Android taxonomies

For reference, and, perhaps, discussion: 'Android' means lots of different things, and there's a lot of confusion about forks, Xiaomi, China and AOSP, as well as 'the next billion'. So this is how I try to think about this. First, there are actually (at least) six types of 'Android' in the market today:

  1. 'Stock' Android, as seen on Google's Nexus devices, complete with Google services (but with tiny unit sales)
  2. 'Modified' Android, as seen on phones from Samsung, Sony, LG etc, complete with Google services - generally, these are modifications that no-one especially likes, but which Google explicitly allows 
  3. 'AOSP' or open Android, as seen in China - essentially these phones are the same as number 2, but with no Google services and apps from the Chinese portals embedded instead. Hence Samsung, Sony etc sell their phones in China without Google services, but few other changes
  4.  (or perhaps 3.1) 'Modified' Android as seen on Xiaomi phones and those of its followers, which people actually seek out, and which comes without Google services in China and with them elsewhere
  5. ROMs and third-party implementations of Android that are available for any handset, such as both Xiaomi's MIUI and Cyanogen (an a16z portfolio company), which may or may not have Google services included or accessible. Again, these contain optimisations and improvements that make people seek them out
  6. Forked Android, such as the Kindle Fire phone: Android heavily modified to produce a different experience, and Google refuses to allow Google services to run on them (other than plain old web search, AKA POWS). Note that Xiaomi and Cyanogen are not forks. 

Th first two or perhaps three I would describe as 'closed' Android and the second three are 'open' Android, certainly from the perspective of device manufacturers. The first two (actually just number 2) have over a billion users outside China (as of the numbers given at IO last summer). Versions 3 and 4 have a further 400-500m users, almost all in China, and there are perhaps 50m users of 5 ( a very rough estimate) both inside and outside China, partly overlapping with the others. Six - well, ask Amazon. 

In parallel, it's worth breaking down Android users in a similar way:

  1. ROM users (very roughly, perhaps 50m people)
  2. People who like to install the kinds of apps that do things Apple doesn't allow on iOS and Google does allow on Android (note that Apple now allows rather more things and Google does not, oddly, allow gambling apps). I had a go at quantifying this here
  3. People with a personal preference for Android, who none-the-less do not actually install ROMs or do many things that are blocked on iOS (the difference between this and 2 is a grey area, obviously)
  4. People who don't actually care very much one way or the other between Android and iOS, and (for example) got a good deal, preferred the handset design or (especially) the larger screen size that used only to be available on Android, and indeed might switch back and forth between iOS and Android 
  5. People who can't afford iPhones or other high-end phones and so got Android as the cheaper option. 
  6. People who actually don't care about smartphones at all, and so just bought a 'cheap phone' (or just a  phone with a good camera, say), and happened to get an Android since it's taken over most of the mid range and low-end, and who don't do much with the ecosystem
  7. People in emerging markets who really can't afford anything other than a $50 or $100 Android phone but are enthusiastically taking advantage of everything it can do.  
  8. As above, but have a relatively expensive data plan, limited 3G coverage and, often, limited access to power to charge their phone (this one is is where the 'next billion' will sit) 

 Some of these categories (but obviously not all) also apply to iOS, of course, but selling phones only at $600 for the latest model creates a more uniform customer base. 

Layered across both of these is huge geographic variation. The must-have phone for teenagers in San Francisco and Jakarta is very different. But the underlying point about both lists is that tech and mobile have grown far past the point that there is really a single market for anything. When you connect everybody you get, well, everybody, and they're not all like you. 

Why is Apple making a gold watch?

As expected, the gold version of Apple’s watch is very expensive by consumer technology standards - $10,000 and up, depending on the band you take. And, also as expected, this made a lot of people’s heads explode. 

There are really two different conversations here: will people buy a $10,000 Apple watch, and why did Apple make one? 

To the first question, Apple is clearly breaking lots of rules with the watch. There are plenty of $10k watches on the market already that sell just fine, but those are normally mechanical ones, and mechanical watches are sold primarily on the complexity of the mechanism, where of course the gold Apple watch is internally identical to the $350 aluminium model. Mechanical watches are also expected to last: no-one quite knows how long an Apple watch will last (the battery is replaceable, but is the screen? And how long will the software be viable?) but it’s probably not something that your grandchildren will own*.

