In the wake of Microsoft's early release of its patch for the WMF problem, lots of Windows users are unhappy. They complain that Microsoft's patch is designed for Windows XP and 2000, not Windows NT, ME, 98 or 95, even though those operating systems are also vulnerable and tens of millions of copies are still in use. Of course, we all know why Microsoft didn't patch those older Windows versions: they're obsolete.

Hewlett-Packard's midrange MPE operating system is obsolete too. Anyway, that's HP's story. MPE users, several thousand strong, believe MPE still works just fine. In December, HP agreed to extend support for those users for three more years.

Obsolescence is a much more slippery concept than it first appears.

That shouldn't be a big surprise. Vendors call something obsolete when they can no longer make money selling it. IT shops say the same thing isn't obsolete until we can no longer make money using it -- or maybe just until it no longer fits into our corporate IT architectures.

And end users, the people at their desks? Many of them believe a familiar IT system isn't obsolete until the pain involved in getting it to do what's needed is a lot greater than the pain of migrating to something new.

Still profitable, still useful, still bearable. Whose definition of obsolete is right? All of them. And they interact. When a vendor declares that a venerable product is going end-of-life, it's likely to be tagged for removal from corporate IT's plans. But users don't like that. Change means disruption. They push back against IT. IT pushes back against the vendor. And obsolescence gets redefined again.

Obviously, IT is in the middle. And we understand why vendors want to push out new products. It's about money: from upgrades, from customers captured from competitors, maybe even from a whole new customer set.

We have a little more trouble understanding why users are so willing to stay with aging products for so long. Sure, there's money involved. And OK, the old stuff is familiar and comfortable. But IT deals with change all the time. We learn new languages, master new systems and figure out new technologies on a regular basis. We accept that it's just part of the job. Why won't users?

Because for them, it's about pain, remember? And not just the existential pain of change. A new system means users will lose their habits -- the well-memorised procedures and tricks and shortcuts that let them do business while paying minimal attention to the technology. Users don't care that hardware or software is obsolete. But they do care when their habits hit the junk pile.

That's what hurts for them. That's why they'll put up with a lot before giving up some old systems as obsolete.

How can we help them redefine obsolescence? By paying attention to pain. Training won't be enough to get users fully up to speed on a new system -- we know that. But we can watch and listen. We can collect new tips and shortcuts and work-arounds from enthusiastic power users, and pass those along to other users. Everything we do to make the new system less painful will help.

And before that, everything we can do to make clear our pain with the old system will help too. Users don't see the cost and effort required to keep some older systems alive. And they won't know about it until we explain it to them. Just as we don't know how crippling the loss of their old habits is until we start paying attention.

That way, working together, we can decide what's obsolete -- and then move beyond it.

And dealing with vendors? That's much easier. Just as users are about pain, vendors are about money. To negotiate a new meaning for obsolescence with them, we'll have to show them some green -- eventually -- and they'll have to make it worth our while.

That's a part of the IT business that will never be obsolete.