In Louisiana, there is a bridge on Interstate 55 that crosses the Manchac Swamp. One of the world’s longest bridges, it’s over 22 miles long.

I mention that because I think it’s important that you be visualizing a very long bridge, because that is, in so many ways what Microsoft is building with Windows 8. Win8 on the legacy (x86/x64) architectures and Win8 on ARM are in many ways two different platforms, and WinRT is the (very long) bridge that Microsoft is trying to build between them to help Windows on ARM succeed.

When discussing the potential of Windows 8, especially the opportunity for developers, Microsoft touts numbers such as “450 million” or “500 million” as a potential market for Windows 8. But is the immediate potential really that large?

A key strength that has repeatedly sold Windows upgrades for more than two decades is application compatibility. That is the premise that last year’s (or last decade’s) applications will still run on next year’s version of Windows.

This has traditionally been especially important when discussing complex, custom-built line-of-busines applications. Enterprises expect to keep running the same applications with as little (as close to no) modification as possible on a new version of Windows. It is important to not dismiss application compatibility, because no version of Windows that didn’t have flawless application compatibility has ever been commercially successful, as I’ve said a couple of times before. The Intel Itanium architecture took a while to deliver the performance value proposition that was initially promised, and – very importantly – had a very, very bad Windows application compatibility story for existing applications throughout it’s entire life. Conversely, the AMD/Intel x64 architecture delivered exceptional performance relatively early on, at a marginal cost premium over x86, with almost no application compatibility cost. True, drivers were an issue, and led to Windows XP x64 having very little uptake, but Windows Vista largely resolved this, and Windows 7 all but removed any hurdles of running x64 Windows. Today you can buy an x64 system and, unless it has esoteric (or crappy) piece of hardware or software, ancient 16-bit applications, or VB 6 applications, you don’t pay a price to run Windows natively on x64 hardware.

Let’s look at a couple of facts about x86 PCs running 32-bit Windows 7:

1. You can run almost any legacy application, going back over 20 years.
2. You can use ancient hardware peripherals, as long as you have a relatively current driver.
3. You can use ancient utility or security software, as long as you have a relatively current driver.
4. The majority (almost all) applications available to you today are not optimized for touch, or even for stylus-based input.
5. The majority (almost all) are either a desktop PC or a laptop. Very few are tablets.
6. The processors in the system are designed with an emphasis of performance over efficiency.
7. The applications (especially legacy applications) on the system give very little consideration to power management, efficiency, or not running unnecessary, inefficient, or repetitive/looping background tasks.

In addition, the odds are that if you looked at almost any PC in use today, it also doesn’t have a digitizer – which means it doesn’t support touch or a stylus. Some do, it’s true. But the cost has left these devices to be an edge of the market, far from mainstream.

If you imagine the same x86 PC upgraded Windows 8, every single fact above remains true, though the new world of Metro applications begins to help by providing a touch-first, well-managed, power-sensitive application framework (but does so on a PC without touch in most cases).

Microsoft’s take is that every x86 PC that can run Windows 7 is an upgrade candidate to Windows 8. That’s absolutely true.

But let’s look at what would happen if you were to buy a brand new Windows 8 device with an ARM processor. How many of these facts still hold true?

1. You can’t run any legacy applications. None. You might be able to modify them to run on ARM if Microsoft provides some level of tooling and libraries they depend upon are available, but as of right now, that doesn’t appear to be something Microsoft wants you to do. Visual Basic 6 applications, in particular, are forbidden under the terms of Windows 8 application certification.
2. You can’t use any hardware peripherals, unless a new driver (with driver signing) has been created (same problem originally faced by x64).
3. You can’t use any utility or security software, unless a new driver (with driver signing) has been created (same problem originally faced by x64).
4. All applications (unless Microsoft renegs and provides a Win32 application migration story) are Metro apps, provided through the Windows App Store. These are inherently optimized for touch, but work with pen. These can be frustrating to use with a mouse.
5. The majority are likely to be tablets, very few are likely to be desktop PCs or even conventional laptops or “Ultrabooks” until Windows on ARM has proven it’s viability.
6. The processors in the system are designed with an emphasis of efficiency over performance.
7. The applications must conform to the WinRT framework, and as a result are very constrained as to what they can do in the background (for better or worse), but are generally efficient, secure, and tightly managed.

Before I go any further, I’d like you to try 2 levels of Cut the Rope in Internet Explorer, or 2 levels of Angry Birds in Google Chrome; but I want you to do me a favor. If you have a stylus or a touch-enabled system, don’t use those. Use a mouse. Don’t cheat – just use the mouse, like most current PC users would have to. Ready? Go on, I’ll wait.

You’re back? Great. Now, if you’ve ever played Angry Birds or Cut the Rope on iOS or Android, I want you to think for a moment about the experience you just had.

