While most everyone else was wrestling with the specter of Oracle swallowing Sun, I was wrestling with something a little more prosaic: getting my laptop keyboard to not die every time I took my system out of suspend mode while running Ubuntu 9.04 RC. Sometimes all it takes is one thing to wreck your enthusiasm, and I consider a dead keyboard on resume to be a real killjoy. In the aftermath of all that monkeying, I asked myself a question: What is the real goal of desktop Linux anymore, anyway?

What is the goal? Is it to draw people away from the bloat / cost / closed-endedness of Windows (all of which are, I fear, entirely subjective notions; something I'll be happy to go into about in depth)? Or is it to create something that stands parallel to it and offers things specifically to the people who come to Linux rather than people who are invited to use Linux?

I've written before about how Linux in the abstract doesn't need to be any particular thing to succeed. It's raw material, a starting point, and the fact that you can create any number of things with a Linux kernel is its greatest boon.

But it also means that once you pick a destination, you have to go all the way there. If you want to create a useful desktop environment -- especially if you're intending to create it as competition for Windows (if only in the sense of competing for the user's time and attention, which are far more valuable than their dollars) -- then you cannot just copy the outside aspects of other people's work. You need to do truly difficult and thankless research into real-world usage. You need to find out for yourself what people are trying and failing to do, not simply read the sanitized and filtered use cases that come to the programmers through self-submitted bug reports. Or you need to take the even tougher step of saying "this is what we are not trying to do", and stick to that religiously.

I don't think people realize just how incredibly difficult all this is. It's taken Microsoft, with all of its money and resources, decades to get to this point -- and a lot of that sweat was not writing code, but learning about the way people used their machines and complementing that as best they could. They may not get it all right, but they get it more right than most other people I could name (except for Apple, who also have the benefit of a totally insular software and hardware ecosystem).

I mention this stuff to Linux advocates, and I get one of two reactions:

1) "Well, of course Microsoft can afford those things; they've got a lot of money."

2) They pretend I never said that and move on to something else.

The first reaction is pure defeatism. I'm positive there are other ways to gather that information reliably without spending tons of money; it's just that the minute you mention anything at all about collecting even sanitized information about user behavior, people (often folks who have no vested interest in seeing something like this work out) begin to scream and thrash around and foam at the mouth. And the second reaction is, as you can guess, beneath contempt.

What is the goal? If the goal is to simply create something which provides desktop functionality for environments where cost, space, and hardware resources are issues, then good, we're already there. In fact, I can think of a place where all this is already happening, and it's not on laptops in developing countries. It's in your cellphone.

From the standpoint of such a POV, the desktop is not even worth the trouble; it's scorched earth. Most people are going to consider the surcharge for a Windows preload to be more than a fair price to pay to not have to deal with hassles like the one I described above. As tech-savvy as I am, I'm one of those people; I, too, have a frustration limit, and as long as Windows exists as a bog-standard fallback, it'll be very hard not to go back to it.

I'm not going to deny that there are plenty of individual Linux success stories. I have plenty of my own. I had an old notebook that ran XP which I outfitted with Puppy Linux and gave to a friend, for instance. Not a speed demon, but it allowed her to get some writing done when her desktop was unavailable. But there was never any talk of getting her to leave Windows behind -- and despite good experiences with Puppy, she was still staunchly uninterested in switching.

Her next machine ran Vista, and guess what: she hasn't run into any of the litany of virus/spyware/bloat/slow issues that most of the switch-to-Linux crowd use as their main terms of evangelism. She's already where she wants and needs to be. Saying that she needs to come to Linux so she isn't locked into anything she doesn't want is starting to take on the flavor of the argument that black-and-white movies need to be colorized because they're "missing something".

So, again: What is the goal? Is it to create as compelling a case as possible for people like her to switch, people with zero interest in changing OSes because the issues in question are entirely unconnected with the work they do? That way, I suspect, lies madness.

There is the possibility that the open source development model -- which revolves around transparency and inclusiveness -- is simply not the best way to develop a desktop OS, where stability and consistency and uniformity of approach from the inside out are all paramount. I would like to think that's not the case; I'm grateful we've gotten as far as we have. It's just that it's all far more difficult than people realize, for both philosophical and technological reasons. And a consistently working keyboard would be nice, too.

InformationWeek has published an in-depth report on the most innovative startup technology companies. Download the report here (registration required).

Follow me and the rest of InformationWeek on Twitter.