Microsoft Windows: Is It Time For a "Do Over"?

[Jason Dunn]

<img hspace="5" align="left" alt="" src="http://images.thoughtsmedia.com/dht/2008/vista-logo.jpg" />Digital Home Thoughts is a <a target="_blank" href="http://windowshelp.microsoft.com/windows/en-us/community/allcommunities.mspx">Featured Community</a>, and has a part of that, one of the things I enjoy the most is getting to interact with leaders from other technology sites. In our private forum we were discussing the issue of whether or not Microsoft is out of touch with its customer base; what the customer needs, what they want. From a consumer point of view, I think there's definitely some truth to that. I think that Windows XP was the &quot;right&quot; operating system for consumers at the time, but we're now six years later and Windows Vista is having some trouble getting accepted - even by the geeks that normally flock to anything new and shiny. There's a lot of <a target="_blank" href="http://en.wikipedia.org/wiki/Fear,_uncertainty_and_doubt">FUD</a> out there about Windows Vista, but even if I toss out 80% of what I hear from others, the remaining 20% is enough to make me realize that Windows Vista simply didn't deliver the way it was supposed to. Windows Vista is fundamentally broken.<br /><br />The more I think about it, the more I believe with every brain cell I have (hey, no laughing!) that Microsoft needs to do what Apple did with OS X: create a new OS from scratch. Create something for the modern era, something designed for a constantly connected era. The OS we have today in Vista is still one created based on thinking from 15+ years ago. Whether it's the registry, drivers, codecs, DLLs, or the way programs are installed, Windows Vista has the same root problems that Windows 95 had. Microsoft needs to start over, to build something fast, lean, stable, and secure. They'd have to implement a purely virtualized XP/Vista layer for compatibility, because one of the reasons Windows is so big and slow is because there's so much legacy code in there for compatibility (what about <a href="http://www.svsdownloads.com/" target="_blank">thin-slice virtualization</a>?). Microsoft needs to create something beautiful and easy to use, but more than that, they need to fix the problems that Windows has at it's core. <MORE /><br /><br />A drastic step? Yeah, absolutely. Impossible? Maybe - Microsoft is a big company and big companies are hard to turn around. But mark my words, Apple is going to keep pushing ahead and people are going to keep switching. Microsoft's current strategy of relying on Moore's law to save them is no longer working. Throwing hardware at speed problems works to a certain degree, but hard drives aren't getting much faster, nor are CPUs (they're moving to multiple cores instead). Sure, RAM is cheap, but going from 2 GB to 4 GB of RAM doesn't do a heck of a lot to speed up the overall performance of Windows. Faster hardware might make things happen faster, but it doesn't fix most of the problems that people have today with their computers.<br /><br />Faster hardware doesn't solve codec and DLL problems that are deeply rooted in the way Windows is architected. Why is it that installing a new video editing program has a good chance of breaking the other video editing program I already had installed? It's the shared codecs and DLLs - it's the way Windows works, and it's fundamentally flawed. The concept of shared DLLs dates back to the era when hard drive space was at a premium and it made sense to have multiple applications all use the same DLL. When is the last time you ran out of hard drive space? We no longer live in that era. Applications crash too easily, and while often it's the fault of sloppy programming, the operating system has a lot to do with that - the way Windows applications stomp all over each other means headaches for users.<br /><br />And remember all of this is coming from a guy who really likes Vista, uses it on all his computers, and defends it from the slings and arrows of Mac users all the time. And if I see something wrong with the direction Windows is going on, there's a serious problem.<br /><br /><em>Jason Dunn owns and operates <a target="_blank" href="http://www.thoughtsmedia.com">Thoughts Media Inc.</a>, a company dedicated to creating the best in online communities. He enjoys <a target="_blank" href="http://photos.jasondunn.com">photography</a>, mobile devices, <a target="_blank" href="http://www.jasondunn.com">blogging</a>, digital media content creation/editing, and pretty much all technology. He lives in Calgary, Alberta, Canada with his lovely wife, and his sometimes obedient dog.<br /></em>

[micaels]

Jason,
I've been running Vista since the first days and I really enjoy it. Have switched over all my computers to it. I've just started getting into hosted exchange and i've installed it onto my N95 8GB without any trouble at all, but when I try and add it into outlook it says that i've to go to control panel and do it through the email configuration. So I think okay leat's do it there then but what do I discover that email configuration tells me that I need to configure it into outlook directly. Im just going into circles here.

