Android fragmentation and the cloud

Game Oven's post on Android gyroscope support is a nice illustration of a general issue (another good illustration here): Android fragmentation is both massively overstated and massively understated, depending on what you want to do. 

On one hand, Google has been quite successful in reducing the impact of Android version fragmentation. Around three quarters of the Android devices that hit Google Play are running 4.X, but more importantly, Google has moved its own key services (maps, payment, notifications etc) out of the OS itself and into a software layer, 'Google Play Services'. By putting key platform APIs into a software layer that can be updated in that background over the air, Google has reduced its dependance on OEMs to produce firmware updates - it can update its own tools any time it wants, across the great majority of phones. When Google announces new APIs for Maps or notifications at this year's IO, they'll be available on devices running year-old and two-year-old versions of Android. There is no more fragmentation if you're using Google's cloud.

On the other hand, hardware fragmentation is if anything accelerating. This chart from OpenSignal, from last summer, is a nice visualization of the market dynamic. 

Android fragmentation isn't of itself a bad thing - it's inherent in the choices that Google made. This is what 'open' and 'choice' look like. And I doubt if it's possible to have an 'un-fragmented' device landscape that includes both $600 devices and $50 devices: some scattering in capability is part of the deal. If you want to have thousands (literally) of OEMs, and a huge range of choice and price points, well, you're going to have different devices with different capabilities. 

This is only interesting, then, to the degree that it has broader consequences. The consequence of Apple's approach is that pretty much everything behaves in predictable ways, but you have a very narrow range of devices at a narrow range of prices (and screen sizes), and that severely restricts the addressable market. More people can afford $50 phones than can afford $600 phones. The consequence of the Android approach is that you have a much wide range of devices and prices, and a much larger market, but anything on the bleeding edge doesn't work predictably at all. This doesn't just apply to the gyroscope - it also applies to varying degrees to almost anything trying to do clever things with the hardware. This is also true even if the API does actually work as advertised - there's not much point trying to do a mass-market Android NFC deployment when you have no idea how many of your users even have NFC Androids (and the users themselves don't know).

One result of this, as I've said before, is that Apple and Google are focusing their innovation in different areas. Apple is moving down the stack with integrated hardware/software experiences (iBeacon, fingerprints, M7 etc) that are hard for Android to match, and Google is moving Android up the stack with Google Play Services, the cloud and machine learning, which is hard for Apple to match. 

The paradox for developers, meanwhile is that the more open and extensible platform can actually be harder to hack on. When you buy two Samsung phones of the same model and brand in two different countries and find they have different camera drivers and your app will crash on one or the other (or both), where do you focus your seed funding? You have limited resources and limited time and you need to hit milestones to get your next round, after all. Plus, the users who want to install your cool new apps are still concentrated on the iPhone. Again, this is a paradox: Android is the platform best for early adopters and iOS the one best for late adopters who just want something that works, but the market adoption is the other way around. That's one of the reasons this chart is both unfair but relevant.

Which type of innovation is crucial to a platform? With the current dynamics, people like Game Oven are going to keep doing iOS first and Android second (if at all), and that keeps the majority of the best users on iOS and Apple's machine turning over - the classic ecosystem virtuous cycle. But if, as many people suggest (Fred Wilson most recently) the most interesting and important innovation will happen in the cloud, then Google's tradeoff might win over Apple's trade-off - Google's comfort zone beats Apple's.

There's another paradox here, though: if all the best stuff is happening in the cloud, then you'll buy the device based not on apps and developer support but on design, quality, fit and finish... that is, all the things Apple always leads on. The web saved Apple 20 years ago, because with the web you could choose the best hardware and UX regardless of the ecosystem - you could buy an iMac and not worry about the software. So if everything goes to the cloud again, is that really an existential problem for Apple?

(As an aside, note this post I wrote a year ago about the different issues facing iPhone and Android.)

Notes on TV

One of the reasons it's difficult to talk about the future of TV is there are really several separate sets of issues that are in play, which are pretty much self-contained, and depend on quite different factors, and all of which need to move before things can change. 

The US problem

First, the USA is a massively over-served Pay TV market. Pay penetration is very high (90%+), ARPUs are very high by international standards, bundling is very rigid, and so there are lots of people who feel obliged to buy much more than they really want.

