I'm at Microsoft's MIX conference in Las Vegas, where the big news has been the unveiling of the Windows Phone 7 development platform, and the platform preview of Internet Explorer 9 with extensive HTML 5 support.

There is more to say about both topics; but one thing I want to highlight is that IE9 will not work on Windows XP, the venerable release that will not die.

It is not only that XP remains in use on existing PCs; there are also new machines on sale with this old version of Windows. I've just purchased a netbook, and when making my selection I noticed that many of them still come with XP, usually the Home edition. Some companies still specify XP when buying new PCs, to avoid the compatibility hassles that come with a move to Vista or Windows 7. There's also doubt over the benefits of upgrading. A friend said to me recently, "I really like XP, it does everything I need. What is the point of moving?"

Here's what IE General Manager Dean Hachamovitch told me:

Building a modern browser requires a modern operating system. There are facilities in Windows Vista and Windows 7 around security, for example the integrity level work that gave us protected mode, there are performance improvements that enable a variety of things in the browser, there is graphics infrastructure to take advantage of the GPU, that doesn’t exist in previous operating systems.

A measure of scepticism about such comments is reasonable. Microsoft wants users to buy its latest stuff; there's no surprise there.

Nevertheless, Hachamovitch is right. Windows 7 is more secure and more powerful. Personally I find it easier to use as well, partly thanks to Microsoft's design work on the user interface, and partly because desktop composition in Windows Aero enables richer preview of minimised applications and other good things.

Microsoft is also making a statement with IE9, to the effect that XP users can no longer expect to be included in major product releases. Office 2010 does support Windows XP; but you can bet that the next one will not.

The long life of XP is a side-effect of one specific thing, which is the failure (relatively speaking) of Windows Vista. I used Vista from its first release and regard it as better than its reputation suggests; but nevertheless, it was greedy for hardware resources, prone to annoying slow-downs, and less polished overall that it should have been.

There is also a real issue with application compatibility, introduced with Windows Vista, mainly thanks to User Account Control. This feature protects access to system locations such as the Windows and Progam Files folders, and parts of the Windows Registry, causing problems for some applications that expect full access.

These factors extended the life of Windows XP, resulting in the situation we have today: an operating system coming up to nine years old still in widespread use.

Very often the continuing use of XP is not something we can control. Rather, it is something we have to live with. Nevertheless, it is becoming a liability. Windows 7 improves on XP in every way that I can think of; and we even have XP Mode and Med-V to assist with migration, by running obstinate apps in virtual instances of XP.

Windows XP is something we have to live with, but no longer something to recommend.

Postscript: arriving at the gate for my flight from Las Vegas, I could not resist snapping what seemed the perfect illustration: Windows XP with a sad little error message.

 

There is a kind of Windows 7 fever sweeping the Web right now. Some observers are getting carried away:

As a person who performs almost every computing task on a Mac and tells anyone who will listen that at this point, the average consumer should be using a Mac instead of a Windows machine because of security and usability, I'm starting to prep myself for the single moment that I thought would never come: I'll be using a Windows 7 machine as my main computer and telling anyone who will listen that, believe it or not, using the latest Microsoft operating system really is worth it. [Don Reisinger]

I've spent a lot of time in Windows 7 these last few weeks, and in some ways I see what he means. It has a smoothness and elegance that is lacking in Vista, though it is there to some extent in Server 2008. But let's be clear: the popular view that Vista is rubbish and Windows 7 is great does not stand up to close analysis, for one simple reason. It is just not sufficiently different. Microsoft actually makes a positive of this characteristic, explaining how preserving the core architecture of Vista ensures good application and driver compatibility and therefore a smooth upgrade.

What about Windows 7 versus the Mac? I agree that Windows 7 is superficially more Mac-like; and more important, that the factors that made Vista such an unpleasant experience for some (not all) early adopters will not apply. I'm thinking of things like laptops that were underspecified, or had buggy drivers, or barely adequate graphics hardware, or that were weighed down with so much foistware that Windows would hardly run. However, other factors that make users prefer Macs will not have changed. There is the control Apple exerts over both hardware and software; its design excellence; the fact that Apple is less tolerant of legacy software, whereas Microsoft has Windows jump through hoops to keep it running,adding complexity and inconsistency; and there is the absence on the Mac of the culture of chaos that afflicts Windows. I'll leave aside religious arguments about Unix vs Windows, though it is a matter of record that OS X has to date proven far more secure, for whatever reason.

Let's ask some awkward questions. Will a substantial number of Windows 7 machines fall victim to viruses, worms, trojans and botnets? Almost certainly. Will Windows 7 from time to time flummox users with obscure errors like "the system could not find the file specified"? Almost certainly. Will some Windows 7 machines be built down to a price and be sold with obvious design flaws and insufficent attention to quality? Almost certainly. Will some vendors wreck the sweet install experience Microsoft has created by imposing their own clunky utilities and third-party trialware on top? Almost certainly. Will the Windows 7 event log become populated with perplexing entries like "The Workstation service terminated with the following error: The redirector is in use and cannot be unloaded." (plucked from my Windows 7 beta 1 laptop)? Almost certainly.

Let me add that I realise how smooth and reliable a well-managed Windows machine can be. Further, while I have had some frustrations with Vista it has always been stable for me and I have never wanted to go back to XP. Vista, then, is better than its reputation; Windows 7 is better than Vista but will have trouble living up to the proclamations made by its more enthusiastic admirers.

