Linux for Real

July 2001

We need to coin a new term: a measure of constant change that exceeds the pace of "Internet time" to such a degree that it embodies Heisenbergian uncertainty: it's changing so fast you can't say what it is. Because that's what's happening in the Embedded Space, where developers are picking up Linux faster than slang.

How about Unreal Time? Or Interbed Time? Hmm. Not quite right. How about... Protoreal Time? I like that one. In Greek "proto" means first. It's the root of prototype. It also suggests protean, the adjective for shape-changing derived from the Greek sea god Proteus, who could assume many forms. Protoreal puts change in the domain of reality, which is a fundamental concept to people who design stuff they call "real time".

In fact, time isn't the only proteal variable. So is space.

Business folks like to call categories "spaces", whether they are real or not. Such is the case with the embedded Linux "space". The virtue of Linux to embedded developers is that it radically reduces the threshold for a) designing, prototyping and making all kinds of stuff; and b) equipping that stuff to use the Net, which is the prevailing environment for pretty much everything that can conceivably find a reason to use it -- which is pretty much everything, period.

The same goes for BSD, by the way. Both Linux and BSD are forms of UNIX that grow wild in nature, rather than in the private habitats where captive OS species have been bred by their owners in relative isolation for private purposes.

This threshold-reducing power is not well understood by those whose knowledge of operating systems, however extensive, is informed mostly by the domesticated varieties. These folks think that well-bred operating systems help their owners "capture" or "control" some market "space".

For example, in a May 4 ZDNet story by Richard Simm, IDC analyst Kevin Burden says the Linux PDA "market" is "fragmented and new." He also says, "Without a standard OS, devices may not be able to speak to each other or use the same software, essentially making them islands when it comes to sharing information, and that is a big detriment," adding, "The key to any OS is application support. Without it, you're starting from ground zero with every OS and device using that OS."

This ignores the infrastructural context in which every operating system now lives: the ubiquitous world of protocols like http, xml, tcp/ip and http. These protocols were ubiquitized by their wild and free nature. Every species of Linux and BSD are native to that world's -- the Net's -- ecosystem. In fact to a significant degree they define it.

The last I looked there were eight Linux-based PDAs and three Linux distributions intended for PDAs, not counting Linux itself, which any manufacturer is free to hack into any shape they like (and which, in most cases, is exactly what they do). Meanwhile Palm has one OS and two manufacturers (though none Palm is willing to name any other than itself), while Microsoft has one OS with a pile of OEMs. As for applications, AgendaComputing, which makes the "pure Linux" Agenda PDA, says a "couple thousand" developers are working on Linux PDA apps, and that the AgendaVR will roll out with two hundred in May (two months ago as you read this).

Credit where due: both Palm and Microsoft are very good at their businesses, or they wouldn't be so successful. But their OSes don't grow on trees. Linux does. So does BSD. And therein lies the advantage -- not for those OSes (which no vendor owns), but for you.

If you want to make a PDA, there's less to stop you in Linux and BSD than in any vendor-owned operating system. There's also less to stop you from remaking that PDA. And remaking it again. And again. There is also a growing multitude of folks ready to help, even if all they do is share their hacks on the Net. In other words, most of your OS-related resources also grow on trees.

And PDAs are just one obvious embedded morpology. Thanks to Linux it's a lot easier to make, and re-make, anything. And so what if some of those things use Linux while others use BSD or Windows CE or PalmOS? Out here in Nature, applications do matter -- but communications matter more. If two devices are exchanging XML streams over TCP/IP stacks, does it matter what operating systems are involved? It only matters if those OSes impede functionality or slow down the rate at which their makers can innovate.

Here's another angle on the matter: If you're living and working in protoreality, nothing softens space and speeds up time better than wild and free OSes like Linux and BSD.

And if there is, we'd love to write about it.

(Speaking of which, I like the word protoreal so much I bought the domain name. If you can think of a good way for us to use it, let me know (mailto:[email protected]). If not, it was still a protoreal move to make.)