While I normally turn a cold shoulder to all those increasingly arcane ways of dividing people into two groups we keep inventing, there's one that I think holds the key to Linux's prospects for success on the mainstream desktop.

The first group thinks Microsoft's virtual monopoly of the mainstream desktop is a Very Bad Thing For Everyone, and that the contender with the best chance of ending that monopoly and restoring genuine choice is Linux. This group also believes that Linux programmers can and should take on this very formidable task, regardless of what Linux or any part of it "was originally meant to be". I'm in this group, and I hope you are, too.

The second group are all those who either want Microsoft's monopoly to endure or don't think Linux can compete in that toughest of all computer markets, the mainstream desktop, or they simply don't want to see the masses using Linux, since it would change their favorite OS or remove a "special" part of their identity, i.e. being able to master an OS that's too tough for the washed masses. If you're in this second group, please stop reading here and go take part in the latest food fight in a Linux newsgroup or on Slashdot.

Now that we've whittled the audience down (but hopefully only a bit), the obvious question is: What will it take to make Linux a Windows competitor in the mainstream desktop market? As I see it, this requires the following 12 steps, some of which are ongoing, essentially endless tasks, some of which have discrete ends:

1. Recognize that the mainstream desktop market is just as different from the expert user desktop market as it is from the server market. The easiest way to see this is through a little painless self-education. Frankly, nothing will boggle the mind of the prototypical, overly insulated Linux expert more than will than spending an hour or two sitting next to a non- or semi-expert Windows user while he or she uses the computer. You'll be both amazed and depressed, particularly in terms of Linux's prospects in that market.

2. Never forget that not all costs are measured in money, and that the mainstreamers have much a different balance between financial and non-financial costs than the rest of us. In blunter terms, mainstreamers are much more willing to pay for software than is the average Linux aficionado, but they're also far less willing to endure hassle and inconvenience. Perhaps most important of all is the fact that mainstreamers think of forced education as a very big hassle. Most of them don't like computers or OS's or even their apps. They simply view all that hardware and software as a steaming pile of necessary evils to be endured in order to get what they want from a computer. They want to finish their work (or homework), play a game, download music, burn CD's, send e-mail to a distant friend or relative, and shop or get product information on the 'net. These kinds of activities are the sole payback for computer use by mainstreamers, and unlike true computer geeks, learning about the computer or mastering it is a cost, not a benefit.

3. Recognize that adding features to help mainstreamers doesn't necessarily take away anything from the geeks. A program with a really slick, GUI-based installer can still include all the manual configuration options you could want, full source code, and more. If anything, it's a far greater test of your programming skills to create something that both Joe Sixpack and Joe Overclock will use.

4. Put a sign on the edge of your monitor, where you can see it at all times, that says: "New isn't better, different isn't better, only better is better." If Windows or OS X or who-knows-what OS does something better than Linux, then by all means copy the idea (if you can do so legally). If you can cook up an even better solution than go ahead an innovate, but don't forget for a second that making something a little better than Windows but very different will drive Windows users away from Linux. This is true whenever users perceive that the hassle cost of conversion outweighs the benefit.

5. Continue to improve all aspects of installation under Linux. The installation of Linux itself was once a nightmare for the average user, and has been dramatically improved with most distros. The people who did this work should get a standing ovation, in my opinion. But the system installation procedures still aren't good enough, and the installation situation for apps is downright dreadful. This is one area where Windows is still light years ahead of Linux. In particular, a user should never be forced to do any of the following to update or maintain a Linux desktop system:
a. Use a compiler.
b. Edit a script or configuration file.
c. Know which directories contain special files on their system.
d. Know about special details of their hardware, whether it's the refresh rates of their monitor, which CPU they're running, or almost anything else.
e. Deal with differences between the various major desktop environments. This includes things like the infamous cut and paste headaches between KDE and GNOME, but also smaller things like making sure an app's installer has the option of putting an icon to the new program on the user's desktop.
f. Endure pain because apps don't uninstall cleanly. All apps should have a fully automatic uninstaller that restores the system to the state it was in before the app was installed, which the exception of the presence of documents created by the user with that app.
g. Do anything special to get TrueType fonts to work and look very good. On a Windows 2000 system, you install a TT font by copying it to c:windowsfonts. Yes, this requires you to know about the special font directory, but that's it. I've spent a lot of time messing around with TT fonts under Linux over the years, and it's not a procedure I'd recommend for mainstreamers.
h. Endure the hassle of running as user vs. root. I'm not saying that all Linux systems should run as root all the time, ala Windows 9x, but that the hassle factor of running the right way should be lowered as much as possible.