On the other hand, there is no a priori reason why a watch should have to follow those rules. Plenty of other $10,000 luxury items are far more ephemeral, and once you're selling things for other than purely utilitarian reasons questions like 'value' and 'resale' miss the point. Apple is certainly trying something that’s new to both the tech industry and the luxury goods industry, but it's not necessarily outside the bounds of (rich) consumer behaviour. If we only ever bought things that had rational use cases and the best value, we'd all be wearing boiler suits, or hoodies. 

Ultimately, though, how many people buy the gold one is probably immaterial. The Swiss watch industry sells about half a million precious metal watches a year, and though the size of the overall watch market is not a good indicator of the market potential for smart watches, the size of the market for precious metal watches is probably not far off as an indicator for the gold Apple watch. Even a hundred thousand gold Apple watches at (say) $15k each would be ‘only’ $1.5bn a year, or less than one percent of Apple’s 2014 revenue.  

So (and this is the question that actually matters) why bother? One could argue that it’s a vanity project, or that Apple’s doing this just because it can, or that a few hundred million dollars still matters at Apple (as indeed it does). But I think it’s more interesting to compare it with Apple retail. Despite its prominence, this is only about 10% of Apple’s revenue. It’s much more important as marketing. And it's great marketing. 

Apple stores are huge rich-media billboards on every major shopping street in the developed world: I can't think of any other company that has shops as big as that in such premium locations in as many places. Apple retail is a self-funding marketing operation. So too, perhaps, is the gold watch. Apple might only sell a few tens of thousands, but what impression does it create around the $1,000 watch, or the $350 watch? After all, the luxury goods market is full of companies whose most visible products are extremely expensive, but whose revenue really comes from makeup, perfume and accessories. You sell the $50k (or more) couture dress (which may be worn once), but you also sell a lot of lipsticks with the brand halo (and if you think Apple’s margins are high, have a look at the gross margins on perfume). 

Meanwhile, though other companies are already making metal smart watches, I struggle to imagine Samsung making solid gold watches. Apple's brand might or might not work there, but no other CE company's does. That is, if this is marketing, and if it works, it's marketing that no-one else can do. 

On another tack, perhaps the biggest message that this sends is that the Apple watch is not a technology product. It’s a post-‘feeds and speeds’ product. Today we have prices and release dates for the watch but no tech specs at all - because they’re irrelevant to the user experience. This is a product sold on delight, and experience, and on the feel and pleasure of owning and wearing it and looking at it (which of course means Apple retail is a huge advantage). It’s sold on a butterfly, not on the storage capacity. The value of the gold may be just that message - it’s not a geek’s product at all. One might call the gold a marketing detox - an emancipation of tech from the tech industry.

Finally, whatever your opinion of all this, it doesn’t really matter. Apple’s watch is, after all, coming to market at a lower entry price than any previous new category from Apple. So we can go out and buy it for $350 or $1,000 and get on with working out what, beyond delight, it’s good for - how it changes attribution, and user acquisition, and dwell time, and changes how you use your smartphone, and all the shifting metrics of the mobile internet. 


* There was some speculation that Apple might have a plan here - that it might buy back the gold watches, or replace the insides, perhaps - because a $10,000 object that's outdated is different to one that's $600. This may well still happen. 

Why do we care about Xiaomi?

'If you have nothing to tell us, but that on the banks of the Oxus and the Jaxartes, one barbarian has been succeeded by another barbarian, in what respect do you benefit the public?' - Voltaire

The Android smartphone business can feel like it's a rerun of the PC business, but compressed into 5 years instead of 20 or 30. The component layer is mostly a commodity, especially below the high end, and so is the operating system layer, and manufacturers are stuck in the middle, all of them using the same basic components and the same software (Windows for PCs, Android for phones), and so unable to differentiate on much beyond than price. The result is a race to the bottom with distribution, marketing and manufacturing scale the bases of competition. 

