Linux world domination is a playful contradiction of terms. As Eric Raymond jokes in the February 2000 issue of Linux Journal, such a domination would not be a domination at all. Freedom is too interwoven into the overall Linux fabric. And Linux is only a starting point.
If Linux were to replace MS Windows as the predominant desktop operating system, what would this really mean? What if ninety percent of all PCs were run by Linux? How would this affect us?
First, if everyone were running Linux, Linux would scarcely resemble Linux as we know it today. As Linux user numbers grow, diversity multiplies exponentially. So, at this fictional point in time when all PCs run Linux, operating systems themselves will probably already be a commodity. Much like other things are commodities today, like motherboards, PC cases, hard drives, and many small software utilities that PCs use today, such as graphics viewers, file managers, text editors, and TODO managers.
The name ``Linux'' would probably be a euphemism for a free operating system kernel by this point. In reality, it might be BSD, QNX, Hurd, Linux, or one of many new Unix-like kernels available. Perhaps everyone will be running several operating systems under some multiplexed hardware using something like VMWare. So, the term ``Linux'' would probably mean some generic freeware Unix-like operating system by this time. Call it ``Freenix.''
Next, if Linux dominated the world, OS licensing would scarcely resemble itself from a legal standpoint. The money in overall software development and marketing would shrink to smaller proportions compared to where they are now. For example, MIS departments wouldn't have to pay through the nose for licenses ``by the seat.'' That would be like going into a restaurant and being charged for water by the glass. Some people will still try to do it, but the practice will seem very ``20th century'' and outdated.
When Linux dominates as the de facto PC operating system, it won't be because of marketing clout or naive consumers, it will because, thanks to it, compatibility and interoperability are no big deal anymore. Smarter consumers in a NetCentric world will just expect to use what works, and do so at a reasonable price.
Unix geeks will be more commonplace, like the corner gas station in the 1950s. So no one pays huge prices for support and consulting anymore. Of course, there will probably continue to be esoteric pockets of expertise where the dollar doesn't go as far as it used to go. Since everyone is online nowadays, in this imaginary Linux-centric future, certain specialties will continue to fetch large fees, such as DNS configuration, security consulting, and integration management (my own term for integrating network services that haven't been integrated ever before). But through all this, Linux, also known now as ``Freenix'', will simply chug along and give people what they need or want. It will come in many shapes, colors, and sizes, much as Linux does now with its many distributions, only much more so than today.
While the OS itself will be a commodity, certain applications will evolve and continue to boldly go where only big bucks can take them. Gaming, simulations, and virtual reality based applications will require the latest hardware and software innovations. People will continue to pay for the thrill of simulated experiences: warfare, love affairs (also known as lust), thrill seeking, and the old favorites, such as football, basketball, baseball, and military strategic simulations. Since these applications are so hardware and OS intensive, a commodity OS may not work as well, and large portions of the OS might have to be rewritten by these game developers. Or, they will be loaded as specialized modules into the generic OS, and other modules will have to be unloaded before these applications can be run.
Actually, if OS design evolves where it should, the microkernel will become the norm. In this case, monolithic kernels that are familiar today will be replaced with smaller ``miniservers'' supported by a kernel that is so small it is fully pageable to memory. This allows entire aspects of the system to be accessible without special privileges yet without making the system's security vulnerable or its services unstable because of untested or experimental code. Yes, it sounds very strange, but it's a brave new OS design with many advantages and very few disadvantages. (The primary one being it's very new to non-scientists and difficult to explain! See www.gnu.org/hurd/.)
When Linux becomes the dominate OS, a lot of software stores won't be selling software as their only products. Software stores will become more like novelty stores or ``head shops,'' as they were known in the 60's. Most software will be available for free over the Internet, as it is now, only most people will know about it and depend on it.
It's just that buying software in a store will seem more like a novelty; real software upgrades and installations will be acquired automatically by packaging software or by system management software that manages dependencies with a friendly GUI smile, or invisibly while you sleep and when the Internet isn't as busy. You'll just be notified by email from your system adminbot that certain upgrades have been applied to your system and will print a detailed report if you're interested.
When you want the latest version of a new window manager, just point your package management tool at the nearest software repository, and your desired window manager will be installed, along with whatever other pieces of software it depends on to go along with it. Yes, most Linux/BSD people do something like that already, but this idea will blow a lot of Windows converts away in the future.
Since software retail stores will largely become software novelty stores, chances are you won't be seeing a single dominate set of file formats in the future. In that sense, file formats will resemble, on the surface anyway, the nasty old mid-80's when your partner worked with WordPerfect 4.0 and you worked with WordStar 3.2, and you had to find a way to share your files. Thankfully, there are geeks out there for whom writing a conversion script is not that big a deal, so you can usually have your cake and eat it to when it comes to file formats. What will make this possible is that people will write SGML-like tools for their favorite document environments. All you have to do is run some program on your document that will flatten it into a universally readable version which you may share with anyone. (Structured authoring may be a tough sell for some people, but perhaps these conversion scripts may apply some intelligent interaction with a document's creator to overcome this obstacle.) That document may then be converted into Postscript, HTML, RTF, or whatever. Actually, that is all possible today, but only geeks seem to know about these tools and use them. Everyone else seems to be an MS Office file format slave.
One thing that most of the various Free Unix-like operating system communities currently have in common is the use of SGML for their documentation projects. It seems natural to me that pretty-looking front ends to tools like these will become popular once the Microsoft file formats become less and less popular. There will still be few people who understand SGML, but the tools will work and nobody will care about why or how. In fact, they will probably become integrated into server-side filtering services through email or web-based file transfers.
Through all this Linux revolution we're witnessing today, several themes have emerged which will continue to exist throughout the Linux (or Freenix, more aptly) reign as dominate OS:
To those of us who have been using the Freenixes for some time, it may seem incredible that they have caught on so. Or that they could become the dominate operating system in existence. But just notice how much they have changed themselves in the last five years. Do you remember what Slackware looked like 6 years ago? I do. And it would be nothing short of a miracle if that product, in its 1994 state of being, were to become dominant. It sucked back then, though it was still better than Windows. But it would not be so surprising if the year 2010 product came to dominate. The rate of improvement has accelerated dramatically.
I will be surprised if Freenix doesn't dominate in the future, simply because the open source software development model works very well with operating systems. I will be very surprised if Microsoft can compete with Freenix in five years time. MS is already playing catchup in terms of quality and cost. They currently have to compete on the basis of FUD (Fear, Uncertainty, and Doubt) and a naive customer base. In ten years, it won't just be ``Linux'' or any other single kernel name, it will be an entire worldwide open source community whose name will be ubiquitous and whose products and methods are the first choice among experienced computer users. At this time, there will, of course, be a Microsoft Linux (and a Microsoft BSD, Hurd, etc.), and it will actually be quite good. Finally it will be finally okay to buy Microsoft!