6. Do whatever is needed to make a GUI Linux system look like a single, monolithic entity. New Linux users go ballistic every time they trip over one of the many seams between the kernel, X, their window manager, and their desktop manager. Users understand that apps are add-ons, if only because they often add them, but they expect "the system" to look and act like a system, and not a bunch of software flying in loose formation that was written by different people with radically different philosophies. A similar problem of the left hand not knowing what the left hand is doing involves the never ending hassles of mounting and unmounting removable disk volumes, particularly floppies. One of the most counterintuitive things for a new Linux user is the notion that you have to tell the computer that you just put a floppy in the drive. Under Windows you put in the disk and just use it. If you forgot to put the disk in the drive, then Windows complains and gives you another chance to insert the disk.

7. Act like adults in public. If you want to flame each other into Bacobits in private e-mail, knock yourselves out. But when you're talking in public, which includes newsgroups, web-based threads, and discussion mailing lists, treat each other civilly. I can't stress this enough: Every time you dismiss a newbie's question with nothing more than "RTFM" or tell someone who's asking for a feature to program it themselves or start a flame war with another Linux project you're handing Microsoft a big stick they can use to beat back Linux, and you're scaring off more would-be converts. If nothing else, rely on every mother's and grandmother's favorite piece of advice (after the one about not running with scissors): If you can't say something nice, keep your mouth shut.

8. Be smarter in how you promote Linux. Look for specific benefits that Linux provides that will make mainstreamers happy. For example:
a. Tell the average mainstreamer with kids about how thoroughly you can isolate individual user accounts on a Linux system and protect other users and the system itself from mistakes (i.e. sandboxing the kids), and his or her eyes will light up. (Note that this is an advantage that's quickly evaporating, thanks to the spread of Windows XP.)
b. Use loopback distros like Phat Linux to demo Linux for users. I've never been able to figure out why the Linux community hasn't exploited this resource far more than it does. These distros are the perfect way to let users play with and learn about Linux on their own system with zero risk to their Windows installation, and then back out just by deleting one directory of files, if they decide Linux isn't for them.
c. Drop the "MS is the evil empire" and "free software will improve your sex life" propaganda. I know from experience that nothing will convince a mainstreamer that you're a nutcase zealot quicker than such statements.

9. Continue to knock down as many of the barriers to conversion Windows users face as possible. This requires looking at a PC the way mainstreamers do: A pile of mysterious hardware and even more mysterious software that happens to contain their precious documents. In fact, I've long believed that mainstreamers understand better than most self-avowed experts that their documents and configuration settings are where their investment of time and energy lies. Figure out how to move over as much of the Windows configuration as possible in an automated way, from user accounts and passwords to TrueType fonts to desktop settings to app settings. Imagine a loopback distro that included a utility like this- -run it once and all those little things users hate to re-discover and set up, like how to set the default font in their spreadsheet or how to configure their word processor to display paragraph marks. This would make it much easier for users to get comfortable with Linux very quickly and not spend hours wrestling with configuration hassles with an OS and a whole slew of apps, all of which are new to them. Yes, dealing with all the sources of such data in Windows and the even larger number of formats and targets in Linux would be quite an undertaking.

