Paranoid Android
Nov. 6th, 2007 11:39 am![[personal profile]](https://www.dreamwidth.org/img/silk/identity/user.png){kind=small}
daveon So Google have finally come clean about what they want to do in the phone space. It's been widely leaked inside the industry for a few months now, but it's nice to have it out in the open. The noise the more astute of you may have heard yesterday was Microsoft, Symbian and Nokia letting out their breath in a relaxed fashion. Why?
Platform wither thee? The whole point of a Platform in a mobile phone is to handle the issue of hardware abstraction properly and form a common framework for running applications on. That way you have a lot of 3rd party application developers making cool stuff for your phone platform - it's certainly the Google vision.
The trouble is, in practise, it doesn't quite work that way. Because every silicon core is different in the way it handles multimedia, graphics, input etc... (this package of drivers and integration work is called a BSP in the industry) you tend to build up small but significant changes in every solution you build, even if you're using the same core OS. The way that Microsoft and Symbian handle this is by actually having a core platform but locking down a lot of functionality and having a security model in place to block applications if they're not welcome.
What does this mean for Android: it means the same for Android as it has for all the other Linux, Linux/Java platforms over the last 5 years - they end up as a bunch of competing proprietary solutions that have been built up at non-trivial cost by the manufacturer. So each of the key OEM and Operator partners will end up with their own variations which will, to a greater or lesser extent, be locked to their particular build. Without somebody acting as policeman and judicial system for the platform, it will tend away from being a platform for a host of sensible engineering issues arising from how BSPs work.
Lock it down, security: Now you've spent money getting the platform working on the silicon of your choice (and no, the Silicon Vendors will not do that work for free and sell it as a product) you have the problem of security for applications and exactly what they have access to. Like with PCs you don't just want an application having full run of the OS and all device functions - especially the radio. The operators are not going to let Android become the preferred platform for cell tower hackers (believe me 3G can do some wacky things if you have access to the stack). This is why Apple really really didn't want to release an SDK for the iPhone and why Apple tend to be anal about their hardware platforms. Nokia are pretty much the same as the more astute of you will have noticed with the supposidly "open" Series60 platform.
So, to mitigate against these problems Google have announced the Open Handset Alliance - there's lots of these alliances, the Open Mobile Alliance was quite big for a while, AMD annouced one around embedded Linux for phones a few months ago. The issue here is that mobile phone companies join partnerships. Let's look at some of the partners...
The OEMS - the actually handset manufacturers - announced were Motorola, Samsung, LG and HTC.
- HTC: Trying to be a OEM and moving from the ODM space in which they made most of Microsoft's devices - they also manufacture phones for lots of people like Palm etc... They'll probably get some devices to the market first.
- Samsung: The world #2 like to sign software licenses - they have pretty much a license for every platform there is. They tend to not bother launching unless somebody else has
- Motorola: The world #3 - see Samsung (same story) although they have their own Linux strategy and Linux Java strategy AND 2 of their own platforms AJAR and UIQ
- LGE: See above, what Samsung does, so does LG
Do you see a pattern there? None of them, with the possible exception of HTC really tend to push into platforms in a structured way.
Software Partners: TAT, Nuance etc... all cool companies, I like the TAT UI stuff. But nobody who actually writes phone applications that launch...
Integration Services: Nosser - who? I work in the phone software integration business and nobody we know has worked with them. They've got a good biz dev person, or they've friends at Google (it happens).
Operators:
- China Mobile: They like Linux, this makes sense - China's probably a good market for this.
- Telefonica - the Telefonica alliance has a track record of this sort of thing starting with millions on i-mode in Europe
- T-Mobile - a more credible brand
Missing are the major operators in Europe and the US.
It could be an interesting play, but nothing here will make the current players lose any sleep.
On a final note: the guys behind Android come from Danger. The Sidekick might be so simple that a Blonde can use it, but their track record of working nicely with others, especially partners in the mobile space was awful.