Today, Samsung's dominance of Android (with close to half of unit sales) appears to have peaked and a wave of companies appear to be able to use the entire Shenzhen manufacturing cluster or ecosystem to achieve many of the benefits of scale without having vast factories themselves. We see this in Chinese companies such as Xiaomi, OnePlus or Gionee, and also in local companies around the world that are having their phones made in China and basing their business on branding and distribution - some examples are Micromax in India, Cherry in the Philippines, Blu in Latin America or Wiko in France, which claims over 10% market share. Many of these are simply picking standard models from the big contractors in China and adding a brand - hence one big OEM told me 'when those guys say they designed it, they mean they went to Foxconn and asked for models 4, 23 and 39'. It's not clear quite how big you can get on this model, nor how like the PC clone business (i.e. eventually consolidated) it will end up. Equally importantly, it's far from clear how global these companies will be, but that may not matter. 

In this environment, just as in the PC business, some companies and some business models do better than others, but that doesn't necessarily matter to anyone else further up the stack. Gateway 2000 disappeared and Dell prospered, but that didn't much matter to PC buyers, let alone software developers or anyone on the web. Equally, Samsung rose to dominance and now slips, Lenovo buys Motorola and invests in mobile, HTC faltered and Sony goes sideways and Chinese OEMs are rising, but the phones keep coming.

(This issue is also one of the problems with Mobile World Congress in Barcelona, from which I've just returned. When I started going to this conference, in 2000, operators and handset makers drove the agenda for the whole consumer experience and the show was the place to see that, but today that agenda is driven at Google IO, Apple's WWDC and, perhaps, Facebook's F8.) 

So these companies, XIaomi included, are interesting to people who are interested in the handset business, then, but why do they matter to anyone else? Xiaomi sold 75m phones last year, yes, but isn't it just another OEM with a new branding and distribution model, and its rise relative to Samsung, say, no more interesting to anyone not in the handset business than the rise of Dell relative to IBM's PC division was? And it's mostly in China, for now  - isn't it just one 'barbarian' defeating another in a far away place

Well, up to a point. Though it's certainly possible to get over-excited about Xiaomi, this sort of dismissal misses several important threads worth pulling on. 

First, PC OEMs mostly failed to differentiate in design, and this has also been true of the Android market (and especially of the biggest OEMs), but the Chinese appear to be learning design, and quickly. Xiaomi, Gionee and several other companies are making phones that are visually appealing, using materials and finishes to differentiate from the run of black plastic rectangles that predominate in Android below the high-end (there are several consultancies composed of former Nokia people serving this market). A couple of these phones look superficially similar to Apple products - most do not. They're nice-looking, interesting, and half the price. (Interestingly, the way this has played out in China so far is that Samsung's share has fallen fast but Apple's has grown: they're selling to different customers.) 

Second, PC OEMs never managed to create any meaningful differentiation in software, and neither again did the Android OEMs, but Xiaomi has, and so have some of its imitators. The difference is that unlike most previous OEM attempts at software, they are not trying to compete with the whole internet. Instead of relying on vertical point solutions (photo editing and music services etc, though it does have some of these too), Xiaomi builds a thin horizontal layer between the commodity AOSP OS and third party applications. It has rebuilt the Android front-end - the whole UI from the launcher and notification panel down to the preference panels - to make it simpler and cleaner, changing the experience of using Android. This sometimes looks superficially rather similar to iOS 7, but this is a little misleading, which is hard to get from screenshots: a big part of the uniqueness is the animation. A Xiaomi phone feels very different from an iPhone or 'stock' Android phone, across the whole experience.

In addition, Xiaomi has packaged together a suite of integrated OS-level services that remix what Google and Apple provide in a similarly integrated way. In this it is of course helped by the fact that those Google services are mostly unavailable inside China and Apple's require you to spend 2-3x more on a phone. People tend to talk a lot about Xiaomi's services as a way to sell the phones at lower prices, but we know from various leaks that the revenue from these is actually immaterial, at  least for now - the point again is the experience that they give. 

It's a common reflex in the USA to call Xiaomi and similar companies copycats, on both a hardware and software level, and there's certainly a lot of design inspiration going on, but dismissing these phones as rip-offs is rather lazy. When you actually hold them and use them you would never mistake them for iPhones, any more than you'd mistake a $500 coat or bag for the $2,000 example that sometimes inspires the colour or trim. You can sometimes see where it came from, but it's not the same, nor is it trying to be. Rather, they deliver a similar approach to the user experience at a lower price in a way that's often been absent from Android so far (it's hard not to look at some of Nokia's beautiful Lumia devices and wonder what impact they might have had on Apple if they'd run Android).   

