With all the hype (and momentum) surrounding Web 2.0, I believe now is the time to start thinking seriously about replacing the traditional desktop operating systems (Windows, Linux, and OS X). Today, our desktops are fraught with numerous unnecessary problems and complexities.  You have to deal with messy install/uninstall programs, compatibility and versioning problems, and of course application bugs and patches.  You have to be diligent about patching your OS in order to combat worms, viruses, and spyware. As apps and the OS bloat over time, hardware limitations with processing, memory, and disk space make periodic system refreshes still necessary.  On top of all of this is the fact that it takes vendors an enormous amount of time to deliver new application or operating system functionality to the masses, both in terms of sheer logistics, but perhaps more significantly in terms of the length of time it takes to create a new version that customers are willing to pay for.

On the whole, people have to spend too much time maintaining their systems.  Computers should be more like TVs from a maintenance perspective.  With a TV, you only have to think about the size, resolution, and display capability of the screen and the type of connection you want to a content provider (digital/analog/HD/etc.).  You don't have to worry about installing new software or patching the system on a regular basis.  All the content and functionality you get is provided from an external source.  The web can do this for the personal computer. 

The idea of a WebOS (or what I refer to as the web desktop) is nothing new.  Sun tried to push a Java-based network computer in the late 90s and several startups went down a similar road during the dot-com boom, but failed.  Back in January 2004, I sent an email to Tim O'Reilly, which he excerpted here where I briefly outlined my thoughts on how the browser could be the primary (or only) application you need on the desktop.  (Unfortunately, Tim focused on the IOS tangent in his reply.)  More recently, Jason Kottke had a blog entry titled GoogleOS? YahooOS? MozillaOS? WebOS? that is similar to what I've been thinking (there are some important differences that I'll get into shortly).  Also, this seems to be the general direction the Web 2.0 community is going.  The recent gossip about Sun and Google working on a web-based office suite suggests people may be ready to rely on the web to provide more functionality to the desktop.

My vision for the WebOS is simple.  There would be a very minimal operating system to boot, load the necessary drivers to support IO, video, network card, etc., and to run the primary application—a web browser.  It could use a stripped down version of Linux with a very small footprint.  The standard CPU and memory available today would be fine, but it would need little disk space.  Most everything goes on the web and little is stored locally.  Of course this necessitates a constant network connection.  This is one divergent point from what others, such as Kottke, have suggested before.  I don't think we need to cache stuff locally.  Have a web server running on your local machine?  That's just another complexity (and security vulnerability) you don't need to deal with.

Bandwidth is constantly improving and more people are getting connected to the net everyday.  Airlines are starting to provide wifi (long overdue). There is a big fight over who will provide wifi in municipalities.  Broadband wifi/wimax is becoming more popular which will help combat the last mile problem.  Eventually, network connections will be ubiquitous, just as cell connections are today.  All we need is a reason to force the hand of providers. If computers had to access the network to do anything productive, that might do it.  However, I understand this "connection-required" approach isn't the only one--I could foresee some caching on the local system, but not to the extent where you run a local web server.  Maybe cache just enough to launch the desktop and some basic apps.

After you boot up (which would be extremely fast), the web browser connects to your "web desktop."  This could look like Windows, OS X, any other traditional desktop, or something radically different.  From here you could launch any of your "apps" from icons on the desktop or from a Start-like menu.  These apps run inside a browser shell and function very similar to traditional apps today.

The web desktop would be a commodity and should really be nothing more than a skin.  The value is in the apps.  We wouldn't have anymore cross-OS compatibility issues.  If an app is good enough, a vendor could charge for it.  Or perhaps the vendor would give it away and provide some value-added service on top for a fee. 

In this model, the GooOS or YahooOS doesn't exist.  Those vendors would provide applications you add to your web desktop, but they wouldn't actually provide an OS, which is a good thing.  It is important to decouple the OS (and desktop) from value-added functionality so we level the playing field and reduce dependency on a single vendor.

In order for this model to work, we'd need to create some basic web apps we don't have today to replace the traditional apps (office, productivity apps, etc.).  You can accomplish a lot with DHTML, Ajax, Flash, and Laszlo, but we may need to do some innovation with the browser to support additional UI controls.  The "web desktop" could be a supped up browser of sorts.  We'd also need a standardized web desktop spec that developers could create applications for.

The power users and hackers out there are undoubtedly scoffing at such a system.  That's fine because initially the web desktop would be most suitable to undemanding users.  Take my parents for example.  They turn their computer on, connect to the Internet, check email, browse the web, and use Microsoft Money.  That's it.  But they also have to deal with anti-virus and anti-spyware software and four different update packages that run periodically to check for patches.  To them, the web desktop would alleviate many of the headaches associated with "using a computer."  I could see this following the classic disruptive innovation pattern.  The web desktop may have a lot of problems initially that impede its use on a large scale, but those problems are solvable.  Over time it would become better and better to the point where it would be "good enough" for knowledge workers and continue to go up the food chain of computer users.

I personally would be grateful to have less dependency on a particular computer.  No longer would a hard drive crash mean you are out of action for several days.  You could access your full desktop from any computer in the world with a connection.  There wouldn't be anymore synchronization issues between your home computer and work computer or work computer and your mobile device.

I believe over the next ten years, the importance of the desktop operating system will continue to decrease.  Windows XP/Vista, Linux, and OS X, at least on the desktop (servers are a different story), will become obsolete or significantly less important.  That's a pretty bold prediction, but it isn't nearly as bold as when I started talking about it just two years ago.  With Web 2.0 gaining traction, it is just a matter of time.