Wired on Friday: The educationalist Alan Kay sits behind an ultraslim laptop and begins the same old talk. This time, he's giving the keynote at Silicon Valley's "Emerging Technology" conference

He starts by gently ribbing the audience of developers and entrepreneurs. Nothing has happened in the last 30 years of computing, he says. From his point of view, that's true. He's seen most of it before, back when he was inventing it.

To rub the point home, he shows his audience flickering black-and-white movies from the 1960s and early 1970s.

Ivan Sutherland's SketchPad program, which ran on a machine as big as a room in 1963, looks better than many drawing programs today. Demonstration videos from Mr Kay's team in the mid-1970s at Xerox Parc show them using the first mouse, the first laser printer and perhaps the first webcam to collaborate long-distance.

Part of this conference will consist of hackers trying to re-learn how to write this "social software" for today's PCs.

What's the delay? Mr Kay blames the grown-ups. It's hard to believe, now that computers are a multi-billion dollar industry, that the windowing systems they use (and the way they are programmed) was originally planned to allow children to use computers. Learning from Mr Kay's thoughts on "KiddieKomp" - computing for children, his team at Xerox Parc invented the windowing environment (windows, icons, mouse and pointers) 25 years ago. Even the first laptop - rarely seen outside the business world - was plotted out by Mr Kay as an educational tool, the DynaBook.

In trying to make computers easy enough for a child to use, Mr Kay designed a system easy enough for a businessman to make money with. But the task he set himself was far tougher than that. He didn't want his children to be passive users of the PC. He wanted them to be its masters: programming complex tasks as well as the highest-level modern programmer.

This was his second innovation; one more hidden away from everyday computer users, but achingly familiar to all programmers. Mr Kay was one of the father's of object-oriented programming, or OOP, a way of thinking about how to control a computer that lets coders create programs more quickly and easily than in the 1960s and 1970s.

Again, it was envisaged as a way of breaking down computers into a model that anyone could manage. Mr Kay envisaged all the different powers of computers as divisible into separate, simpler, objects. Objects could be copied, or built into bigger, more complex objects.

You didn't have to know how an object worked - you could just treat it as a thing. You don't need to know what is in a file (or indeed how it was created), to copy it onto a floppy or move it onto another computer.

More subtly, you could create an object that represented a tax bill. You could have another that behaved as a background accountant which could find out your tax details by asking the tax bill. Programs became less like long lists of complex instructions, and more like a conversation between objects.

This is how programmers create most programs used today. But like most analogies OOP has broken down. Eventually, programmers have to stop treating the computer as a bag full of imaginary objects, and start remembering that it's a pile of real electronics. And while the user sees much that has Mr Kay's OOPish inheritance on screen, the object nature of behind the scenes is mostly hidden.

But not in Mr Kay's world. Since inventing the modern desktop, and kick-starting the OOP revolution with his programming language SmallTalk, he has carried on his work. He has continued to design systems where everything - numbers, keyboards, users, even the computer itself - is represented as an object, all of which can be taken apart, tinkered with, and then fitted back together as needed. And not just by expert programmers, but by anyone using the system: mostly children. Mr Kay's work has continued in an almost unbroken line since his Xerox Parc work, being sponsored at different times by Atari, Apple, Disney and now via a non-profit supported by many of the people whose industry Mr Kay inspired. The strangely atomic world has not made such an easy transition from the world of children to that of grown-ups. A revolutionary feature of Mr Kay's work is that it obliterates the idea of a separate software application.

Smalltalk systems, and the current environment he works in now, called Squeak, don't have Word, Excel or Access. Smalltalk has, instead, Goodies: additions to the core system that tweak and transform everything - sometimes digging deep into the objects at the heart of the computer and changing the very idea of how the machine works.

This idea is powerful; perhaps a little too powerful. In the beginning of Smalltalk, such universal flexibility was seen coming at too great a cost to speed and memory-usage to survive the transfer to the first commercial windowing systems. Later, the power to fiddle under the bonnet of the home PC was reserved to one software maker - the operating system manufacturer. And Apple and Microsoft have defended that right and the profits that came from it.

There is also the point to be made that businessmen don't "want" to tinker. Providing so much openness can easily lead to the whole system falling apart. Too much freedom on your PC might lead to endless tweaking, and endless inconsistencies between machines.

But perhaps Mr Kay's time has come, once again. His latest, patient, teasing of the industry he spawned ended with a spectacular demonstration, aimed at exciting his classroom of Silicon Valley geeks. A 3D environment, called OpenCroquet, is his new attempt to push past the desktop he envisaged. Croquet appears as a three-dimensional world, filled with tinkerable objects. This world is shared, real time, over the internet. Other users appear - as objects, naturally: in the demo Mr Kay was a rabbit, his colleague Alice. Some problems that gnawed at creators of previous generation of computing vanish in Croquet. Machines are now fast enough to cope with Mr Kay's object-driven world (he showed Croquet on a laptop smaller than his original Dynabook). Security is being considered (kids don't always want to share all their toys, after all).

And for business? Well, you couldn't sell someone an object instead of an application in Smalltalk without them downloading it onto their machine, tearing it apart and reusing it. But on the Croquet model, you can keep your object safe on your machine, and charge others to converse with it. They call that Web Services, and it's hyped everywhere in the Valley these days as being the next big thing.

But then, most next big things turn out to be Mr Kay's last little project.