a16z Podcast - a16z Podcast: Searching for Mobile’s Own PageRank

Episode Date: March 24, 2014

The mobile experience is still in its 1995 Yahoo phase, a sea of apps and websites without an easy way to find what you want and need on your smartphone. What will be mobile’s version of PageRank, the algorithm that made the web manageable? What is the interaction model and the form - app, card, URL – that will help us find, explore and engage with people, products and services from our mobile devices? Andreessen Horowitz’s Chris Dixon, Benedict Evans and Balaji Srinivasan delve into the options.

Hi, this is Chris Dixon. This is a 16-Z podcast. I'm here with Benedict Evans and Bologi Shrinivasa. Benedict, you've been talking a lot about this idea that we are in a pre-page-rank era of mobile. Can you explain that? Yeah, so I think the issue is really this. When you look at an app store today, it's like looking at Yahoo in 95. You know, you can browse a million apps. You could browse 100,000 websites, but it doesn't scale. and there's no way for you to find the right things that you want to use on that device. There's Google, but Google doesn't cover everything that you can do on a device. App stores don't, you know, the app stores are like Amazon. They list everything that exists, but they're not ways for you to discover and explore that. So we're kind of in the phase that the internet was 20 years ago, that there's all these different things going on on the device,

but we don't really know how to find and explore them. And, you know, in parallel with that, we're also kind of in like a pre-netcape. face because you know what happened on the internet was you had the web browser as a box on your screen and everything happened inside that box and you had a few exceptions like you know Spotify and Skype and stuff but basically everything happened in the browser and again on smartphones that's not the case so there's this kind of swirling mass of how do we find stuff how do we explore things what is the interaction model is it an app is it a card is it a web app Is it a URL, is it a messaging app?

How do you find stuff and explore and engage and acquire customers within that? So I think there are at least two ways to interpret that. One is no one's invented that yet. And the second is the more pessimistic view is it's a fundamental architectural limitation of the platform, which is there's no browser because there's no HTML, right? Because obviously there is HTML in the web browser, but I'm speaking about apps. You know, the app model of having to download software, there's no way to really deep link sort of on Android, not on iOS. So you don't have the same linking structure and you

don't have the same sort of openness and open standards. You don't have a universal resource locator in a sense for every single piece of content. And you can see Google poking around this with deep linking into apps. But I'd almost kind of turn that. Even if you deep link into an app, you still have to have the app. Yeah. So this whole idea of download as opposed to a page load, right? I'd kind of turn it the other way. I mean, there's an ideological issue in here because there's a whole group of people that will say, well, webs are like a hack and a fork of the web, and everything will converge back on to web in some form.

Not HTML, but actually web pages with URLs and you can click on and go to other URLs, because obviously half of apps are actually built on HTML. I'd kind of propose turning the other way around and say, well, the reason that we only had the web on the desktop internet was because it was impossibly difficult to do anything else. And you had to have this incredibly strong selling point to get somebody to actually go and download an application and install it.

Well, and you had to, in many cases, go to the store, literally, and buy a shrink-wrapped. Yeah, exactly. Yeah, right. Yeah, kids, there were these stores that sold, like, physical stores that sold software. In cardboard boxes. So, yeah, I mean, the reason that we only had the web was all that there was, was because it was so hard to do anything else, and now it's incredibly easy to do other things. And now smartphones themselves are communications platforms in a way that PCs never really work.

So, I mean, you see this really obviously with WhatsApp. because it taps into the phone book, it taps into the photo library, and a push notifications, gets an icon on the home screen, and there's a whole bunch of other applications is trying to do the same thing. And so the smartphone itself becomes a social platform.

Whereas Facebook, you know, whereas on the desktop you had the browser, and then inside the browser you had Facebook, and then inside Facebook you had a social platform. Whereas on a smartphone, that's turned upside down, it's the device itself is a platform. So you're not limited to the browser. You can do all these other things.

And you give up stuff in return, for that. There's pluses and minuses on both sides. Isn't one outcome here that the OSS have all the power that iOS and Android? Is it the OSS is? Is it Facebook? Is it line? Is it, you know, is it a combination of all of those things? And how much does that carry on changing? I mean, the, the interesting thing, looking at the bunch of rumors about, you know, what Apple would do about wearables and we're looking at what Google is doing with Android wear is around the, that these interaction models carry on shifting.

So the way Android Wear is, you've got this device, which is a computer on your wrist, but it doesn't run native code. It just displays notifications that come to it from an Android phone. Now, if you look at what Apple did with Passport, you look at some of the rumors about what they might do with Health Book, you see them pulling away from a native app and doing like a kind of more of an XML card sort of application, which isn't necessarily installed from the App Store. I mean, if you look at Google Now, Google Now is, I see, now much, probably HTML, but it's not the web in any meaningful sense. If you were to go and stand outside a restaurant,

you can pull your phone out of your pocket, should you have to open a restaurant review app and wait for it to get a geo-lookup and then get a list of 10 restaurants and then pick the one you want, or should it use Bluetooth eye beacon to know where you are and pull up a view for that automatically, whether that's from within an app or from the web or scraped out of Google Now or something.

