Google search

Google Android – Closed Source

I think Google may make Android proprietary in 2017.

Google has launched Android N but, without the ability to distribute updates, the software is virtually useless. To make matters worse, I think Google is effectively doing research and development where its competitors benefit more than it does.

Android M (6.0) is currently on just 18.7% of Google’s Android devices, despite having been available for almost a year. That corresponds to the penetration one would expect with virtually no updates being made. In contrast, iOS 9 is available on well over 90% of all devices and is about to be quickly replaced with iOS 10. This is a massive problem because it means any innovations Google makes to Android to compete against iOS, Windows or China will take four years to fully penetrate its user base. In my opinion, this renders the innovation worse than useless as it will be fully visible to the competition who can copy it and get it to market long before Google can.

A great example of this is Now on Tap which allows context based search from anywhere on the device. I have long believed this is a stroke of genius as Google currently only has 41% coverage of the Digital Life pie but this feature allows Google to collect data as if it owned 100%. The net result will be greater understanding of its users and better targeting of its advertisements — meaning higher prices, driving revenues and better margins.

Unfortunately, this service requires low level changes to be made in the Android Open Source Package (AOSP) so the device has to have version 6.0 (Marshmallow)or later in order for this service to work. Currently, only 19% of Google’s ecosystem users on Android have access to this feature despite it being available for a year. This, combined with the endemic fragmentation that hobbles the user experience relative to iOS, is a major reason why I think Google services in Android devices generate 50% of the revenue they do on iOS.

The only way I think Google can fix this problem is to take complete control of Android, culminating in the migration of the Android Run Time (ART) from the Android Open Source Package (AOSP) into Google’s own proprietary Google Mobile Services (GMS). This would render the open source piece of Android to being just a kernel with the real functionality of the device being controlled from within GMS which is fully under Google’s control. This would fix both the software distribution problem and the endemic fragmentation but could or would probably result in an outcry from the open source community as well as attract more scrutiny from regulators.

Google has long been an advocate for open source software and the backlash it would receive from developers when/if it moves in this direction would be severe. However, the recent loss in its war with Oracle has given Google the perfect excuse to close down its version of Android and blame Oracle when developers complain. In Marshmallow, Google has been forced to use Oracle standard libraries for the Android Run Time — Google has effectively lost control of the software roadmap for the runtime. This is something Google simply cannot afford and, when it presents its proprietary version, it can point the finger at Oracle as the reason for having to make the move. Furthermore, the AOSP will remain in open source but its relevance will have been reduced to being just a kernel rather than a fully-fledged OS.

There are already signs of this beginning to happen as Google is set to launch hardware which it has “put more thought into” and become “more opinionated about”. The Nexus line of devices is set to disappear to be replaced with devices designed by and branded with Google with the manufacturer (HTC) being completely absent.

For the creators of Android forks such as Alibaba, Xiaomi, Tencent, Cyanogen, and so on, this means they will also be forced down the same road, resulting in a series of proprietary operating systems all based on a common kernel. For developers, this will make their lives generally easier as developing apps for Google Android devices will become much easier but more work will be required to also develop for others. I suspect this will force makers of Google Android devices to take Google’s software as they have almost nowhere to go until the EU decides to step in and force Google to change its practices around how it licenses the right to put Google Play on devices.

This has been a gradual process as the scope of GMS has been increasing for the last three years but, at Google I/O 2017, I think this move will become much more visible. I think Google has very little choice because, at the end of the day, its fortunes will be driven by revenue growth from Android as iOS is grinding to a halt. I think a bit of developer anger is better than a $100bn loss in value.

Published by

Richard Windsor

Radio Free Mobile is an independent research producer specialising in the digital and mobile ecosystem. During Richard’s 11 years at Nomura Securities, Richard focused on the equity coverage of the Global Technology sector in Equities. During that time, Richard began looking at Handset software and over a period of 5 years became an industry leader in the space. As a result he was regularly consulted by small companies, start-ups, VCs and the like for strategic input and views on where both the mobile phone industry and the handset software space were headed. Richard has always strived to maintain an independent and objective view and would not pull any punches when concerns arose about the viability of a company or an idea. Radio Free Mobile is the land of the one man band. This research is created, written, owned, operated and maintained by Dr Richard Windsor PhD CFA. Dr W. also makes the tea.

