Linux and embedded converge at the technical level, but they clash at the cultural one. At the technical level, Linux is like an element in the periodic table, or a pile of 2x4s. You can use it to build all kinds of stuff. Sure enough, all kinds of stuff does get built, including embedded stuff that most hard-core Linux hackers would never develop, or that they would develop in a very different way. We're talking here about stuff that is typically closed as a vault. Examples include electronic picture frames, security cameras, airplane avionics, vacuum cleaners, assembly-line robots and medical equipment. And, of course, mobile phones.
Linux has been at the heart of millions of phones made and sold during the last few years. Few of these, however, are what you would call “open” to the degree that you can do what you please with them—that is, as you would with a PC. At the low end, phones are dumb devices that do little more than telephony. At the high end, they're smart devices that do lots of different things. But alas, not in the same ways—meaning, there is no “write once, run everywhere”.
Right now, the Linux smartphone market is split between Android, LiMo and Palm. You write apps for Android in Google's own version of Java5 (described by one developer as “not J2ME, but not quite Java5 either”). You write apps for LiMo in C or C++. You write apps for Palm Pre in JavaScript. Of those three, Palm is the only standalone vertically integrated platform, kind of like the iPhone. The LiMo vs. Android situation is a lot more complicated. The LiMo Foundation has a pile of big names behind it, not including Google. Google's Open Handset Alliance (all developing for Android) includes its own pile of corporate heavyweights, some of which also are involved with LiMo. The differences are easy to play down. For example, LiMo focuses exclusively on middleware, while Android focuses on applications (but covers the whole stack). But, Linux for phones is beyond forked. Which brings me to another F-word.
Freedom.
Phone makers and phone companies don't come from freedom. They come from control—specifically, of whole markets. Take one example from page 12 of the LiMo Foundation's Introductory .pdf (dated July 2009). There, it explains where LiMo's middleware platform fits in the scheme of things. Its middleware sits below three other layers: 1) UI/Applications, “As selected by the handset maker/operator”; 2) General Content, “As selected by the mobile consumer”; and 3) Applications, “As selected by the mobile consumer”. In other words, LiMo primarily serves markets that are run by handset makers and phone system operators. Of course, so do Android and Palm. (Not meaning to pick on LiMo here; they just provide a handy illustration.)
Why, as we enter the second decade of the 21st century, should we be forced to choose only from applications provided by handset makers and mobile phone network operators? Can't we finally have an open phone marketplace—one where users and developers are free from control by both makers and operators, where you don't need to get your apps from a “store”, where you can make any app you want, for any purpose you want, without confining your innovations to what phone makers and systems operators allow? Hey, that's what we've had in the PC marketplace for going on 30 years. It's what we've had with the Internet for 15 years. Can't we have it in the phone marketplace too?
In a word, no.
Or, in two words, not yet.
The problem is that the phone business has never been open or free. It's one of the most tightly closed and highly regulated businesses on the planet. True, the Internet started cracking open landline phone systems nearly two decades ago, but it's still just getting started with mobile phone systems and the devices that run on them.
Mobile phones today are a bridge across a chasm between protocols, technologies and business categories. To see the big problem writ small, consider the natural conflicts between SIP (Session Initiation Protocol) and SS7 (Signaling System 7). SIP is a peer-to-peer protocol defined by the IETF (Internet Engineering Task Force). It arose from the obvious need for IP-based calling that looks and feels like what we're used to with phones, but without all the other hassles, such as centralized switching and billing for everything. SS7 is defined by the ITU-T, or the Telecommunication Standardization Sector of the ITU (International Telecommunications Union), which began as the International Telegraph Union in 1865. SS7 is the heart and soul of every phone system, including all the mobile ones. While the IETF values “rough consensus and running code”, the ITU values top-to-bottom definitional completeness. Yet to various working degrees, SIP and SS7 coexist in our lives, devices and applications.
Trends favor the IETF and the Net, but I wouldn't bet that way in the short run, which may be a decade or more. We see positive signs with smartphones and handhelds for which telephony is one application among many. But the telephony space is a billing space, and it still rakes in the cash. The Net may be a “world of ends” (worldofends.com), but its means include a vastness of phone company cabling, routing and switching. And, most of all, billing.
It is essential to recognize that billing is the core competency of telephony. The Net threatens that. I'm betting the Net will win. But it will be a long, hard fight. And Linux will be right in the middle of it, serving both sides.