The result of this is that we now appear to have at least a couple of Chinese companies doing what was supposed not to be possible - low-margin companies using commodity components and a commodity OS, yet achieving differentiation in design, software and services. Looking at China always challenges your assumptions about what's inevitable in technology. Hardware companies doing good software and UI? Commodity box-shifters learning design? How far might that spread? 

Finally, this shift in what handsets might look like also has a broader implication: the spread of new types of Android OEM might change Google's control of Android. 

Historically, Google's lock on Android outside China has therefore been based on three things: 

  • You can't experiment outside very tight constraints: making even one forked device means Google won't allow you to sell a single phone running Google services. And all the OEMs have too much to lose to risk experimenting
  • There's a widespread belief that an Android device without Google services (really, this means Maps and the app store) is unsaleable outside China (I'm not entirely sure about this, as I wrote here)
  • No OEM managed to build a compelling set of services or tools of its own that might offer alternatives to Google, because, well, that was impossible (see above)

These new trends place all of those in question. The growth of smaller operators pursuing different models, with no existing base of sales and hence nothing to fear from  Google ban, may mean more experiments with forks. Xiaomi and its imitators point to a new potential model to differentiate (and note that Xiaomi is not a fork), and Cyanogen (an a16z portfolio company) offers the tools to do it. Smaller OEMs are less powerful than Samsung as a counterpart to Google, but also harder collectively to impose upon - Google can't shout at them all. This isn't to say that I necessarily expect to there to be lots of local attempts at the Kindle Fire, but we may start seeing a lot less uniformity in how Android comes to market, and what it looks like. 


Notes on cars

A circular building

A circular building

Apple we are told, is working on cars, and there's enough smoke that some fire somewhere seems likely. Apple has enough cash (over $150bn) to do this, if it wants, and this prompts all sorts of investing questions, but I've been wondering how one should think about the market opportunity it might be able to secure, and how that fits into the other incursions of the tech industry into cars. 

First, can Apple create new value in the industry in the way that it did in phones?  With the iPhone, Apple created a new price segment and (with Android following) made the phone industry's revenue much bigger - the average price of a phone sold has more than doubled since 2007. But cars are, pretty obviously, more expensive than phones. Many people can find $400 for a better phone or, this year, a smart watch, if they're persuaded that they really want one, but rather fewer can find an extra $40,000 for a better car, or to replace their car every two years instead of every 4 or 8.  If you're in the market for a $20,000 car, there is very little that anyone can do to a car that will put you in the market for a $60,000 car. Cars do not come out of discretionary spending.

That is, lots of people never thought they'd spend the extra to get an iPhone or one of its imitators, but they did, and it wasn't actually that much money. A billionaire and a teenager have the same phone. Conversely Jonny Ive could invent a car that flies and makes perfect espresso, but if it costs $60k or $80k then people driving second-hand Corollas aren't going to buy one. 

In addition, Apple created a premium segment in phones, but there's already one in cars - Apple could take share of that, but it's ipso facto too late to create it. 

 So, it seems, at the very least, much harder to increase the overall size of the market than it was for phones.  This isn't necessarily a problem. The major premium brands BMW, Lexus, Audi and Mercedes Benz sold 5.5m units in 2013, and had revenue of around $220bn. (The total market was around 65m cars, with a further 22m commercial vehicles). Taking a share of $220bn a year, even without changing the overall market size at all, would be just fine.

Meanwhile, there's certainly scope to change the product and take market share - indeed, we are now rethinking what a car is for the first time in generations. Electricity leads to different manufacturing economics, allowing (potentially) lower costs and lower maintenance costs. It also allows the car to be reconfigured, at least to some extent, though obviously less than a self-driving car with no manual controls at all. And the experience of driving the car itself about adding more and more software - one wonders what one would do with a dashboard if one started from zero as a software company? It seems that there might be quite a lot that could be done to reinvent the experience. And though we should assume Apple will retain a premium experience, we can make too many assumptions about the price - remember that the iPad was certain to cost $1000.  

However, there are other reasons besides electricity for the reinvention of the car - the rise of on-demand and the possibility of self-driving cars. These do have the potential to change the size of the market, but by making it smaller, not bigger. 