[Stinger]

Apple managed to bring OSX to the table in a relatively short space of time because they based it on BSD. I doubt Microsoft would ever consider using BSD or other Unix-like OS as its core.

I've also heard horror stories about the legacy code within Windows. Horrible spaghetti code that hardly anyone understands.

Starting from scratch, from the kernel level up might take 10 years of development. Who knows where technology might be in 2018? Maybe the operating system or even the PC will be irrelevant by then.

[Janak Parekh]

Starting from scratch, from the kernel level up might take 10 years of development. Who knows where technology might be in 2018? Maybe the operating system or even the PC will be irrelevant by then. The OS will almost definitely not be irrelevant by then. If anything, embedded OS usage is going up -- witness CE, OS X Touch, and Linux successes. To me, that's even a bigger call that maybe Microsoft should start a new tack. Take the NT/XP/Vista kernel (or perhaps one of their Embedded platforms wholesale) and write a new OS around it. Make sure it's scalable up and down to a broad array of devices, and go from there.

And, if I understand correctly, Microsoft's doing exactly some of this stuff (http://en.wikipedia.org/wiki/Windows_7#MinWin) for Windows 7, which will be very interesting to see. (Although I don't think they're starting from scratch...)

For those of you who have some developer background, there's an interesting set of articles brewing over at Ars Technica (part 1 (http://arstechnica.com/articles/culture/what-microsoft-could-learn-from-apple.ars), part 2 (http://arstechnica.com/articles/culture/microsoft-learn-from-apple-II.ars)) about the cruft problem in greater detail.

--janak

[EscapePod]

I too have been using Vista, since RC1 and beyond. With the exception of a few test bench PCs, and one ubuntu model, all others are Vista w/SP1 now. I love this OS. Any minute issues I had with drivers in the early stages (nVidia SLI was beta for a loooonngg time), have been cleared. I may never buy another HP printer or scanner again, but that's another story -- the one about equipment and software companies that didn't want to pony up the cost to certify their product with Vista.

That brings up the one issue I have with the DLLs. MS surely "wanted" to prevent DLL conflicts, but somehow cannot find the right code to prevent it, or, cannot get the software and equipment manufacturers to pay attention. If they "certify" conflicts are more easily prevented.

There's no doubt that sometime, MS will have to create a new OS from scratch. I like the idea mentioned above regarding using virtual machines to accomodate the legacy programs (and hopefully drivers, too). Something built-in and transparent to the enduser would be awesome.

[Janak Parekh]

I too have been using Vista, since RC1 and beyond. With the exception of a few test bench PCs, and one ubuntu model, all others are Vista w/SP1 now. I love this OS. Any minute issues I had with drivers in the early stages (nVidia SLI was beta for a loooonngg time), have been cleared. I may never buy another HP printer or scanner again, but that's another story -- the one about equipment and software companies that didn't want to pony up the cost to certify their product with Vista. I do think Microsoft should be harder on the vendors who don't bother getting their drivers up to speed faster. Microsoft's gaining a bad rep for allowing these crappy drivers to promulgate, and it isn't entirely their fault.

That brings up the one issue I have with the DLLs. MS surely "wanted" to prevent DLL conflicts, but somehow cannot find the right code to prevent it, or, cannot get the software and equipment manufacturers to pay attention. If they "certify" conflicts are more easily prevented. Yeah, I think you're right. DLLs are so embedded as a system design that without a ground-up scratch redesign, it's tricky. Certification is a slow and expensive process and may not scale for all things. However, I think this also feeds into Jason's argument about starting anew.

There's no doubt that sometime, MS will have to create a new OS from scratch. I like the idea mentioned above regarding using virtual machines to accomodate the legacy programs (and hopefully drivers, too). Something built-in and transparent to the enduser would be awesome. Mac OS Classic is a great model for this. The biggest challenge, though, is ensuring that existing business distributed enterprise apps still run. I suspect that's one reason Microsoft's held up. Windows Server 2008 is supposed to eventually have more virtualization--maybe we'll see that filter down to newer Windows workstation releases.

--janak