46 thoughts on “Google Android – Closed Source”

  1. Having multiple versions of their OS in use is a huge burden for third parties that create accessories and apps that have to support, not only different manufacturers, but different software versions of Android on each device. But contrast that with developing products for iOS devices that have just a few models and perhaps just two versions of the OS in use.

      1. But closing it down doesn’t change much w/o re-architecting for modularity, and re-architecting makes closing it down fairly pointless since it means the OS can evolve independently of all the rest.

        For example on the hardware side, taking OEM OS tweaks out of the equation, you still have to deal with, say, 30 Socs, 10 screen sizes, 15 screen definitions, 20 front and 20 back cameras, and 10 each of digitizers, touchIDs, compass, battery and sensors… that 30x10x15x20x20x10x… = tens of thousands of different configs to support. And that’s pretty much x the number of carriers. Update problem not solved.

        Plus on the flip side, no OEM OS tweaks would mean no KNOX, no multi-tasking/windowing in 2011, not Android laptops (requires OS tweaks), no Android desktops (ditto), no Android TV sticks (ditto), gaming consoles (ditto) no touchID in 2010, no eyeID in 2016, no gamepad support years before native support, … Not happening.

        1. Two things.
          1) I think you will see Google being far more prescriptive when it comes to hardware starting with the launches next week.

          2) When has KNOX, multitasking ever generated any differentiation for the handset makers?

          The economics of Andrrois are dire and only seem to be getting worse

          1. 1) OK, we’ll see. I’m not sure why though, next week is the launch of Google’s own hardware, nothing to do with OEM guidelines. But at least we’ll have a quick answer on that one.

            2) ??? Are you serious ? When have exclusive features generated differentiation ? When haven’t they ? Most of the Android articles on this site are of the “must differentiate” category (I’m more of the “must execute” church), but now you’re preaching “must homogenize” ? Granted, a majority of users probably don’t notice, but all the new features on Android (and iOS !) first popped up as OEM tweaks. Taking that away would take most of the vitality and innovation of the mobile hardware market. Plus where would Apple get ideas ;-p

            The economics of Android are wonderful for Google who is getting a few billions Internet users at a cost of certainly less than 1k employees (was 250 in 2010).
            For OEMs, the economics are dire… unless you compare them to any other Mobile OS except iOS. I know, Apple is very successful, which means everyone else is a failure. But Android OEMs sit rather pretty compared to Nokia, RIM, Palm…. most of them are profitable too, though not western PC companies, which seems to taint analysts’ perceptions.

          2. 2) yes I am serious. ask yourself what differentiation means. It doesnt mean that the product is different, it means that those who buy it will notice a difference and most importantly be willing to pay for it. If any of the android makers (except samsung special case) had suceeded in this regard you would have seen it in their profit margins… No joy so far. to me that means no differentiation.

          3. I’m always a bit amazed at the added “must be successful” (or “profitable”) criteria tacked on to various concepts: innovation, now differentiation.
            That makes those things kind of self-fulfilling requirements: to succeed, you must innovate and differentiate, but if you do and aren’t successful, then your innovations and differentiations… don’t count ? To me, you can have successful/profitable differentiation, and unsuccessful/unprofitable differentiation. Bundling the two concepts confuses the issue and confuses product and finance.

            To differentiate is to do something different. I’ll give you “and to make people aware of it” (because we’re talking business and marketing; in a design context, I wouldn’t). But not “and to be successful with it”. You can have failed differentiation that is differentiation nonetheless. Customers understand and perceive your difference, they just don’t want it / don’t see enough if at all value in it. Examples of differentiated failures: the Yotaphone, the HP Touchpad and phones, the BB PlayBook, the MS Surface with Windows RT, the LG Prada, Nokia’s N-Gage…

            Also, the profit criteria seems a bit random. Why not market share or sales ? Why not influence/legacy/awards ?

          4. simple. if you the purpose of innovation is to create a products with a high gross margin in order to earn a return on the R&D invested. If you are unable to do that then your R&D has been wasted. at the end of the day these products are designed t make money. No cash flow = going bankrupt = failure.