Both on-demand and self-driving cars would appear to drive a reduction in car ownership and certainly car use (which means slower replacement), to the extent that they become a major part of the urban landscape. That obviously means fewer car sales. They also change what cars get bought. If you don't own the car yourself, and don't even see it before it arrives, the brand and styling matter less than efficiency. That effect is probably strengthened if we move to a fleet model (as many taxi systems work) rather than owner-operators - a fleet manager will choose the vehicle based on metrics, not the fit and finish. That is, the car market would be both smaller and might look more like the corporate PC market. It might also start to bifurcate - people buying $15k and $30k cars substitute $20-25k on-demand vehicles, while the high-end is less affected. Or, high-end sales might be affected most, if those people are best able to afford going entirely on-demand. We don't know, but there are lots of moving parts and will be many unanticipated consequences. Who looked at the Model T Ford and predicted Wal-Mart?

On the other hand, self-driving cars might support both an on-demand model and an AirBnB model for cars - does your car drop you off at work and then roll off into the city to earn you some extra money driving other people around? Would people want to do that? Would that reduce the opportunity for 'dedicated' on-demand vehicles? Who knows. Of course, it's also possible that self-driving technology, said to be a decade away now, will remain a decade away indefinitely, as so many other AI projects have done. 

In a sense, all of this is the unbundling of public transport. Instead of large vehicles aggregating passengers on fixed routes, you have many small vehicles, with many different ownerships, on almost infinite routes - packet switching instead of circuit switching, if you like. 

The challenge for Apple and anyone trying to make premium cars in all of these questions is that they are matters of AI and routing and algorithms - they are matters for Google, not Apple. They shift the value away from the hardware to the cloud, and turn the car into a generic commodity - dumb wheels instead of dumb glass.  Apple doesn't really do algorithms (though it does do privacy, which may become a lot more relevant). And meanwhile a shift to self-driving and on-demand is focused precisely on the urban 18-35s who are its best customers. 

There's a counter-argument to all of this, of course, that the correct place for intelligence is in the device you hold in your hand, take everywhere with you and replace every two year, not the large piece of moving metal that you replace every 5 or 10 years. Cars are for car makers, and though Apple could make a nice one with all its design skills and capital, it could also make a nice retail bank, or chain of restaurants. This is especially the case without self-driving. On this view, the car should be dumb glass, with all the intelligence in the smartphone.

This of course is the real problem, which I talked about here - forecasting how the tech will evolve is often easier than how it will be used. Agatha Christie supposedly said that when she was young, she could not imagine being rich enough to have a car or poor enough to have no servants. The ways that both would change in the next half-century were totally opaque. Now we're reinventing the car again, but it's much easier to see what might be possible than what people would choose. 

Ways to think about market size.

When you try to work out the market potential for something fundamentally new, you’re actually trying to resolve two, linked problems.

  • First, you have to look past what it is now, and see how much better and cheaper it might become
  • Second, you need to think about who would buy it now, and who else would buy it once it is better and cheaper, and how it might be used. 

The second problem is actually the hard one. Anyone with a sense of history ought to have been able to look at a phone the size of a brick and say ‘well, this could come down to the size of a pack of cards and cost $100, given time”, just as anyone should have been able to look at the Krieger electric landaulet above and see that it would get much better and much cheaper, just as trains and steamships had done. If you understood technology, that much was pretty easy. But if automobiles had only replaced existing horse-drawn carriages and carts then the market would have been much smaller. The hard part was to forecast Wal-Mart, and Los Angeles. 

That is, it’s easier to predict 'cheaper and better' than to predict the changes in behavior that will come from that. And pricing is only one dynamic - once the price falls below a certain level it stops mattering. Cheaper and better is necessary but insufficient: if billions of people can afford it, it doesn’t follow that billions of people will buy it. You need to have a theory as to why more and more people will care.

Hence, if I’d shown you a 2015 PC in 1975, would you have predicted that there’d be 1.5bn of them on earth 40 years later? Why? If I’d shown you an iPhone or Android smartphone in 1995, would you have predicted that we’d now be on track to have 4bn of them on earth - fourth fifths of all the adults on earth? 

So, to work out market size, really, you have to work out who will care, if it is cheap. To do that, as for any estimation, you look for numbers that might tell you - other, similar products, or the products you'll compete with - something that can act for a proxy for how people might look at what you're making. There's a sliding scale here, I think: there are some markets where you have a lot of data, and others where, really, you're just guessing. 

