http://arstechnica.com/articles/culture/modular-windows-will-suck.ars wrote:Modules and the next version of Windows
There is a growing consensus of opinion forming that Windows "Seven" will be "modular," the concept being that you buy the core OS first and then add to it individual "modules" with logically distinct units of functionality. There are two ways the OS could be modularized in such a fashion, the first being that it could be split into functional "roles," such as "music" or "movies" or "mail & chat." The other option, which is a bit more radical, would be to build on the "Windows Live" software that updates/replaces some of the OS components. For example, Vista's Windows Photo Gallery is replaceable with Windows Live Photo Gallery. The Live version is similar in concept, but includes greater online integration and features.
With both Windows Vista and Windows Server 2008, we can see the first few steps in this modular direction, albeit in different ways. Windows Server 2008 has as one of its major features the idea of "roles". Rather than installing everything and the kitchen sink, with 2008, you install the base OS and then choose one or more roles, such as Active Directory domain controller, Web Server, or Print Server, and the software components to support those tasks are installed accordingly.
This "roles" tack has resulted in the now infamous multiplicity of versions of Windows Vista. If you want Media Center, you need to get Home Premium; if you want hard drive encryption, you need to get Enterprise; if you want Aero Glass, you mustn't get Home Basic; if you want everything, you have go get Ultimate.
The differences in capabilities among Vista's versions are fairly arbitrary; there isn't actually anything particularly business-y about the full-system image-based backup, for example, but it's not found in either of the Home versions. This annoyance is mitigated somewhat by the fact that Vista also made it easy to change between versions. Unlike the days of XP Home and XP Professional, where reinstalling from scratch was the only way to switch between variants, Vista has Anytime Upgrade, which allows someone with, say, Home Basic to upgrade to Ultimate relatively seamlessly.Why go modular?
There are a few motivations behind the modular approach that appears to be Windows' future. With Server 2008, the rationale is that by not installing unwanted services, there is less (potentially exploitable) code running on the server, and so the server is more secure. Similarly, by offering components as "Live" tools, there are fewer antitrust concerns (they're no longer bundled with the OS, so Microsoft can't be exploiting its desktop monopoly).
But the most compelling reason—for Microsoft, at least—is an economic one. For a corporation with business-critical Windows-only custom applications, Windows is worth more than it is to a home user who just does a bit of e-mail and web surfing. The home user isn't really depending on Windows. It may be what he's used to, but if push comes to shove, he could always try one of those fruity computers with the funny TV ads that seem to be so trendy. Many businesses, on the other hand, need to run custom applications; if they can't, they're not doing business, and the cost of redeveloping the applications is huge. The upshot of this is that a business would pay more for its copies of Windows than the home user would because the Windows licenses are more important to the business than to the home user.
With only a single version of Windows, however, everyone has to pay the same price. This means that the price has to be relatively low, because Microsoft doesn't want to scare off the home users to whom Windows is worth relatively little.
Unfortunately for Microsoft, this means that business users can get away paying less for Windows than it's actually worth to them. They need to pay only the cheap home user-friendly price, even though they'd be willing to pay more than that if they had to. To Microsoft, this represents lost revenue; "lost revenue" that arises from people buying things for less than the highest price they would be willing to pay is called by economists the "consumer surplus."
If each customer could be charged what they were willing to pay rather than an amount that's identical for every user then Microsoft would get more revenue. Normally, though, it's quite difficult for companies selling commodity products to adjust pricing in this manner. Companies don't negotiate a price with each customer; they (or "the market") set a single price that everyone pays, and they just have to hope that it's the price that will give them the best trade-off between units sold and revenue per unit.Recouping the consumer surplus
For the accountants and economists, selling multiple versions of the same software is a partial solution to the "problem" of the consumer surplus. Instead of selling one version of Windows at one price, two versions are sold—one for home users and one for businesses—for two different prices. As long as there's some minor feature that makes the business version a must-have for businesses but is irrelevant to home users, the corporate buyers will shell out for the pricier software, and the home users can stick with their cheaper home version. This discriminatory pricing allows Microsoft to increase revenues by extracting some of the revenue that would otherwise be "lost" to the consumer surplus.
Windows XP was the first example of this; XP Home and XP Professional were, for most purposes, identical. But XP Home couldn't join a Windows domain and so was unsuited to typical business usage. Vista took this idea further still. Home Premium and Business more or less correspond to the old XP Home and XP Professional, respectively, but Vista also added Home Basic—for the very cheapest, lowest-value systems—and Vista Ultimate for the geeks and technologists who've gotta have every feature of the OS, whether they'll use them or not.
Because Vista has more versions, there is more need for the products to be differentiated to justify the different prices. What's more, because Vista attempts to break relatively uniform markets into smaller groups (XP Home is replaced by Vista Home Basic, Home Premium, and Ultimate, while XP Professional is replaced by both Business and Enterprise), that differentiation became a whole lot harder.