          5. If you bake a cake, and it turns out not very good; or it turns out OK but people don’t like it anyway (looking at you, nephews !!)… does it mean you didn’t bake a cake ?

          6. In the context of product differentiation that is exactly right, if you fail it means there was no product differentiation. The changes themselves are not product differentiation (I could add a wind turbine to the top of a car, and while that is different it would likely not be product differentiation). You must also create a competitive advantage that is real in the marketplace, the consumers must value the product differentiation (vote with their wallets so to speak). If they do not, then you have not created product differentiation. Hence the emphasis on sales and profit in relation to product differentiation.

          7. If you baked the cake to sell it for a profit because it had cool icing but no one valued it and you had to sell it at a discount it then indeed you failed to differentiate your cake.

          8. what’s the… difference… between “differentiated” (ouch !) and “successful” ?

            To me, you can be differentiated, but unsuccessful: say you messed up your target market and it’s too small, too price-sensitive, plain inexistent. Or you differentiated on features no-one cares about, or a set of features that work at cross-purposes (say, a phone that’s a) rugged for the outdoors and b) w/ an innovative screen that’s color-perfect but unreadable in sunlight and unoperable w/gloves). Isn’t that still differentiation, but failed differentiation ? I mean I know MS makes Windows phones, and those are utterly different. Nobody wants one; they’ sell more bog standard Androids just on the strength of the Nokia brand.

            Reciprocally, can’t you be successful w/o product differentiation, just via lower production costs, better availability… ?

          9. “To me, you can be differentiated, but unsuccessful”

            No, you really can’t be differentiated and unsuccessful. Maybe what you’re having trouble with is equating ‘being different’ with ‘product differentiation’. That is wrong. Product differentiation is ‘being different’ + ‘buyers placing value on the difference’ (you create a competitive advantage). We could debate what threshold of success qualifies but there is no debate about what defines product differentiation, it requires more than just ‘being different’ in some way.

            “can’t you be successful w/o product differentiation, just via lower production costs, better availability”

            I think here you demonstrate once again your confusion, you seem to think ‘being different’ equals ‘product differentiation’. That is not the case. Lower production costs or better availability could very well be aspects of product differentiation, since it isn’t just about making the product physically different in some way.

          10. Ha Ha. How very true but that is what Universities and research institutes are for… not businesses..

          11. Actually he did it on his own, not as part of his day job. At the time he was a lowly patent clerk at the Swiss Patent Office. His work was completely unrelated.

            In fact, when he eventually won the Nobel as a result of his work, the prize money was given as a settlement to his former wife.

  2. I’m not seeing Google waver in their “open” commitment. From time to time we do see a feature that does require a new OS version, but mostly
    a) even core features piggyback on Google Market Services, not the OS itself, so OS version is irrelevant. That has been the case for Pay, Wear, Family…
    b) ever more Google apps/features/services are being taken out of the core OS and moved to the Playstore: Keyboard, Now Launcher, Play, Books/Music/Videos, Chrome+Webview, gMail, 2FA, Health, …

    So yes, once in a very long while there’s a feature that’s OS-dependent. Mostly though, people mistakenly project iOS’s situation to Android. iOS is much more monolithic, a lot of apps and features are utterly dependent on an iOS update but just require a modular update on Android. I’d say over 95% of Google-originated new stuff doesn’t require an OS update, just app/GMS updates.

    Plus Google has already solved the OS update for those who care: get a Nexus.

    And the current setup makes it really easy to add the Google stuff onto Android forks (download the market and PlayStore, then the rest of the Google stuff you want via the PlayStore), reeling more users in for Google, while letting OEMs do whatever (and giving them the illusion of empowerment) w/o any real consequences for users and Google’s business.

    1. Publically you are correct. Nor will Googe waver in its commitment to an open AOSP. However the AOSP will become just a kernal and therfore largely irrelevant when it comes to fixing the fragmentation and sdistribution issues.

    2. “So yes, once in a very long while there’s a feature that’s OS-dependent. Mostly though, people mistakenly project iOS’s situation to Android. iOS is much more monolithic, a lot of apps and features are utterly dependent on an iOS update but on Android the equivalent just requires a specific, partial, modular update to an app or to GMS. I’d say over 95% of Google-originated new stuff doesn’t require an OS update, only an app/GMS update.”