Hence, there is a lot of pent-up demand for some sort of unbundled approach - to be able to get just the channels that you want and pay less. But the bundling model is very deeply embedded in the structure of the US TV market, at multiple levels of the value chain, and there are some very strong incentives for a lot of industry participants to continue with the status quo. In other words, everyone hates the way the US TV industry works, except for the US TV industry. This makes the current model very rigid, but also of course potentially very brittle. 

Tech industry attacks on this model have tended to come from the distribution side - they put together a new distribution platform and then go and ask content owners to give them the content to offer an unbundled service. But they very quickly discover that coming to Hollywood with a distribution platform doesn't get you a cup of coffee - you need to propose a more profitable economic model, and back that up with cash on the nail. A physical distribution platform itself is not where the value is - it's the possession of a huge audience or valuable content that gives you power. And it's tough for a platform with no customers to offer better economics than a platform with tens of millions. 

More broadly (i.e. beyond just the US bundling issues) it's helpful to think about TV as a virtuous circle. Audience gives revenue, revenue lets you buy great content, and great content gets you more audience. This is self-sustaining at each level. A big TV channel is big because it's big, not because it has access to a linear distribution path. 

This is a little like orbital mechanics: if you want to go to orbit you need to burn a lot of fuel, and the higher you want to go the more you need to burn - and having a better-looking rocket doesn't make much difference. Money for content is the fuel of the TV business. 

Hence, the interesting thing about Netflix is not so much the physical distribution platform as the use of data to attack the cost of content - if you can make hits more reliably you can get the same audience for less money spent, since you waste less of it. To the TV industry Netflix would look like just another TV channel without the use of data.  

Finally, if the US TV market is over-served, most others are not. The UK is arguably a 'goldilocks' market - half of households have pay TV and are broadly happy with it, and half don't and are broadly happy with that. Meanwhile some European markets are probably underserved for pay TV - more people might like a pay TV product than are currently being given one. So the whole US 'cord cutting is the future' discussion is not necessarily broadly applicable. But the size of the US market (and the physical location of most of the tech industry) tends to shape the debate. It may also focus attention on the wrong problems.

The user experience problem

Next, there is the user experience. It seems pretty clear we're in a 'pre-iPod' phase at the moment. That is, all of the technology is in place, more or less, but no-one has quite managed to package it up in the right way to give the right user experience. There isn't yet a totally fluid way to browse, choose and display what you want on your TV. Lots of people are poking around it (including Apple and Google) but it doesn't seem like we have that magical 'aha' moment just yet.

There are also a bunch of route-to-market problems here. Integration inside a TV is great but TVs are only replaced every 5+ years. Pay TV tech, unlike mobile, is a balkanized mess of standards, with no GSM/UMTS that you can implement and sell globally. Pay TV operators are the gatekeeper to any other equipment in the living room (at least in markets where pay has a large share) unless you can make it really cheap - a couple of weeks before the Chromecast launched I suggested that the next Apple TV should be a $50 HDMI dongle, and the tech for that is moving forward in interesting ways. And of course we also have tablets themselves, which are certainly replacing second TV sets and may take a real share of primary set viewing too.

At the moment there are a lot of rumours that Apple has a new product - but the rumours mainly focus on how it would deal with the unique US content ownership and distribution structures, not the user experience, which is actually the real question. That is, in the US a big part of the user experience gap is the simple availability of content per se, but elsewhere that is much less of an issue. In the UK, for example, all the main broadcasters put the last 7 or 30 days of content online for free, on any device, with no restrictions or messing around. iPad, iPhone, Android, Smart TV, games console - anywhere, any device, any time. And yet peak streams on the BBC's iPlayer service are 600k or so, where peak linear TV viewing is over 20m, and linear TV viewing shows no sign at all of declining. 

How do people really want to watch?

This in turn points to another question, and perhaps a slightly subversive one: how do people actually want to watch 'TV' (or whatever we call it)? Hundreds of millions of normal people really do just come home, turn on the TV and watch whatever's on - if you offered something less passive, do we really know how many would do it? That is, the idea that no-one would watch linear broadcast TV if on-demand worked 'properly' (whatever that might mean) is really just an assumption. 

