www. O S N E W S .com
News Features Interviews
BlogContact Editorials
.
Android is a dead end
By Thom Holwerda on 2017-07-16 23:26:41

Dieter Bohn at The Verge:

So while Microsoft didn't do itself any favors, I'd argue strongly that all these machinations and flailings weren't a response (or weren't only a response) to the iPhone. The real enemy was the company that had set its sights on Microsoft's phone ambitions since before the iPhone was released.

That company was Google, of course, and it only tangentially wanted to take on the iPhone. Google's real target was always Microsoft, and it hit the bullseye.

This article looks at the past, so let me take this opportunity to posit something that might come as a surprise to some.

Android is a dead end.

I really want to write a far more detailed and in-depth article explaining why I think Android is a dead end, but I can't yet fully articulate my thoughts or pinpoint why, exactly, I've felt like this for months now. All this doesn't mean Google is going to get out of mobile operating systems, and it doesn't even mean that the name "Android" is going away. All it means is that what we think of today as "Android" - a Linux kernel with libraries, the Android Runtime, and so on on top - has served its hackjob, we-need-to-compete purpose and is going to go away.

Android in its current form suffers from several key architectural problems - it's not nearly as resource-efficient as, say, iOS, has consistent update problems, and despite hefty hardware, still suffers from the occasional performance problems, among other things - that Google clearly hasn't been able to solve. It feels like Android is in limbo, waiting for something, as if Google is working on something else that will eventually succeed Android.

Is that something Fuchsia? Is Project Treble part of the plan, to make it easier for Google to eventually replace Android's Linux base with something else? If Android as it exists today was salvageable, why are some of the world's greatest operating systems engineers employed by Google not working on Android, but on Fuchsia? If Fuchsia is just a research operating system, why did its developers recently add actual wallpapers to the repository? Why does every design choice for Fuchsia seem specifically designed for and targeted at solving Android's core problems?

I don't like making broad predictions based on gut feelings and spidey senses, since they can be incredibly misleading and hard to read, but I'm still pretty confident on this one: over the coming two to three years, Android will undergo a radical transformation. This transformation will be mostly transparent to users - their next Android phone won't actually be "Android" anymore, but still run the same applications, and they literally won't care - but it won't be a Linux device, and it won't suffer from Android's core problems.

In a few years, Google's Pixel phone will have a fully custom, Google-designed SoC, and run an operating system that is Android in brand name only.

Bookmark this.

 Email a friend - Printer friendly - Related stories
.
Read Comments: 1-10 -- 11-20 -- 21-30 -- 31-40 -- 41-50 -- 51-60 -- 61-70 -- 71-71
.
RE: Sounds Familiar....
By name99 on 2017-07-17 17:46:47
Actually I think the more important analogy is to the OTHER change that MS tried to ship at the same sort of time --- the great .NET experiment.

Remember the POINT of .NET was to do exactly what Google presumably wants to do now: throw away the godawful legacy of the past and replace it with something architected for the future. And how did that play out?

So the real issue for Google, IMHO, is not details about the technology of the new OS, it is the extent to which they have the DESIRE and the CAPABILITY to do things differently from MS. In the case of MS, we had internal company antibodies that fought change (of any sort) tooth and nail, the many third parties who were, likewise, uninterested in change if it meant any sort of short-term pain, and a general philosophical belief in the user-base that preferred stasis (and all its problems) to change.
You could make much the same argument for Intel and EPIC.

So how is Google different? They probably have the same internal politics (perhaps not as bad as MS because more of Google are engineers and aware of how badly what they have sucks).
They certainly have the same user base that complains about any change whatsoever.
They have the same vendors allergic to change (and with more power in that they can do the equivalent of when the nascent PC industry rebelled against MCA and created EISA --- they can fork 'Open Android' or whatever and all switch to using that).

Rapid change IS possible --- but in the past it required a company like Apple that controlled the entire stack, that was willing to force short term pain on its customers for long-term gain, and that has created an expectation of constantly abandoning the past in its users and developers. (eg we expect to lose 32-bit iOS compatibility this year. Perhaps in two or three years, we'll be seeing macOS switch to ARM).

The analogy really is with .NET, not with XP --- XP is just like the usual annual Android update, eg bringing in ART. Hell, MS couldn't really pull off the less extreme versions of this sort of switch in the form of Windows 8, or Windows RT, both of which eventually deteriorated into giving up or some sort of stupid compromise that's an even bigger mess than when they started. (Do we land up with phones that are running Android on some cores, Fuchsia on others, a hypervisor underneath both, two different UIs --- plus inconsistent skinning --- and 4x the resource requirements of iOS?)
Permalink - Score: 2
.
Comment by jeffedsell
By jeffedsell on 2017-07-17 18:28:29
Android is where Apple was when it reached OS 9. They realized that they just couldn't continue on the same path. A major change was needed, from the ground up.