      So you keep saying in comments to many articles.

      Conveniently, you basically simultaneously downplay, a) the fact that Apple can update most devices very quickly, and b) the fact that Google can’t.

      According to you, Apple *needs* to be able to update the OS in order to deliver new features and improvements; apparently Google doesn’t. Again, very convenient.

      I would like to see more expert discussion on your opinions:

      For example, how much is Google’s approach (“genius” as it may be) in reaction to the “predicament” that they find themselves in (lack of control regarding carriers and OEMs, etc.), and how much is what they would have designed it that way if they could do it from scratch?

      For example, are the improvements Google can make and widely deliver through the PlayStore *really* equivalent to those in OS updates that Apple is regularly making available to the vast majority of its user base. Or, again, is this just your opinion?

    3. “So yes, once in a very long while there’s a feature that’s OS-dependent. Mostly though, people mistakenly project iOS’s situation to Android. iOS is much more monolithic, a lot of apps and features are utterly dependent on an iOS update but on Android the equivalent just requires a specific, partial, modular update to an app or to GMS. I’d say over 95% of Google-originated new stuff doesn’t require an OS update, only an app/GMS update.”

      You have said something similar in comments to many articles.

      Conveniently, you basically simultaneously downplay, a) the fact that Apple can update the OS on most devices very quickly, and b) the fact that Google can’t.

      And, apparently, Apple *needs* to be able to update the OS in order to deliver new features and improvements; while Google seemingly doesn’t. Again, very convenient.

      I would like to see more expert discussion on this view.

      For example, how much is Google’s approach (“genius” as it may be) in reaction to the “predicament” that they find themselves in (lack of control regarding carriers and OEMs, etc.), and how much is what they would have designed it that way if they could do it from scratch?

      For example, are the improvements Google can make and widely deliver through the PlayStore *really* equivalent to those in OS updates that Apple is regularly making available to the vast majority of its user base? Or, again, is this just your opinion?

        1. Sure, I get the chart: things are moving from left to right. I got that the last time.

          Again, it just seems a bit convenient that this movement to the right is considered so wonderful– since users don’t get updates to those things unless they do move right. And again, you haven’t shed any light on why it is so wonderful, just that it is happening.

          One still wonders how good a thing that is, technically — other than the fact that more Android users now get those things updated (since they can’t get core OS updates). One wonders, for example, what happens if there is some kind of update (say one that affects security) that should, or can only, occur in the core OS — most users go without it, apparently.

          Maybe Google is just doing an end run around OEMs and carriers in order to get control back — moving everything to the Store, so that in the end the phone ships with nothing at all except a connection to the store 😉

          1. 1- things don’t need to *move* right, most of them already *are* in the “GMS” or “PlayStore” columns. The “move” thing just means ever more app/features/tools are in these 2 “good” columns.

            2- most new apps/features also directly appear in the GMS or PlayStore columns, making them backwards-compatible with older versions of the OS (for example, Pay and Wear were backwards compatible at launch)

            3- indeed, the one good thing about being in the GMS/AppStore columns is that you’re OS-version-independent (also, smaller, individual updates, but that’s a detail), which partially compensates for Android’s update issues: basically, OS updates don’t matter much for features/apps, (they do matter a lot for security).

            Indeed, Google is moving more stuff the GMS and the Appstore. The rule is simple and has never, ever changed: stuff that needs Google’s servers is GMS-based, stuff that doesn’t is AOSP. AOSP is kind of an empty husk, but it’s a fully working OS. It is very convenient for Google that apps and cloud services are what makes an ecosystem valuable, and Google is certainly making sure that you need GMS + Playstore for that. Especially GMS: the PlayStore thing is more about convenience, there’s no rule preventing 3rd party appstores, devs from releasing on other stores, there’s easy sideloading. Even GMS isn’t always required, lot of apps don’t use any cloud stuff, Amazon has a quasi-plug-in competitor for many of those services…
            That’s why Google can be so generous about AOSP, other appstores: it doesn’t really matter, and where it does (China, piracy), having alternatives to the PlayStore actually help spread more Android.