First, at one end of the scale, there are those people who are entering an existing, fairly mature market, with a superior product or price, expecting to take market share. In that case you already know how the market size works - you know why and how people use these things. For example, the US market for, say, refrigerators is xm units a year, with ym homes having them and replacing them every z years. Prices are low enough that every home already has one and the product lasts for a decade or more, so you only change them when you move or rework your kitchen. So annual sales in the overall market (for the sake of argument) are outside your control, but you can take share. You can get people to buy yours, but not to buy more than they did before, so the question is how much market share you can take with a better operating model.

Second, at the other end of the scale, there are companies that are creating something entirely new. The personal computer was an example: imagine trying to forecast this in 1980. You know what typewriter sales are, you know how many middle class households there are and you can assume that only corporations and middle-class households will be able to afford one for the next few decades. But you don’t know about the internet as the key driver for consumer PC adoption,  and you don’t know how many office typewriters will become PCs, nor that typing pools will disappear and every executive will write his own emails instead of dictating letters to his PA. 

The same problem applied to mobile phones. You could do a bottom-up analysis that counted business travelers, taxi-drivers, fleet dispatch and so on, and get to maybe 10-15% of the population. Lots of people did that in the 1990s. They were all wrong. For phones, as for PCs, you had to make an imaginative leap into the unknown. You had to say ‘I believe’ that this experience will be transformative, and everyone on earth who has the money will get one. Moore’s Law takes care of ‘having the money’ meaning 4-5bn people, but it's the imagination that gets you to teenage girls living in text messages. You could predict that phones might get really cheap, but not what that might mean. 

In that light it’s worth comparing these two mobile phone ads from the early days of the industry in the UK. The first, perfectly rationally,  starts from the mentality ‘how many people will need this?’ This is the '10-15%' argument. The second, from Orange, assumes that everyone will want one and it’s our job to get it to them, because we're changing the world. Phones don't have specific use cases - they're a universal product. Hence, the CEO at the time, Hans Snook, went around saying that the UK would go to 150% penetration and most people thought he was mad (note that the Cellnet ad was made two years later). 

This is the problem with forecasting sales of the Apple Watch. Annual watch sales are a bit over a billion units, and people buy watches at anything from $5 (China exported over 600m watches last year at an average wholesale price of $3) to $500, $5000 and $50,000. But this doesn’t tell us anything useful. The fact that you buy a $10 watch, or a $1,000 or $10,000 watch, or buy no watch at all, tells me nothing about whether an entirely new product that you also wear on your wrist would be appealing. The fact that you bought a watch x years ago and the average replacement rate for watches is y tells me nothing about whether you’d replace it with an Apple watch, tomorrow, if you saw one. 

That is, there are, in principle, hundreds of millions of people available to be persuaded to buy a smart watch, but we cannot draw any firm conclusions about how many will do so from looking at the existing watch market. That's like looking at the typewriter market to forecast PCs. It might be more helpful, perhaps, to look at the broader luxury goods market (how many women buy how many $500-$1,000 handbags each year?), or the camera market before smartphones killed it, or the phone case market. One can look at the high-end segment of the phone case as a proxy. One can triangulate one's guesses. But we're really just going to have to wait and see. We have no data for how many people will find a place for this in their lives, just as, 20 years ago, we had no data to support the idea that almost everyone would find a place for a mobile phone.

Third, you have companies that sit somewhere in the middle - companies that are entering a market in which the top line dynamics are mostly fixed  but there remains plenty of scope to change things. This is where the iPhone and Android came in. The global mobile phone market has somewhere between 3.5bn and 4bn users, growing steadily as a function of macroeconomics and increasing distribution. Apple and Google didn’t change that - they couldn’t. By reinventing what a phone was, Apple did not change how many people bought a phone, or even (really) how often, but it did change what they paid. It converted $200 phone sales to $500 iPhone sales (only partly helped by operator subsidies), and Android followed at lower prices, such that together they now make up about 70% of unit sales.  As a result, the average selling price of a mobile phone more than doubled from 2007 to 2014, from $80 to around $185. 