The recent Linux plumbers conference included a session on getting Linux to boot in 5 seconds (see also the write-up here). It was great to see the report, because performance gets far too little attention. Most of the business world runs Windows rather than Linux, at least on the desktop, and in most respects Windows seems slower than Linux on the same hardware. I would give anything to have Vista boot in 5 seconds on my laptop. In fact, the main problem with Windows Vista is not driver compatibility, or annoying security prompts; it's that little spinning bagel that appears only too often. When I start Microsoft Outlook 2007, I brace myself for an extended pause while it starts up, during which the whole system becomes unresponsive. By contrast, I still enjoy using an ancient version of Paint Shop Pro for working with images, even though I have Adobe PhotoShop installed, because it starts in a blink and does exactly what I need.

The old joke is that what Intel giveth, Microsoft taketh away; but it is not a joke any more. Time spent waiting while a computer boots, or reboots to apply an update, or sits there doing who-knows-what in one of those sulky pauses, is time when we could be getting on with our work. Those delays cost real money, every day. They are also aggravating, sometimes not only for the immediate user. If I am on the telephone, for example, Outlook's slowness is not only a problem for me, but also for someone else trying to arrange a meeting.

I suspect that Microsoft made two wrong assumptions. First, that hardware improvements would remove performance issues. Fast hardware does mitigates problems, but really only disguises rather than fixes slow code. Further, the popularity of mini-laptops and other low-power devices means that fast hardware cannot be assumed. Second, Microsoft intended Vista to be always on, relying on sleep and resume instead of fast start-up. Unfortunately sleep and resume tends to be unreliable, and many Windows updates still require a system restart.

The same factors apply to web sites and web applications. One of the things that keeps people going back to Google is the consistently excellent performance of its site, especially in search. I am convinced that this has a subliminal impact, and that we instinctively prefer the sites that are more responsive. PHP inventor Rasmus Lerdorf speaks frequently about performance - last August at Drupalcon, for example - and is intolerant of slow applications; it would be great if more industry leaders shared his attitude.

There is little we can do about Windows boot times - well, apart from happy hours spent messing with msconfig, the System Configuration Utility - but that does not apply to our own development work. A project is not really done until the performance is satisfactory, even if all the other features work as specified. Performance is easy to measure, and there are plenty of profiling tools that show up bottlenecks in the code; tools like Rational PurifyPlus, Intel's VTune, or for .NET Redgate ANTS; there are also free tools available. The trick is to focus attention on what is too slow, rather than wasting time on what is already fast enough. It is rewarding work, since applications that perform well are a delight to use. Performance goes hand-in-hand with design, about which I posted a couple of weeks ago; they are both essential parts of the user experience, and a good user experience means a high level of satisfaction. There's nothing better for keeping customers coming back for more.

"Developers, Developers, Developers" is the famous rant of Microsoft's Steve Ballmer. Maybe he picked the wrong word. I'm just back from the Remix 08 conference in Brighton, where the company's Principal Researcher Bill Buxton spoke on software design. He made a big impression, especially when he suggested in an interview that unless Microsoft changes its culture in a way that put design at the core of its software, the company would die.

Underlying this is the ascendancy of Apple, widely admired for its design, and the mixed fortunes of Windows Vista, which was meant to have great design but in practice has plenty of irritations many of which are design issues. There is a tendency to think of design as primarily about graphics and layout, but this view is too narrow. Design is about the user's experience, and encompasses all kinds of user interaction.

Few companies have the resources of Apple or Microsoft; but raised expectations of what constitutes good or even acceptable design affect the whole industry, including developers of internal applications as well as web site authors. Design failures are frustrating for users; and the consequence is that software is used less, and satisfaction is lower.

An example which comes to mind has no graphics at all. A certain telecom company with a colourful name has a telephone menu system which is unavoidable if you need to call support. Whenever I have to use it, I feel my blood pressure rising before dialling the number. I know I will have to go through numerous layers before being allowed to speak to a human. I know I will be given mandatory choices, none of which apply to me, forcing me to lie. I know each step will be ponderous, with a recorded messages I have heard before, and invitations to use the web site instead - if I thought that would do, why would I call? If you call outside working hours, you are taken through all the steps and told right at the end that the office is closed, goodbye. When I do eventually speak to someone, I am already in a bad temper.

This is the consequence of poor design. I could reel off countless further examples, such as Microsoft Outlook 2007, with its obscure dialogs - click Message Options to show email headers, very clear NOT - its habit of blaming the user for its own shortcomings, "The data file was not closed properly," and its slow performance with large mailboxes. Some of these are technical issues, but because they directly damage the user experience, they are also design problems.

Enough ranting. The interesting question is how developers and designers can work together to improve software design at every level. Buxton talks about continuous design, where designers participate in the development process throughout, rather than only at the beginning of a project, or even worse, only at the end. It is a good thought; but as the panel discussions at Remix demonstrated, there is no consensus about how the developer/designer relationship should work.

There are signs of change. Tools are appearing which have both developers and designers in mind, such as Microsoft's Expression Blend which can open Visual Studio solutions, or Adobe's product codename Thermo, which converts artwork into application code. That said, it is people rather than tools that hold the key. It is about developers ceding some of their power to designers; and about designers learning more about the constraints under which developers work; and about learning to treat poor user experience as a defect. It is not easy; but teams that get this right will gain a substantial commercial advantage.