Long-term IT watchers feel much the same – this is supposed to be a fast-moving industry, but it's run out of steam. Has there been much on the desktop after windows, icons, mice and pointers came out in the early 1980s? Not so as you'd notice. Can you get more than a handful of hours on your laptop between charges? No: but once you could word-process on the move with a Tandy Model 100 for twenty-odd hours. Things have gone backwards since then.
Or take the 8086 architecture. At the time of its birth twenty-five years ago, it was a clever compromise that was somewhat compatible with older 8-bit chips but with some 16-bit functions. However, those compromises didn't age well – newer chips from Motorola and others were quickly introduced that were better in almost every way. By the mid 1980s, the 8086 was seen as a rather ugly device to write software for. Programmers often had to spend as much time worrying about how to work with the chip's idiosyncrasies as they did thinking about the useful bits of their software. Yet even by late 2004, the chances are overwhelming that you're reading these words on a computer that still speaks fluent 8086.
Everywhere you look, the old guard is holding up well. The big battle for mobile storage is between micro-miniaturised hard disks – Samsung and others are working on devices roughly the size of your middle fingernail – and flash memory. Neither of these is new technology: flash was invented in 1988 by Intel, while hard disk technology will be celebrating its 50th birthday next year. However, both remain at the cutting edge of storage technology – I've lost count of the number of exciting new memory ideas that have come and gone without leaving a trace. Bubble memory – Intel had a one-megabit part on the market alongside the 8086 – phase memory, ferromagnetic, holographic: all have either been and gone or are “the technology of the future, and always will be” in the industry cliché.
The dynamics of the market make it almost impossible to unseat a successful incumbent just by doing the same thing better. An established technology – even one with many flaws – has enormous momentum. Several fortunes have already been made by those that own the technology, and you can bet a Segway to an SUV that they'll be ready to spend several more to keep things that way. Then there's the experience behind that success: is it really true, as Intel likes to think, that plain old silicon can conquer optoelectronics, microwave engineering and nanotechnology? Perhaps, perhaps not, but that company's done so many impossible things with silicon in the past that sheer force of will should see it through. What chance for exotica with the potential for more, like gallium arsenide?
And then there's the four minute mile. If something is perceived as impossible, then it doesn't get done -- but as soon as someone demonstrates it once, the target is not only visible but achievable. Cathode ray tubes are enjoying something of a revival: despite a hundred years of evolution, they've picked up their game in the face of large LCDs by shrinking their forward-to-back size to fit into ever slimmer cabinets. It's very hard to do this safely if you've got a vacuum inside the glass envelope, but given enough impetus -- provided in this case by the commercial success of plasma and LCD screens -- the engineers managed. Because the rest of their design is so well understood and the production techniques so established, they cost a great deal less to make than equivalent LCDs while offering better performance and – the important bit – much higher margins.
Evolution only takes you so far, though: at the same time that the Wintel standard is comprehensively dominating the market, it's running out of places to go next. Both Microsoft and Intel are looking short of ideas and long on delayed launches. If you can't add anything the punters want, why will they buy?
So we should look elsewhere for our innovations. You want all-day portable computing? Instead of hanging around waiting for fuel cells to replace lithium in our laptops, expect breakthroughs in very low power portable clients that hook to powerful servers over ultra-fast wireless. That'll need new ideas in infrastructure and network management, not physics, but these are the areas most desperately in need of innovation – and most promising for revolutionary new business models.
Or look at open source – not for the software it provides, but for the new way it provides of thinking about intellectual property, perhaps even how we order our affairs for the greater good. Expect the old ideas to fight back, even to prevail for a while, but these new ideas aren't just replicating the old, they're making things possible that are otherwise impossible. That's the key for successful innovation -- and why there are no jet-packs while there are cheap seats on a Boeing 737.