia, 80% of the things we're searching for is the same stuff. And then second, it's like we're reading the same news articles. We're reading the same things. So here is where this little refresh functionality comes in. The first thing you do is pile on all of the things that you can possibly imagine. Have basically an internet without internet. And the refresh functionality— carriers, every night while all of us are asleep, have 97% of their capacity sitting idle. Refresh it. So that you wake up in the morning and your news app is all there. Your Twitter feed is full. Your sports scores are there. Your weather is there. The next month's worth of educational videos that you needed for your school, also there. Everything else, I pay per megabyte. So that data plan, Now we're speaking with piles of carriers. Every single carrier we've spoken to says, yes, we want to make that. Why? Here's where business model comes in. It's super, super important. You have to be realistic about business. OK, what is their incentive? Their incentive is to sell data plans. We will help you sell data plans that today you can't sell to a whole market, which by the way in an emerging market is most of your market. Mm-hm. And it will cost you not a cent to push that data plan. We're using existing infrastructure. So the cost of that data plan, a couple dollars a month. Is that a couple dollars a week or a month? A month. And one of the things that I should clarify is as we talk about people in emerging markets, it's really easy. The first thing that comes to a lot of people's minds is like the Oxfam ads and starvation and poverty. And that is not actually statistically most of the world. Those are the pictures, but that's like going to San Francisco and taking pictures of homeless people and being like, this is San Francisco, or going to coffee shops and being like, this is San Francisco. And like, no, most people are somewhere in between. And so statistically, if you look at the world, it ain't a pyramid anymore. It's a bell curve, right? So 40 years ago, the bell curve was skewed so far down that they called it a pyramid. And in the past 40 years, we've all heard the statistic about how billions of people have been brought out of poverty. What does that mean? That means that moved up here. It's now a normal distribution. Most people are in the middle. China, great example. China 40 years ago, everyone was in poverty. China today is a prosperous country with people who have, people who don't have, and most people in the middle. So a couple dollars a month, totally affordable. So So this insight came along and what we realized was it wasn't enough to build an operating system. We actually had to build like almost an alternate internet, to make it sound more grand than it is, but we had to build a whole infrastructure of how content comes in, gets sanitized, turns out this is really tough stuff, to then build a modular framework so that you can click, click, click, build apps with a tiny team, with a tiny, tiny team. So there was another big problem that we had. So now we've got an operating system. Hardware manufacturers, it turns out, move in herds. So their margins are 2%, 3%, 4%, 5%, which means they have zero margin. for error, which means they do what they know will sell. If they know this will sell and they know that they can now crack open a whole new market who couldn't afford data plans, they will make these. If they don't know that, they won't make these. So there was a company in Brazil called Positivo. It's the largest computer manufacturer in Brazil. And the guy who was their head of innovation, awesome guy, within the first 10 minutes was like, I've got it, that's it. Like he saw it, all he does is see companies come through his door day in, day out, and he's like, all of them are junk, this is real. We built a relationship with him over a very long period of time, and finally he went to his CEO and he put his, literally, Not literally, figuratively, he put his badge on the table. If you don't do this, I'm leaving. They didn't do it, he left, he joined us. They still didn't do it. So here it is, we have an operating system underway and we don't have anyone who's willing to ship it. What do you do? The only answer was, we've got to build hardware. Every person that we spoke with was like, Matt, you have no idea how hard hardware is. They were right. Hardware is freaking hard.