Customers don't like change. They complain. They feel betrayed somehow. They swear to leave and never come back.

But if the change is successful and you can keep things accessible enough to your old users in the process, you can become even better.
Permalink - Score: 1
.
Of course...
By rft183 on 2017-07-17 19:00:43
Hey, it's Captain Obvious! And he's making predictions! Don't forget to bookmark the article so that you can come back in a few years to see how right he was.
Permalink - Score: 1
.
RE: Comment by jeffedsell
By rft183 on 2017-07-17 20:01:19
I think that's a pretty decent comparison.
Permalink - Score: 1
.
RE[2]: Not linux that is holding it back.
By cdude on 2017-07-17 21:11:32
The NDK is, of course, still well supported for Android Things. See https://developer.android.com/thi...

Fuchsia uses a wide range of tech including C, C++, python, dart, go, etc. See https://github.com/fuchsia-mirror

Edited 2017-07-17 21:23 UTC
Permalink - Score: 3
.
RE[3]: Comment by CATs
By leech on 2017-07-17 21:21:58
You could feasibly get one of the Samsung Gear S2/3 watches with 4G. Though I am not sure if there is a way to side load apps onto them without a smartphone, but they have uber apps and such for them, can handle your calls (through the watch or bluetooth headphones) and you could even feasibly send/receive messages from them.

They are Tizen based instead of Android.
Permalink - Score: 2
.
RE: The Good OS.
By leech on 2017-07-17 21:28:42
> Fuchsia? Could it be... The Good OS? :)

A successor to BeOS that itself claimed to be a successor to the Amiga.

I expect atleast 0.2ms latency hardware accelerated audio, so that it can compete with the original "instant response" Amiga. (realtime response as many amiga-sources say).

..and it should play my old modules https://www.youtube.com/watch?v=l...

:-)


That's like to this day, there aren't many systems that can get as tight a response on MIDI as the Atari ST can. Got to love the old systems, the OS seemed much more tightly integrated and optimized for the hardware.

Granted these days your hardware could be so many different combinations of things....
Permalink - Score: 2
.
RE[2]: The Good OS.
By tidux on 2017-07-18 01:37:32
That's the tradeoff. Going back to system-specific OSes and bare metal responsiveness means trashing 30+ years of application compatibility for DOS/Windows, and nearly 50 years for Unix-likes. Even most research OSes slap on a POSIX compatibility mode so they can leverage the vast bulk of open source software.
Permalink - Score: 2
.
RE[2]: Sounds Familiar....
By tidux on 2017-07-18 01:40:18
Fuschia will have Linux binary compatibility for the NDK and the usual VM for ART. Linux kernel binary compatibility isn't exactly a hard thing if you have a dedicated loader for them - the BSDs have had it for years, and now Windows 10 does as well.
Permalink - Score: 2
.
RE[3]: Not linux that is holding it back.
By moondevil on 2017-07-18 06:33:18
> The NDK is, of course, still well supported for Android Things. See https://developer.android.com/thi...

The NDK is not what was promised for Brillo developers.

"Brillo/Weave Part 1: High Level Introduction"

https://www.youtube.com/watch?v=A...

"Brillo/Weave Part 2: Deep Dive"

https://www.youtube.com/watch?v=V...

It was supposed to be 100% C++, not a single line of Java, with framework APIs written in C++.

Instead we got the constrained NDK, in case you didn't noticed, only these APIs are allowed, anything else will kill the application.

https://developer.android.com/ndk...

https://developer.android.com/abo...

https://developer.android.com/pre...

Only who never used the NDK can still think it is like coding for Linux.

> Fuchsia uses a wide range of tech including C, C++, python, dart, go, etc. See https://github.com/fuchsia-mirror...

Sure it does, but that doesn't change the fact that the current plans appear to be that apps written by us can only be done in Dart.
Permalink - Score: 2

Read Comments 1-10 -- 11-20 -- 21-30 -- 31-40 -- 41-50 -- 51-60 -- 61-70 -- 71-71

No new comments are allowed for stories older than 10 days.
This story is now archived.

.
News Features Interviews
BlogContact Editorials
.
WAP site - RSS feed
© OSNews LLC 1997-2007. All Rights Reserved.
The readers' comments are owned and a responsibility of whoever posted them.
Prefer the desktop version of OSNews?