          2. OK, so it’s not so much that things are “moving” to the right, as Google “putting” as many new things as it can into the center and right columns, and keeping them out of the left column.

            It would seem there are downsides to this approach. For example, however well a few Android phones are specced on paper (high-clocked quad cores, loads of RAM, etc.), their actual real-world performance lags. Again, the main upside appears to be that it makes these things accessible to devices that will not get OS updates.

          3. I’m not sure how/why you think moving an app (say, the Dialer) from core-os to appstore has any impact on its performance.

            And it’s really both moving old things to a right-er column and making sure new things pop up directly there. The ubiquitous Webview for example is now appstore-based instead of core-os.

            Indeed, it’s about updating more stuff independently of the update-challenged core-os.

          4. Oh, I would hope that an app like the Dialer doesn’t have any noticeable lag relative to other apps. Obviously, any perceived and real world lag is mostly down to to the things that stay in the left column, like application framework, runtime, and UI.

            “Indeed, it’s about updating more stuff independently of the update-challenged core-os.”

            Exactly. But, it’s still not clear why you seem to think that not getting core-os dates out to the majority of users is no big deal (and seem to use “monolithic” as a pejorative). Apparently you think *every* Android user gets the “full Android experience” (whatever that is), no matter what version of the OS they are on.

            Indeed, you are often at great pains to suggest that they are missing out on nothing (which just says that Google has made no significant improvements to its core-OS from version to version); that the approach of independently updating stuff through the store makes up for the lack of user base on the latest OS.

            Just seems to me that that would leave Google and Android users with a “lowest-common-denomitator” kind of situation.

  3. “I have long believed this is a stroke of genius”

    Evil genius, enabling them to insert their privacy sucking proboscis even further into their user’s veins.

    “This… is a major reason why I think Google
    services in Android devices generate 50% of the revenue they do on iOS.”

    If you ignore all the people who buy android phones because they are the cheap default pushed by their wireless provider, but who are not interested in using the device as a full fledged smartphone and thus seldom use all the revenue generating ad filled services on the device.

    “The only way I think Google can fix this problem is to take complete
    control of Android…. This would fix both the software
    distribution problem and the endemic fragmentation”

    Um… that’s overlooking a lot of barriers to the distribution of software updates. It’s not that Android is open source, it’s that in order to make it pleasing to the wireless carriers and to the OEMs it cedes control over updates to them, so an update issued by Google first has to be passed through the OEM’s process and then through the carrier’s process before it can get onto the phones. Making the OS closed source is not going to change this unless Google renotiates its contracts with the OEMs (with restrictions passed on to the carriers) that take all control or say over updates away from both of them. Which they aren’t going to like. Since updates need to be tested against the hardware in each device, this means Google would have to either test against a ludicrously expansive array of hardware, or dictate what hardware can be used in phones to the OEMs. Again, the OEMs aren’t going to like that, if it can even be done.

    In short, in order to make android updates be timely, Google will have to become a dictator to the OEMs and through them to the carriers. Take away the carrier’s control and you take away most of the reasons they found Android appealing in the first place. And the thing is, the carriers don’t care what OS their customers want. They are perfectly willing to demand a non-google OS that gives them the control they desire. And what do you know, ASOP exists and will still exist even if Google makes all future development of Android closed source.

    1. Updates don’t need to be that hard to do, if they’re well architected. MS has few issues updating tens of thousands of different HW configs, and hasn’t for the last decade.
      I think there are 3 main issues:
      1- the “phone” stack needs to be carefully validated. Luckily, it barely evolves, so what’s needed is simply to isolate it from other updates.
      2- The carrier crapware stack is regular apps that can be PlayStore-based, actually it already is optionally, with a bunch of stuff auto-installing at first run. No reason why those apps can’t be forced onto users via the Playstore, independently of the OS.
      3- OEMs customization fall partly in the same “it’s only apps” category (launchers, custom stores, various system utilities), but also sometimes go deeper (OS-wide pen support, multi-tasking and -windwing pre 7.0…). That’s indeed a problem.