The really big question here is how TV viewing would change if you did move from the current model of TV as a largely undirected, passive experience, to one that required (/'allowed') you to make choices. If you come home and turn on a random piece of generic light entertainment you'll watch it, but you might never choose to watch it, much less search for it. So is that a bundling problem or a recommendation problem? Should we think of TV viewing hours as propped up by filler shows in the same way that CD albums were full of filler tracks, and that if we go to a fluid on-demand environment people might just stop watching that filler? Or would the right passive programming system - 'Pandora for TV' replace one passive experience with another, more tailored and targeted one, with the greater accessibility of long-tail content taking up the slack? Of course, a lot of TV channel branding and programming is about just this - in effect a lot of TV is 'Pandora for TV'. Either way, this is really about unbundling shows from TV channels, not unbundling channels (or on-demand channel brands) from cable TV subscriptions. And (looking back to Netflix) how would that cascade back though the TV production system? How many fewer shows might be made? How would they be funded? And what would happen to the 'golden age of TV'?

Android instability

If you were a PC OEM from, say, 1990 to 2010, you operated in a very clear ecosystem. You outsourced much of the innovation to Microsoft and Intel. You knew exactly what WinTel were doing, both because their roadmaps for the next few years were actually public and because Microsoft and Intel had very clear and widely understood strategies. Moreover, the core, fundamental strategies of OEMs, Intel and Microsoft were pretty much aligned. Everyone wanted more PCs to be bought, and preferably a good number that were high-end and high-margin. Intel, Microsoft and CloneCo all lived for the same things. CloneCo didn't necessarily make great margins (and eventually got killed by Dell, perhaps), but it knew what the game was. 

The Android ecosystem today is superficially similar to the PC ecosystem, but I'd suggest that the clarity and alignment of interests of the PC ecosystem isn't present in anything like the same way. As an Android OEM you have very little idea what Android will be in 3 years - partly because Google itself may not have a fully-formed idea. There certainly aren't public roadmaps stretching out years in advance. 

It's also questionable how much alignment of interest there is. Google certainly wants Webkit everywhere, and arguable Android everywhere (or, more precisely, Google Plus everywhere). But that doesn't translate to a burning hunger for an aggressive phone replacement rate at high prices. Indeed, Google Play Services reduces Google's interest in device replacement as a way to drive service penetration. As an Android OEM, your ecosystem creator doesn't benefit directly from the health of your industry. A healthy PC market was Microsoft's driving objective - 'A computer on every desk and in every home'. That's not quite the case for Google and the sales of Android smartphones - they're reach, and a means to an end, but not the reason why Google exists. 

Next, it's not clear what a sustainable position for an Android OEM looks like. All the brands except Samsung are sub-scale and failing, and while Samsung looks dominant it is clearly feeling paranoid: the growth of the Chinese Android OEMs outside China is a huge question. Lenovo has made its first move by buying Motorola but the real story is whether 2, 10 or 100 others follow it, and if so how. 

Finally, Google's control of Android itself is a question. Amazon forked it, but with limited broader effects. Almost all Android in China lacks Google services but then Google is largely absent from China anyway. The ways that forks of Android might become relevant outside China (and Google's tools for preventing this) are complex and a topic for another post, but we can't rule this out. Indeed, a lot of the most interesting ecosystem innovation is being done on top of Android rather than as a would-be competitor to it. 

On one level, then, the smartphone platforms wars are over - iOS and Android both won. But actually, nothing is finished - we just move onto new questions. 

Interaction, canvases and ecosystems

Suppose that in 5 years or so I send you a Yelp review of a restaurant, from my phone to yours. What will that mean? 

  1. First, I might well use something like Airdrop, or touch my phone to yours to pass it across, or tell Now to give it to you, or indeed Now might decide to give it to you without my even explicitly asking. Or the review might be invoked by a Bluetooth LE beacon as you hold your phone next to the menu on display by the door
  2. But for the sake of simplicity, suppose I send it to you using an internet messaging app - either one built into the OS or a third-party one - Facebook, Whatsapp or more probably one that doesn't exist yet but by then has 15 engineers and 1bn MAUs. 
  3. It seems pretty unlikely that you'll see a dumb URL string on your screen. Rather, you'll get something rich and interactive, within the message.
  4. And you'll be able to go into that experience and tap the number to call the restaurant, or make a live booking, or swipe through photos.
  5. And if you tap 'book', it'll pass them a $10 booking fee in bitcoin, authorised with a fingerprint swipe. 
  6. Now suppose you decide to save this item, as an icon on your home screen, or some other yet-to-be created place. 