At the moment, the differentiations between the Vista versions are pretty arbitrary. If you pick Home Premium over Home Basic, you get a lot of extra stuff: DVD playback, Media Center, Aero Glass. There's no way to buy only one of those extra features; it's all or nothing. This tends to limit the price differentials between versions. If you have Home Basic and only want a couple of the extra features that Home Premium has, Microsoft still needs to price the Home Premium upgrade so that it's worth your while to pay the extra cash and get the better version. If the pricing gap is too big, only users who want everything that Home Premium has to offer will bother with it.
Modularization of the OS could go a long way to solving this pricing problem. Modularization provides product differentiation in spades. Instead of having to go all the way from Home Basic to Home Premium to get the features you want, what if you could choose to buy, for example, just the DVD support and scheduled backups?
This approach could be quite versatile; perhaps you have Home Basic, but you want the DVD support from Home Premium and the imaging-based backup of Business. At the moment, to get both of those features, you'd have to go all the way for Vista Ultimate, and that's a big price difference. With modular Windows, you could choose just those features and pay a fraction of the price. This is not to say that Microsoft would necessarily allow something this fine-grained, but it's a logical extension of what we have now. In practice, I would expect to see something like a handful of different roles (which would each contain a dozen features, say) that could be purchased and installed independently of each other. Stepping stone to subscriptions
If these different functional modules were further tied to Live-branded software, they would also be ideally positioned for a subscription model. While people may balk at subscribing to "core" OS features, if Windows Photo Gallery were coupled with online photo hosting and backup, it might be much more palatable. Rather than one-off payments to add a module to Windows, there might instead be a stream of monthly payments.
How far this could be taken is hard to say. There are places where the tie-ins are obvious, but many more where they are less clear-cut, especially given the decades of expectations that have grown around consumer software. Still, one could envisage a future where all the "extra" features (compared to Home Basic) of the Business or Enterprise editions were available for a monthly fee rather than a flat up-front payment, so businesses would buy a very cheap "core" OS and then pay a few bucks each month for domain membership, central management, drive encryption, and so on and so forth.Not such a good idea
So, modular Windows in the Vista style is good for the accountants. By allowing people to pay closer to what the OS is worth to them individually (rather than the traditional one-price-fits-all), the consumer surplus can be partially recouped by Microsoft, which is good for the bottom line. It's also good for those wanting to push subscription software, in that it provides lots of neat, independent units of functionality that can be subscribe to individually. It might even seem like it's good for the users. Instead of having to buy a much more expensive version of Windows, users can buy just the features they actually want, especially if the pricing is kept roughly in line with what it is today. But I actually believe that such a modular approach would be totally devastating to the platform.
Don't get me wrong: I have nothing against optional components in the Windows Server 2008 mold. As long as I can just pop in the DVD to enable something I didn't install, I have no problem. In fact, this might even be a good thing. I'm not altogether convinced of "surface area" arguments for the desktop OS, but it can't hurt to leave out something that you absolutely positively never use.
But I don't think Windows Server 2008 represents the direction that Microsoft is going for its desktop OS, or at least, it is not their primary goal for the presumed modularization. I think the approach taken by Windows Vista is a much better indication of the shape of things to come, and I think that is bad for consumers, and bad for Windows itself. I think that Microsoft will use the different modules as product line (and hence price) differentiators. XP and Vista alike tried to segment the market in order to extract more money from it; XP split it into two, and Vista split each of those two into two (with Ultimate kind of spanning all four options). Finer segmentation, courtesy of modularization, has to be seen as the next logical step by Microsoft's beancounters. Spinning these modules off into subscribable units would then take this to another level.Screw the customer
In real (inflation-adjusted) dollars, the price of Windows has tended to come down over the years, even as the operating system's capabilities have grown. Microsoft is forced to keep the price of the different Vista versions reasonably similar to ensure that they're within the limits that a customer is willing to spend to upgrade. This means that there is still considerable consumer surplus going untapped.
But modular Windows will give Microsoft the ability to push up the overall price of the software. The process starts with making the individual units smaller, so that they can be priced individually. Instead of buying, say, Aero Glass plus domain membership plus encryption plus better backup tools plus Previous Versions (Home Basic to Business, and yes, I do realize that there are other differences) for $150 at a notional $30 per feature, you might buy only domain membership and Previous Versions for $40 each. For the person who only wants those features that's a net win, $80 instead of $150. But for someone who's after the entire upgrade, it would now be $200. Or how about domain membership and backup for $15 each a year?