      Mostly, the issue is that the underlying Linux (especially on ARM) never hit the “security reset button” the way MS did: kernel, drivers, libraries, system tools and apps aren’t modularized, so for example a fixed libstagefright couldn’t be pushed to all devices. And Linus Torvalds isn’t terribly motivated by the issue. Which is why I’m very intrigued by Google’s Fuchsia, which could be a way out of the Linux morass and on to a cleaner, Windows-like modularity. Since apps don’t really see Linux but only the Android APIs, I don’t think swapping OSes under them would be an unmanageable project ?

      1. I think you bring up an important point. That is “Who will take responsibility when an update screws up your device?”.

        For carrier-issued Android devices in Japan, the carriers assume responsibility. They provide face-to-face customer support on their ubiquitous shops. For iPhones, the carriers tell you to go to an Apple Store.

        Assuming the same structure, if Google went closed source and sent out Android updates on their discretion, then Google would have to provide the customer support, maybe even charging for it like Apple does with AppleCare.

        I think we are putting far too much blame on the carriers and OEMs for the lack of updates. As far as I know, the burden of ensuring that all devices are compatible rests with the carriers and OEMs, and not with Google.

        If Google could assume this responsibility, then yes I think they could go closed source and send out updates on their schedule. I seriously doubt though, that Google would ever want to go down this route.

    2. Because the users already demand Google Play on their handsets, Google is already a dictator to both OEMs and carriers. They cant demand anything as it stands today because no one will buy a non Google PLay device in developed markets. If the EU forces Google to unbundle PLay from GMS then that is another issue but today Googe has them by the short and curlies.

      1. “the users already demand Google Play”

        But do they really? There’s a percentage of the phone buying public that just get whatever phone is cheapest, or that provides the features they want, but who don’t care about more than that handful of basic functions. My mother, for instance, wants a phone that makes calls, takes good pictures, checks email, and lets her verbally ask for directions or information. Everything else about the smartphone is redundant to her. She would be perfectly happy with an easy to use featurephone that also provided a very limited number of smart features.

        Apple is a dictator to carriers — the only concession Apple makes to the carriers is to allow them to lock the phones they sell. Everything else about the Iphone and its software, the carriers have no say whatsoever. Google is the exact opposite. Back when Google was desperate to make Android a success, they agreed to let the phone makers and the carriers be in the driver’s seat. So we get Samsung’s touchwiz, we get Verizon phones with DT Ignite. And so on.

        1. yes they do.. ask any handset maker in the world and t hey will tell you that they are unable to sell an Android device outside of China and Africa that does not have Google Play on it. Thats why they make no money, all the differentation and user preference is with Google.

  4. I’ve gotten an email notification about a reply from Kizedeck, but till can’t see it in the comments several hours laters. I don’t think it got deleted, it was fairly standard (BTW, i gave a list of Android features that got introduced w/o an OS update: Wear, Pay, Health, Family Sharing… iOS needs an OS update for… Music or a new messaging client ???).
    Back to the point: Is Disqus having issues ? I’ve had issues also this week, one of my replies to Joe about Android got borked when I hit “post”.

    1. Thanks, Obart. I found it in my Disqus account, and will try to repost (under your comment below, where I attempted to before).

  5. Open versus Closed source is not that critical to me. What is critical is an installed base to offer a stable (as in longevity) ecosystem with “open access”. None of this censored BS iOS has.

    There should be multiple stores with “open access” and things can then be censored at the store level. No forcing “Christian Bookstores” to sell “Fifty Shades of Gray”, that should be able to be purchased elsewhere. That is the buyer/seller exercising self-censorship, which is the only acceptable kind.

    BTW, all desktop OSs have “Open Access” regardless of open versus closed source.

  6. Can anyone please refer me to the “loss in the war against Oracle” the author is writing about? All I can find is the fact that Google didn’t violate anything with Java use..

  7. Hello! Do you know if they make any plugins to help with Search Engine
    Optimization? I’m trying to get my blog to rank for some targeted keywords but I’m
    not seeing very good success. If you know of any please share.
    Thank you! You can read similar blog here: Sklep

Leave a Reply to Glaurung-Quena Cancel reply

Your email address will not be published. Required fields are marked *