Now, what were you using? An app? a widget? Native code? What programming language? Did you install an app or surf the web? I'd suggest that none of those questions would really mean anything, at least not as we think of them right now. The programming language matters much less than the user flow. And some of this example sounds 'webby', but Google is the first to advance interaction models that are not remotely webby (such as Now). 

This is a pretty simple illustration (an expansion of the super-hot card metaphor) of a broader point I've made before: on the desktop internet, the web was by far the dominant model and that didn't actually change very much for well over a decade (before that, the interfaces of Windows and Mac were also very stable for a long time). But on mobile, not only are other models just as important as the web, but they're not remotely stable, settled or mature. The platform war may be over but that doesn't mean things are settling down. 

So I have very little idea what precisely I would mean if, in 5 years, I were to say 'I installed an app on my smartphone'. Further, I'm pretty sure that if it's an Apple smartphone it will run an iteration of iOS but I'm rather less sure what Google will have done with Android and Chrome by then. And of course I might be running a fork of Android from Amazon or, perhaps, Microsoft. 

This is the key reason why the new social messaging apps are so interesting - not because they have users and inventory now, but because they can be vectors for some of this sort of behaviour - a third acquisition, discovery and distribution channel besides Facebook and Google.

This may also have implications for any discussion about what it means for Apple that its ecosystem will have a minority of mobile users. We need to think about what it means to call a ecosystem that might have 800m-900m live devices 'minority', but we also need to think about what 'ecosystem' might mean. What, if any, 'winner takes all' dynamics operate in this environment? One reason the Mac didn't die was because the web changed what it mean to be a computer ecosystem: the mobile ecosystem has lots of changes to come too. 

Ecosystem maths

I'm pondering Apple's results, sitting in a cafe in-between meetings. It's been apparent for years that Apple was camped out at the high end of both tablets and phones, and that Android would take almost all of the rest. But it's worth working out on the back of a (rather large) envelope exactly how that would play out, assuming that Apple's pricing strategy doesn't change and of course that nothing else changes (which of course are rather unsafe assumptions). Hence: 

  • Right now, on the basis of a 24m replacement cycle, there are perhaps 290m iPhones in use on earth. Depending on the second hand market, this might be larger. 
  • Apple has sold 195m iPads - perhaps 180m are still in use. 
  • There are also a fairly small number of iPod Touches in use - perhaps 20m

That adds up to a rough estimate of 490m live devices. For comparison, around 900m Google Android phones were sold in the last 24m, and probably another 110-120m Google Android tablets ('Google Android' = 'not China'). 

Where might this go? Apple now has about 10% of global mobile phone sales, rising steadily. It's important to note that Apple is not losing share of phone sales to Android - it's just not taking as much share as Android. There are between 3.5bn and 4bn people with mobile phones on earth (there are far more connections but many people with multiple connections). This number is also rising slowly, but all the growth is from the very low end. 

So. 

Over the next few years the great majority of that 3.5-4bn will convert to smart (and indeed the more important variable is affordable data plan penetration rather than smart penetration). If Apple continues the current strategy and share growth, it will end up with (say) 15% unit share. 15% of 3.5bn is 525m. (I told you this was a BOTE calculation.)

The great majority of the rest will go to Android (though quite what 'Android' will mean is an open question). That means perhaps 2.5-3bn Android phones in use. There might be some Windows Phone as well (assuming it doesn't become an Android fork) but we can ignore it for these purposes.

No-one has much idea what the total addressable market for tablets really is, let alone premium ones such as the iPad, and the recent sales trajectory is somewhat lumpy. If we assume a four year lifespan for iPads as the tech stabilises and look at the recent run-rate, that suggests a stable base of, say, 300-350m. This gives us a base of perhaps 850m iOS devices, with a lot of ownership overlap.

There will also be an ungodly number of Android tablets, of course, but we know neither quite how many nor what they'll be being used for (right now, mostly TV, it seems). 

What does this mean? What does it mean for Apple to have a platform with a minority of users, indefinitely, in mobile? An minority ecosystem with only 850m devices? Or even 490m?

Certainly, this isn't 'Windows versus Mac all over again'. There are now 490m iOS devices in use, but PCs only hit that number in around 2000, long after Apple lost the last ecosystem battle. Apple sold 51m iPhones last quarter - total PC sales in 1995 were 59.5m. That is, the iOS ecosystem now is much bigger than the winning ecosystem back then.