So, you know, one thing I think is in terms of discovery and kind of thinking about architecture and so on. So there are companies like, you know, like URX, for example, that are doing like deep linking within apps. But I think more fundamentally, so if you think about, for example, Google Ware and, you know, the phone and so on. From an implementation perspective, you often think of, you know, model view controller. You think of an app as being a view on top of an underlying data store, which is sending things to, you know, a smartphone view and a tablet view and this view and that view. And so one of the more interesting,

projects out there is Erbit, which is thinking about, okay, so we've got the web, which is linking between pages. How did you really link between apps, right? One of the ways that you'd want to do this is you sort of want to make it so that you don't have to generate and think up and document an API for every app, right? So what if, for example, you had, let's say, a phone and it had an internal directory structure, let's just say you have a laptop, it's got directory structure, and it's got front slash contacts, and it's got front slash files, and it's got front slash personal information and so on and so forth. And this file system is what Dropbox accessed and read and wrote to for your files.

And Facebook accessed and Red wrote to for your files, for your photos, but also for your contacts and for your personal information. And LinkedIn also accessed this, and Twitter accessed this. So the data was actually local, and the APIs were sort of automatic, and these applications would read and write to that. And rather than having all the data locally in their own databases, it'd be the billion phones would be this distributed data center, right? And so what's interesting about that kind of model is now you're talking about links

between not just web pages, but between applications based on like data interstices, right? So that's one model for where the future of sort of like mobile architecture is going to go, where you have data local computing, and then, you know, you turn your large centralized, you know, web services into really more like large caches of local data, which is going to be stale, but it'll be updated next time you log, into the website and then they'll download it. So that's like a total inversion of control of where we are now, but that's something I could see happening in like five years, ten years.

I think another way of thinking about some of the stuff you're talking about is if you contrast what Apple and Google are doing just in the short term with the way they're running their platforms. Because Apple runs the whole stack, they're pushing a lot of their innovation into the integration of the hardware and the software with all the Bluetooth stuff that they're doing, all the local wireless stuff based obviously, but also things like the fingerprint scanner.

whereas, which is very hard for Google to do because they've got no idea what's on the phone. It's exactly like Microsoft. They didn't know what PC you've got. Whereas, of course, what Google are doing is lifting everything out of Android and putting it into the cloud, both into Google Play Services but into Google Now

and all the other stuff that they're trying to do that. ChromeOS being the ultimate version. Yeah, exactly. So there's a question, you know, it's sort of one of my kind of the things that I've been saying is that if I was to say I installed an app on my Android smartphone in five years, I don't really know what that would mean.

you know, what would install mean, you know, what would app actually mean? Right. What would Android mean? You could just be enabling it in their services. Yeah, exactly. What would you really actually be doing? And you could say that actually you could probably say the same thing for Apple. I can't be completely confident that if I say I installed an app on my iPhone,

I meant that I went to the App Store and initiated a download at a package of Objective C code. You know, there's probably going to be other stuff that might be going on as well. And religion of that basically is that, you know, at Google, right, like then you remember, been stepping down and Sunder Pitchai now running apps and Chrome and Android, right? Obviously, I think it's, you know, like there's going to be some fusion of Chrome OS and Android now that they're both folding into the same person. One plausible scenario is that every Java API in Android gets rewritten to be a JavaScript API for ChromeOS so that you have a migration path, right?

And then, you know, that scenario that you're talking about, like, where, like, what does it mean to install something? I thought Google fought a whole lawsuit saying they weren't Java APIs. well okay that's right like whatever you want to call them Google Java or something like that legal word and there's an engineering word okay yeah so the Android APIs right so Android like the Java syntax compatible APIs right

those I mean if you were doing a mobile platform today you would not write it in Java you'd write it in probably JavaScript right just because JavaScript's getting fast and Google in particular has put all that effort into trace trees and actually getting that into production with a V8 compiler and so I mean there's one there's one school of thought thought that the web can never have the same ux performance characteristics of native apps but that just seems

well I think the wrong side of Moore's law it feels like it's it will probably always be true technically but it might not it will matter less and less right you know it's like you know it's the same thing all these you know all the arguments between different programming languages or between different network technologies or something you know you throw enough resource at it and eventually the problem kind of goes away and you know you see more you know it again, to me, what programming language is almost the least interesting question. The puzzle is, you know, what is the interaction model?

You know, how does that icon get onto your home screen? You know, do you want that icon onto your home screen? Does it appear there automatically? Do you have to enter a user ID and password and wait for a download? Does it appear there as you walk past the store? Does it bubble up from the bottom of the screen onto your watch? And so those kind of contexts, you know, trying to focus too much on, well, what's the app distribution model,

or how can we deep link into apps, sort of, fades away. I mean, the thing I would sort of, I wrote a blog post about this the other day, but the kind of the thing I wrote on Twitter that prompted the blog post was, you know, who is it that's going to kill the native app? It's going to be Apple. Apple is the pump for me. I think it's going to be Google. Well, Google might try, but I could perfectly well imagine Apple turning around

and, you know, at a certain point, and Apple being the one, who kill. Interesting. That is counterintuitive. Because they don't make any money from it, obviously. It's purely about the platform. But if they were to decide, okay, here is another bunch of things that we can do that will make our platform stronger, that will make it more compelling to develop for iOS,

that will make it more compelling to buy an iPhone. At a certain point. They'll do that. Yeah, sure. At a certain point, if the pressure gets there, insofar as we're throwing out, like, counterintuitive projections, so certainly obviously a large component of the future is mobile. I think a countertive, but true, if you reflect on it, projection is that a significant infraction in the future is also going to be sessile. And what I mean by that is it's going to be

people who are wearing Oculus and they're not mobile at all and they're just sitting there and they've got very powerful computers that are pushing away at it. So I think that's like return of the laptop, return of the server farm, even return of the desktop. So Oculus is is pretty interesting there.