There is a case to be made that such a practice would be as frustrating for users because it is confusing. Many OEMs offer a choice of Vista versions on their hardware, and it serves as yet another technobabble decision that users aren't really well-equipped to answer. What exactly are the repercussions of getting Home Premium on a new PC as compared to Ultimate? Although the information does exist, it's not easy to find and it makes it far from clear which version you should get. This is compounded by the poor division between "home" and "corporate" brandings, with certain features desirable to both groups found in the "wrong" versions.
Greater modularization would tend to make these problems worse. Even more offensive than the existence of higher-priced "does it all" versions of Windows would be the perception that Microsoft was nickel-and-diming Windows users, charging more for every little feature. That feeling already exists, and it rankles. The more fine-grained it is, the worse it feels; the classic example of how much this sucks would be Apple's QuickTime Pro upgrade for QuickTime Player. Until fairly recently, such basic features as "full-screen playback" required a paid-for upgrade to the basic free software. Although that particular feature is now free, many others are still paid-only, and this does nothing but inspire ill will and venom for the company. The very real risk of platform fragmentation
These problems might be relatively minor and manageable, though, if the pricing is right. If users feel that they're getting a bargain, they'll probably overlook all manner of price gouging. The bigger problem, as I see it, is one that is a little more insidious.
The issue is that modularization strikes a blow against the very concept of a platform. When a software developer writes a program for Windows XP, they more or less know what they're going to get. They know what Windows XP does out of the box, and although some parts might be optional, someone can install everything from the XP CD if they want without paying a penny extra. This gives developers a nice big target to hit. And that's the point of a software platform; it's a big set of functionality that you can depend on—it's just there automatically, ready to spring into life as soon as your applications call it.
With a modularized Windows, that could fly right out the Window. And to an extent, it already has, at least for some kinds of application, particularly "media" applications. I might want to write, say, a video management program to help people catalog and organize their video files and DVDs. It would be natural for such a program to slot into Media Center, since it already has some pretensions in that area. Except, not every version of Vista is eligible for Media Center. It's only in Home Premium and Ultimate. So either I have to change my plans and target only the common core features found in all the variants, or I have to persuade people that my program is so good that it's worth not only buying my program, but also buying the necessary Windows components too. That might be okay if my program costs $1,000 and the necessary Windows feature costs another $20, because the marginal cost is very small. But if my program costs $20, well, "modular Windows" has just doubled its price, and that doesn't make me happy. Some might argue that the consumer saved that $20 on the initial purchase of Windows (by not buying things they didn't need at the time), but consumers won't see it like that, because the consumer never feels that "savings"; it's all rolled into the cost of their new computer.
Subscriptions may make this even more unappealing. In spite of copyright law to the contrary, there is a general perception that, like the purchase of a computer, the purchase of an OS is an investment, and when you buy your OS, you own it. Technicalities of "licensing" aside, that's the perception, and people act accordingly. Subscriptions—where it is very plain that you do not "own" the software—are not desirable. Most consumers expect software to be theirs in perpetuity when invest their cash, so I think switching to a subscription model will, for home users at least, go down like a lead balloon. Corporate markets, with their preference for regular payments rather than big lump sums, might feel otherwise, of course.
Fragmenting the platform has other repercussions, like making it harder for Microsoft to sell Windows. Vista has proven to be a hard sell, and consumer satisfaction is low. But a few rough edges notwithstanding, Windows Vista is actually really good OS. At least, it is in its Ultimate incarnation. This is the edition that everyone should be using, because its combination of features is great. The lesser editions are very peculiar because of what they lack. Turning off potential buyers
For an example of how Microsoft's first steps in the modularization direction are already hurting the sale of consumer Windows versions, consider my favorite Vista feature, Previous Versions. Previous Versions aren't new as such; Windows Server 2003 also has them. But Vista is the first OS to bring them to the desktop. With Previous Versions, Windows makes snapshots of all the files on my hard disk every so often. If I accidentally delete or overwrite a file, I can browse an earlier snapshot and retrieve it. It's like having a complete backup made of my hard disk made every few hours that I can instantly restore any files from.
I love Previous Versions. They've saved me from a number of cock-ups; typically hitting delete on the wrong directory or saving over documents when I meant to "Save As". The user interface is a bit clunky, but nonetheless it's a killer feature. The appeal is obvious; when even TV ads are using the worry of deleted photos to sell on-line backup services, it's clear that this kind of accidental destruction scenario is something that rings true with a lot of people, especially with the proliferation of digital photography and the like—we keep a lot of valuable memories on our PCs. So on the face of it, it sounds like it ought to be the kind of thing that home users should love. The appeal is obvious, and universal.