Even beyond that, all the other dynamics are different - the smartphone market is not driven by corporate buyers who demand commodity product based on bullet points and don't care about design or user interfaces, for example. The relative market share of an ecosystem is relevant, but it's not the only thing that's relevant. We need also to think about value share, engagement share, and all the other dynamics that drive the viability of an ecosystem. The assertion that an ecosystem with close to 500m users now and over 800m in a few years will not be viable because there's another that's bigger seems pretty simplistic. It can't be taken for granted that any 'winner takes all' dynamics will work like that. 

More importantly, though, these questions will probably have become irrelevant by the time we find out the answer. That happens quite often in tech. To me, the platform wars are now much less interesting than what happens on top of the platform. On the desktop internet, we had close to 15 years of stability with almost all online activity going through the web browser, but on mobile it is far more complex already and also far less stable. Nothing is settled on mobile. I have no idea what it will mean in 5 years time to say that I 'installed' an 'app' on an 'Android' 'smartphone'. All of those terms could change completely, and with it what it means to say 'ecosystem' or 'market share'. 

Note: for supporting charts, see this post

Mobile interaction models

The interaction model for the desktop internet was pretty much settled 15 years ago. It turned out that the answer was a web browser. Stand-alone apps such as Pointcast were a mostly blind alley, and while apps persisted for email and IM, and for very specific things like music, the words ‘web’ and ‘internet’ became effectively synonymous to anyone non-technical. Over time we added Ajax and better search and better social, but everything really happened inside the browser.

In mobile this is quite different: nothing is settled. We have the web and apps and of course app stores, and then we have many complications - voice, in-app payments, web apps, hybrid apps, widgets, push notifications, social messaging apps, Google Now and Siri. Then there’s the hardware layer - images, barcodes, NFC, bluetooth, location, motion sensors etc. Innovative and disruptive new interaction models can very often find a route to market, far more easily than they could on the desktop internet. Sometimes, they scale to a hundred million users in a year to two. And we have more and more waves of innovation coming, with things like local wireless from Apple and deep linking to within apps from Android, and a very fast-evolving social messaging space, and more things in 2014 and beyond.

So, we can actually have a pretty limited idea of what the dominant interaction models will be in 5 years. 

(There is a dream, of course, that all these nasty choices and options will go away and we can go back to the nice, simple, limited web, but that doesn’t seem very likely just yet.) 

One of the big changes here is the removal of monopolies. If the web is not the only interaction model then web search loses power as a discovery and acquisition channel. And in parallel Facebook’s desktop monopoly on social has not transferred to mobile and it seemly unlikely that it ever will (I wrote about the reasons for this here). So both of the key channels on the desktop are smaller and less crucial, and also work significantly differently, and are pretty poor at driving some key types of engagement, now that you’re not just looking for a click on a link. This changes lots of things, and creates lots of new opportunities.  

The puzzle for Google is how it brings its vast, decade-old machine learning project to bear on this new complexity of data and behaviour. The obvious problem is that data and behaviour within apps are effectively dark matter that it can’t track (hence the deep-linking initiative in Android). But this is balanced by much richer data collection. Your Android phone feeds it with data all the time - where you are, what you look at, where you go after you search and what you did the day before. The challenge is finding the right ways to collate and present that data - Google Now is one example but probably not the only one. The search box and the page of results is just one possible interface to that machine-learning project - what does Google look like after the search box?

Social faces a different set of challenges. It seems to me that on mobile Facebook will never have the near monopoly that it briefly enjoyed on the desktop - smartphones remove most of the frictional barriers that keep you on one social network. But mobile social more broadly is a vast opportunity. With web search no longer the dominant channel, social, on a far more social device than the PC, has an open door to push at. Tencent announced that the first 5 games that it launched with Wechat integration, starting in August, have had 576m registered users. Mobile social is an engagement, interaction and distribution channel, and it appears to be much richer, and probably much bigger, than social was on the desktop. 

If this is the end to near-monopolies in acquisition and discovery, it’s also interesting to think about it as the end to monocultures. If the interaction model shifts away from web search, that change makes different models and different types of behaviour possible. In turn, one might ask - what models and companies and behaviours were precluded by Google on the web? What good ideas didn’t work because of the way Google did search and the way Facebook did social? How did that monoculture shape things, and how does that change now?