Was it fun? Sure.
Was it novel? Meh – you’ve probably played these both on other platforms. That’s why Microsoft and Google paid to have them ported to the browser – it’s important that well-known titles like these be on their respective platforms to begin with.
Was it the best experience you’ve ever had with these games? Probably not. Besides some performance compromises to go from native compiled code to HTML/JS, both of these games compromise the gameplay because you need to use a mouse, not the more intuitive touch user interface that both apps succeeded because of.

Now, you might be saying, “But Windows 8 will be touch-based! People will have tablets!”

That is the plan – and this version of Cut The Rope will probably be a great experience on a touch-enabled Windows 8 tablet, regardless of CPU architecture. But reconsider that 450M-500M number mentioned earlier. Almost every x86 PC in existence included in that number does not have touch.

Any Windows 8 PC can run the Metro user interface. Absolutely. But I can tell you from my personal experience that Metro and the new Start page are very frustrating to use with a mouse instead of touch. Frankly, they’re rather difficult to use. Personally I’ve been contemplating whether Kinect for Windows could help here – it may. That aside, while consumers and enterprises may upgrade to Windows 8 in time, I have to think that there may be some initial hesitance to upgrade, especially on those exact “legacy” systems touted because the user interface is such a fundamental shift – even if existing Win32 applications “just work” on Windows 8 x86 systems as they did on Windows 7.

I worked on Windows XP. I saw enterprise customers become emotional – become angry about the new Start Menu in Windows XP. It had to have an option to disable it and turn theming off. People have become used to it, but the shift to the Start page and Metro are far more jarring than a new Start Menu or Windows theming were.

Now you might be thinking, “Well, fine. Consumers will buy new tablets, largely with ARM, but possibly with x86, processors, optimized for touch.”

But then we have the opposite problem. Early on in this post, I highlighted how significant of a role Windows legacy application compatibility plays in businesses deciding to upgrade to a new version of Windows. Windows 8 on x86 and x64 continue to offer this value proposition, thus indeed resulting in a potential for Windows 8 uptake on many existing systems (that don’t support touch). But honestly, x86-based tablets are a longshot to remove the iPad’s lead, due to the iPad’s name recognition, performance, value prop, battery life, weight, and low cost. Windows 8 on ARM, conversely, throws away the entire “legacy application” value proposition, but begins to counter the iPad (the key details then being the build quality, performance, weight, battery life, and cost of Windows 8 ARM-based tablets – all of which are completely unknown at this time).

Let me summarize:

1. Legacy PCs running Windows 8 will run legacy applications, but deliver a suboptimal experience when running touch-centric Metro/WinRT-based applications.
2. New x86 tablets running Windows 8 will run legacy applications, will deliver the optimal experience when running Metro/WinRT-based applications, but may compromise battery life and/or weight and/or cost over an ARM tablet. This isn’t a guarantee, but at this point, this is the way product offerings have panned out in the past.
3. New ARM tablets running Windows 8 will not run any legacy applications, but will deliver the optimal experience when running Metro/WinRT-based applications, and will quite likely deliver the best weight/efficiency/cost.

I’ve said before that Windows 8 will live two lives. It’s critical to understand this. Desktops/laptops running Win32 and some WinRT applications, and tablets, running mostly or all WinRT applications.

I’m not saying the quoted 450M/500M number is completely incorrect. I’m also not saying that Windows 8 won’t succeed. I hope it does.

I am saying that I believe it may be a false syllogism to link the current market entrenchment of Windows 7 and earlier on x86/x64 desktops and laptops running Win32 applications, to any guaranteed explosion of sales for Win8 tablets which require an entirely new generation of WinRT-based applications.

The notion that “Lots of people run Windows today, therefore Windows 8 will succeed on ARM tablets and outsell the iPad” or “Lots of people run Windows today, therefore Metro application authors are guaranteed to make money” are both false syllogisms to just assume at this point.The current – gigantic – installed base of Windows x86/x64 systems running Win32 applications does not, by definition, mean that you can assume an immediate, or definite, success for Windows 8 on ARM running WinRT applications. The iPad is young, and it’s anybody’s game.

People do own PCs, and many are likely to continue to buy Windows 8 PCs or Ultrabooks. However without touch, they are, I personally believe, not the best candidates to run Metro. Conversely, consumers can buy Win8 Metro/touch-focused tablets when available, but then will abandon legacy Win32 apps.

We’ve yet to see what the stimuli are that lead consumers to prefer a Windows 8 tablet over an iPad (Apps? Form factors? Performance?). Microsoft may well have some secret sauce here that we’re not yet privy to, but as of this point, they are running silent and running deep, and not discussing Windows 8 go-to-market strategy at all.

Correction/clarification (1/18/2012, 11:00 AM) I have had several readers and followers on Twitter note that the 400M+ number specifically speaks to the current run rate of PCs continuing to sell as it has – not the installed base at all. Stupid on my part to incorrectly state that. However that actually only accentuates the issue above, as that run rate needs to continue or grow, while also including iPad competitive ARM and low-powered x86 tablets in the sales mix.