Everyone likes to quotes the Wayne Gretzky line that he was skating to where the puck was going to be, not where it was, but Apple and Google didn't do that - they changed what the game was. In the same way, saying that you’re aiming for x% of a $ybn industry is unambitious - great companies change the y, not the x. It'll be interesting to see what, if anything, Apple is planning in cars. I'm not entirely sure there is quite the same scope for changing the market size that there was in phones. Watches on the other hand, are wide open. 

App store revenue, and selling to the world.

Last summer, at their developer events, Apple and Google told us that the previous 12m months, Apple paid $10bn to developers and Google $5bn. Now we learn that Google's up to a $7bn run-rate. In December, Apple repeated the number - that might mean flat growth, or (I think more likely) that they were just repeating the same number. We don't (yet) know where Apple's grown to since then (hint, hint). 

If we gross up this new Google number and the most recent Apple number, incidentally, we get to close to $25bn in annual consumer spending on apps.

Google also told us last summer that it had 'over 1bn' users of its version of Android (i.e. not counting China, with China also not counted in the revenue numbers). Since at that time my model (and others) estimated that Apple had about 625-650m live iOS devices, that meant that the average iOS device was generating 3x to 4x more app store revenue than the average Android device ($10bn and 625m users versus $5bn and 1bn). We cannot tell from these numbers if that's changed since then. On one hand, Google has improved monetisation, but on the other, all the new users since then will be on lower incomes with lower propensity to spend, and we know neither the Android user base nor the iOS revenue. 

The other point that Sundar makes is that Android is selling a massive range of devices to a massive range of different people, whereas iOS is selling to people who can buy $600 phones. This is true, and it's inherent in the different models, and both are great models. This is why I've said that both Apple and Google have both won the platform wars, for now. Google got reach and Apple got device profits (which Google doesn't care about) and an ecosystem with sustainable scale. These numbers make it abundantly clear why iOS is not going to lose access to the best apps. But it's also why Android ARPU is lower (and of course this difference is seen in ecommerce spending and traffic as well, not just spending on apps). If you sell to everyone, well, the average user is going to be different to the average of people who buy $600 devices. Again, that's fine, and it's inherent in the model. 

One final observation, too: both Google and Apple have smartphone platforms and ecosystems with massive global scale and stable, sustainable dynamics (though the dynamics of the Android OEM space may be another question). That means that working out the horse race between the two is now pretty irrelevant. Again, they both won, so what's next

In mobile, disruption comes from above

The classical description of disruption in business, and especially technology, is that a new product (method, business model etc) appears that’s not as good as the existing way of doing things, but that’s much cheaper. The existing industry thinks it’s a joke, and certainly not a threat. But over time, it gets better while staying cheaper, and then, sooner or later, it’s not a joke at all. 

You can see this basic story over and over again in the history of the technology industry. The future always comes looking like a toy. But right now the tech industry is being reset by the mobile, and in mobile, disruption tends to work the other way around. The new thing tends to arrive looking like an expensive luxury for rich people, doing far more than any normal person would need. But over time it gets cheaper, and the new, unnecessary characteristics turn out to be very necessary, and the the old, cheaper, less capable model gets squashed. 

That is, in tech the cheap weak product generally gets better quicker than the good expensive product gets cheaper. But in mobile, the good expensive product has generally got cheaper faster than the cheap, weak product got good. Moore’s Law is operating in both cases, but the effects are different. 

As for all theories, there are exceptions and gaps (‘all theories are wrong but many are useful’), but it’s worth looking at those cases where it does hold true. 

The really obvious example is mobile itself. Twenty years ago cellular was an expensive luxury for millionaires and drug dealers - status symbols no normal, sensible person would ever need. You have a telephone already - who needed mobile? But once the devices and networks reached a minimum threshold the appeal of mobility was much greater than the appeal of a landline's price. (Moreover, price turned out to be more nuanced - a prepay phone can look cheaper or more expensive than a landline depending on your point of view.)

At the same time, there was a lot of debate about quite what kind of mobile network was best. Anticipating the ‘wifi is good enough’ argument of the early 2000s, there were several attempts to offer cheaper ‘limited mobility’ services that would only work if you were in a specific place. You can see an (appallingly bad) ad below for one of these in the UK, Rabbit: the PHS networks in Japan were the only, Galapogean survivor of this. It turned out that mobile phones need to be, well, mobile. (An old colleague of mine suggested that the root of all the problems in the USA was calling them ‘cell phones’ instead of the British ‘mobile phones’, but then the Germans call them ‘handys’ so who knows.)