There's nothing wrong with Google's strategy, in general. The problem is that it's hardly new - a Linux core OS with a general applications platform on top is pretty standard these days and it's been struggling to get anywhere for a lot of reasons. I'll tackle the problems, and why I have little time for the "Open Handset Alliance" news separately.
What exactly do we mean by Free? In reality the OS and platform is a pretty small part of the effort of getting a phone to market, and while Microsoft, for example, charges anything between $6 and $12(ish) a unit for Windows Mobile - you actually get a lot of the heavy lifting done for you in that. Just being given the OS and core platform doesn't really save that much overall in the effort for making a phone.
Platform wither thee? The whole point of a Platform in a mobile phone is to handle the issue of hardware abstraction properly and form a common framework for running applications on. That way you have a lot of 3rd party application developers making cool stuff for your phone platform - it's certainly the Google vision.
The trouble is, in practise, it doesn't quite work that way. Because every silicon core is different in the way it handles multimedia, graphics, input etc... (this package of drivers and integration work is called a BSP in the industry) you tend to build up small but significant changes in every solution you build, even if you're using the same core OS. The way that Microsoft and Symbian handle this is by actually having a core platform but locking down a lot of functionality and having a security model in place to block applications if they're not welcome.
What does this mean for Android: it means the same for Android as it has for all the other Linux, Linux/Java platforms over the last 5 years - they end up as a bunch of competing proprietary solutions that have been built up at non-trivial cost by the manufacturer. So each of the key OEM and Operator partners will end up with their own variations which will, to a greater or lesser extent, be locked to their particular build. Without somebody acting as policeman and judicial system for the platform, it will tend away from being a platform for a host of sensible engineering issues arising from how BSPs work.
Lock it down, security: Now you've spent money getting the platform working on the silicon of your choice (and no, the Silicon Vendors will not do that work for free and sell it as a product) you have the problem of security for applications and exactly what they have access to. Like with PCs you don't just want an application having full run of the OS and all device functions - especially the radio. The operators are not going to let Android become the preferred platform for cell tower hackers (believe me 3G can do some wacky things if you have access to the stack). This is why Apple really really didn't want to release an SDK for the iPhone and why Apple tend to be anal about their hardware platforms. Nokia are pretty much the same as the more astute of you will have noticed with the supposidly "open" Series60 platform.
So, to mitigate against these problems Google have announced the Open Handset Alliance - there's lots of these alliances, the Open Mobile Alliance was quite big for a while, AMD annouced one around embedded Linux for phones a few months ago. The issue here is that mobile phone companies join partnerships. Let's look at some of the partners...
The OEMS - the actually handset manufacturers - announced were Motorola, Samsung, LG and HTC.
- HTC: Trying to be a OEM and moving from the ODM space in which they made most of Microsoft's devices - they also manufacture phones for lots of people like Palm etc... They'll probably get some devices to the market first.
- Samsung: The world #2 like to sign software licenses - they have pretty much a license for every platform there is. They tend to not bother launching unless somebody else has
- Motorola: The world #3 - see Samsung (same story) although they have their own Linux strategy and Linux Java strategy AND 2 of their own platforms AJAR and UIQ
- LGE: See above, what Samsung does, so does LG
Do you see a pattern there? None of them, with the possible exception of HTC really tend to push into platforms in a structured way.
Software Partners: TAT, Nuance etc... all cool companies, I like the TAT UI stuff. But nobody who actually writes phone applications that launch...
Integration Services: Nosser - who? I work in the phone software integration business and nobody we know has worked with them. They've got a good biz dev person, or they've friends at Google (it happens).
Operators:
- China Mobile: They like Linux, this makes sense - China's probably a good market for this.
- Telefonica - the Telefonica alliance has a track record of this sort of thing starting with millions on i-mode in Europe
- T-Mobile - a more credible brand
Missing are the major operators in Europe and the US.
It could be an interesting play, but nothing here will make the current players lose any sleep.
On a final note: the guys behind Android come from Danger. The Sidekick might be so simple that a Blonde can use it, but their track record of working nicely with others, especially partners in the mobile space was awful.