10. Continue to improve all aspects of hardware support. This is more than just drivers (although that's a very important and ongoing battle); it's the whole "end user experience", as the marketeers say. Strive to make it as easy as possible for someone to go into the local OfficeMax or Fry's, buy some gizmo, like a digital camera, a new and different model graphics or sound board, a scanner, or a USB- or Firewire-attached external drive, and use it without hassles. And remember, "hassles" in this context refers to any need to rebuild a kernel, change boot parameters, compile anything, edit anything by hand, run a configuration program (X project, take note), or even see a command line.

11. Figure out how to get more Windows apps running, and running perfectly, under Linux. No matter how much you love and cherish GIMP and other Linux programs, and no matter how much you can argue about them being functionally equivalent to commercial Windows software, the bottom line is that they're not the apps mainstreamers want to use. They've already invested a lot of time (which they never have enough of) in learning the programs they use now under Windows, an investment they consider a very high cost. Tell them they have to ditch all that hard-won knowledge and learn a whole new set of apps, or, much worse, do without certain things entirely, and they'll lose interest quicker than you can say "blue screen of death." What do people run? Go back to my first point, about sitting down with a mainstreamer and watching. You'll see them running things like AOL, greeting card creation programs, games, and lots of oddball apps that will never have direct Linux equivalents.

As two examples of that last (meta-)category: One friend of mine is a tugboat captain who has some serious naval mapping software on his system while another friend is a CPA who has to use every accounting program on the market in order to handle files created by his clients. Neither of these people like Windows in the least, and they've both asked me several times if Linux would be a good option for them. Every time I've had to tell them there's no point in trying, since there's no way they can run the software they need. And please don't even think about suggesting to such users that they use a dual-boot system or a virtual machine; that's a perfect example of not understanding their world view and pushing the wrong solution, and it will only drive them away. And no, WINE isn't ready yet for such duty. It's come a long way, but it's nowhere near ready for my CPA friend (who also happens to be my accountant) to risk his business on it.

12. Produce first-rate, end-user-oriented documentation for Linux, including both online help and linear print-and-read manuals. I know, I know--no one likes to write docs. I've been in the computer business since 1980, including a long stint as a programmer and designer at IBM, and I hated writing docs as much as anyone, but it has to be done. The current documentation for Linux stinks out loud, particularly the man pages, for more reasons than editor Dennis could possible give me room to cover in this article. Let me say it one more time: Users detest forced education, so the easier you make it for them to teach themselves, the happier they'll be. (They'll be happiest of all with a system that minimizes the need for self-education in the first place, of course, but some will always be needed.)

That's one heck of a to-do list, isn't it? And it's packed with decidedly unsexy work that's atypical for most (but not all) Free Software/Open Source projects. Even more daunting is the notion that nearly all of those things must be done before Linux could hope to grab even 10% of the mainstream desktop. Do you think the Linux programmers are up to the job? I'm convinced they are, and I see absolutely no reason why they can't pull it off. Whether they collectively have the will to do this much work of this kind, as well as make the needed changes in attitude, remains to be soon.

But imagine for a second the payoff if the Linux camp jumped at the opportunity and managed all this: They'd free millions of end users from a monopoly and (eventually) create a huge, paying market for Linux software that would guarantee a flood of new apps, making Linux a far more usable, enjoyable, and productive environment for us all. Perhaps more enticing yet to the hardcore Linuxites is the notion of humbling Microsoft and greatly reducing their influence across the entire computing industry. I have to admit that even for a someone as strongly user-centric as I am, that does have a certain appeal.

However you approach the results, the prospect of doing that much good for that many people would be sweeter than words can express, and the Linux movement would finally, at long last, deserve to be called a "revolution." Ultimately this comes down to how thousands of developers answer the same question: Are you willing to do what's needed to win the war against the "evil empire," or are you content with sniping at Microsoft from the sidelines while they rule the desktop industry for the foreseeable future?