Now, did you notice that little message at the beginning saying 'outgoing calls only'? I'd forgotten that part. 

The contemporary counter to this, of course, is Iridium, which was too expensive and ‘too good’, with global coverage out of the box overshooting customer needs relative to cellular. However, though Iridium could give you a signal in the middle of the ocean or desert, it couldn’t give you a signal inside a car or an office, needing line of sight to a satellite, and the phones needed their own porter, so I’d suggest that actually, Iridium was more expensive and worse than cellular, not better.

Exactly the same thing then happened with the idea that wifi would threaten cellular - 3G got cheap enough and fast enough that ‘free’ wifi data for your phone was irrelevant, while wifi coverage never matched even the ‘good enough’ target of ‘most of a city’. Better and more expensive beat cheaper and good enough.

Arguably, you can see the same thing happening again with projects like Firefox OS. Entry-level Android phones are now well under $50 - the price window between ‘only a feature phone’ and ‘I can’t afford Android but want more’ is moving down fast. It’s tough to compete with the scale effects of the whole Shenzhen ecosystem. Again, the better, expensive product gets cheaper. 

Most recently, of course, the iPhone came in at a very, very high price in the context of the phone market in 2007 (even after Apple realized it needed subsidies after all). It’s certainly valid to say that it disrupted PCs from below, but it did, actually, disrupt mobile phones as well - just ask Nokia, what’s left of it. The new paradigm was a large phone with a multi-touch screen, ‘PC-class’ OS, relative indifference to bandwidth efficiency, target battery life of a day instead of a week, and durability targeted as ‘well, don’t drop it then’. It was also an MVP (no 3G, basic camera, etc, etc). For all these reasons and more the industry laughed at it, but Apple created demand for a $600 phone on a scale that had never existed before and its cousin Android then drove the same model down to much lower prices. The Symbian model, and the feature phone model, didn’t grow to meet the new, expensive challenge - the expensive challenger model took the top of the market at a new, higher price point, and then (incarnated as Android) got cheaper and took the rest. 

The fact that multitouch smartphones make different trade offs to feature phones around durability and battery life does of course blur the question of whether it is 'better' or just different. But I'd suggest 'better' is subjective: what is not in doubt is that the more expensive approach has beaten the cheaper one. 

There are a bunch of different subsets to this story, of course. Another counter-example, stronger than Iridium, is that within the mobile operator industry each country generally has some operators with strong networks (good coverage, high speeds and capacity) and others with weaker networks - the stronger networks generally charge a premium, and some consumers choose to pay the extra and some decide the cheaper one is good enough. Most recently, this is what Iliad/Free is doing in France (helped by heavy regulatory support) and Deutsche Telekom's subsidiary T-Mobile is trying in the USA. But I’m not sure that counts as ‘disruption’ so much as ‘cheaper competition’, which is where I’d put Android as well: a similar product at a lower price, not a different product serving the same needs at a different price.  

I don’t propose a perfect, unified theory to explain why mobile seems so often to work like this, but that's less important than the observation, I think (or, perhaps, I’m just suspicious of unified theories). The tech industry is very used to the idea that incumbents always laugh at disruption because though it’s cheap, it’s also crap, not realizing that history tells us it will get better. But we also have to remember not to laugh at things that are amazing but way too expensive, because history tells us they might get better and cheaper, faster than you can add 'amazing' and put it through a waterfall. 

Podcast: Slack, messaging and institutional memory

You've heard the story: Slack began as a game. But almost exactly 1 year ago today, the internal tool the team built for its own use became a team communication app that anyone (and especially enterprises) can use -- and is now one of the fastest growing ones at that.

It seems like collaboration is "something software should be helping us with” Slack co-founder Stewart Butterfield observes, yet it typically isn't. So what can an app like Slack tell us about how we work today, and how the nature of work will change (fewer meetings? less emails)?

Butterfield is joined in this edition of the a16z podcast by a16z board partner Steven Sinofsky and a16z's Benedict Evans. The trio examines the origins of messaging and task management tools (many of which Sinofsky worked on at Microsoft) -- and how the advent of cloud-based services and mobile in particular have changed the requirements for modern workplace tools and information management.