Unfortunately, for whatever reason, the powers that be at Microsoft have deemed Previous Versions to be a "business" feature. Home Basic and Home Premium don't have them. So, Previous Versions, something that should be easy to sell ("it lets you get back your deleted files"), that anyone would want, aren't available to the people who would most benefit from them (after all, home users don't have some nice centrally managed and backed-up file server to put their data on). So in turn, Microsoft can't use Previous Versions as a marketing point, because the people that the company would be marketing to won't get Previous Versions in their Windows. And similarly, when home users ask their knowledgeable cousin/neighbor's kid/etc. if they should get this Vista thing they heard about, instead of being told "Yes, it's got these neat features," they'll hear "Don't bother, it's just a slower version of XP."
In chasing the consumer surplus by performing a limited modularization of Windows Vista, Microsoft has kneecapped what should have been its flagship product, limiting killer features in the name of market segmentation. It has alienated its users, even—or maybe even especially—those users who would normally be early adopters, who would pioneer migration to a new version and be its number one cheerleaders. And it has diluted the strength of its platform by making it something less than a single coherent target for software developers. Neither of MS's competitors (Apple and the Linux distros) is taking an approach like this. Neither of these segment their markets in anything approaching the same way. People might point to Linux distributions' installers as evidence of being "modular", but Linux has never had the same conceptual strength (as a targetable software platform) as Windows, and this is part of the reason why. And as Linux software tends to be free, the cost/consumer surplus issue doesn't arise either. The model may work for Linux, but that doesn't make it good for Windows.
Modularization—and the discriminatory pricing it permits—might appeal to accountants and economists. But it is bad for consumers, bad for Windows, and ultimately, bad for Microsoft. A modularized Windows, or worse still, a modularized subscription-based Windows, undermines the purpose and value of the Windows OS. If it comes to pass, it will surely sound the death knell of the entire Windows platform.It doesn't have to be all bad
Microsoft has plenty of smart employees, and plenty of them no doubt realize that the kind of market segmentation enforced by Vista damages Windows a great deal. Indeed, we know from the e-mails revealed as part of the ongoing "Vista Capable" suit that "Vista Home Basic" was not wanted by many within the company, precisely because it diluted the Vista platform and branding. Unfortunately for both Microsoft and the consumer, the people making the decisions either didn't understand or just plain didn't care. But Microsoft must have noticed the lukewarm response to Vista, the weakened marketing, and the corporate indifference, they certainly can't help but have noticed the Vista Capable debacle.
Upper management prevailed with Vista, and they're prevailing with Windows Server 2008 and SQL Server 2008. Although both of these product lines have used various market segmentation techniques in the past, with the 2008 iterations they're getting that bit more diversified, and thus a bit more fragmented. But in an ideal world, Microsoft would recognize the folly of its ways and make Windows Seven a more coherent platform. Well, scratch that; in an ideal world Microsoft would, at the very least, scrap Vista Home Basic and cut the price of Home Premium, but to do that would be to admit failure. So in a plausible world, Windows Seven would see a reduction in the number of versions available. Even three versions—home, business, and "everything"—would be a marked improvement over the current situation.
With a less money-grabbing approach to the base platform, useful things could come from modularization. Going beyond mere optional components (as in Windows Server 2008's roles), we could see Microsoft spin off components that are "built-in" with Vista. The clumsy "Windows Live" branding is used for a number of programs that effectively replace built-in Vista features; for example, Windows Live Photo Gallery replaces Windows Photo Gallery, and Windows Live Mail replaces Windows Mail. If Windows were to be totally divested of these components, in favor of the Windows Live counterparts, it could be good for both Microsoft and consumers alike.
The advantage to Microsoft (as previously mentioned) is that free, downloadable add-ons aren't subject to the same anti-trust concerns as bundled features. This would permit the applets that are currently part of Windows to grow into more capable applications, without fear of inconvenient anti-trust lawsuits. It would also divorce the applications from the Windows lifecycle. Instead of having to wait three years for the mail client to get new features bundled in with a new Windows release, it could happen within the much shorter release cycle of a standalone mail client. So Microsoft could keep their OS "current" between OS releases by shipping new and improved versions of the applets, and users would get more free stuff to play with, which would be a win all round.
A precedent for this kind of thing has been set by Apple. The iLife suite (iPhoto for photo management, iMovie for home movie creation, iDVD for DVD mastering, etc.) can be updated independently of Mac OS X, and many of the iLife applications are very highly regarded (especially the GarageBand music program). The real trick that Apple pulls off is to bundle iLife with new Macs, and then charge $79 for upgrades. Monetizing updates in this way is obviously appealing, but probably not something Microsoft can do so readily.
An approach like this would, far from damaging the platform, allow Microsoft to make it a more attractive prospect for its customers. Making the ancillary Windows components interchangeable and replaceable with high quality replacements, thus restricting modularity to non-core OS features, would make Windows a better environment with better software. The only question is whether the managers who make the decisions will go for this approach rather than a continuation down the miserable path that Vista has already started on.