It's therefore tempting to predict that PCs in 2020 will have 16 or 32 processor cores, but I don't think so. First, we have 40 years of software that hasn't been written to exploit multi-core processors, so users will not get the benefits they might expect. Second, the extra transistor budget in a 2020s chips will be used for other things besides CPU cores. Intel has already added what used to be separate maths co-processors, cache memory and graphics chips to the CPU, and will continue to work towards "system on a chip" integration. Third, programs that need repetitive supercomputer-style operations will farm them out to graphics processing units (GPUs) instead of relying on the CPU. Fourth, most ordinary users don't actually need anything more than a quad-core system. I'd love to think we'll all be using 16-core PCs in 2020, but I suspect not. The trend is not towards raw performance, it's towards performance per watt.
I expect many of you disagree with my inclusion of a Blu-ray drive in a desktop PC, and I'm dubious myself. It's certainly true that, one day, chips will replace rotating disks as a way of storing data. However, we've been saying that since 1982, when I first used a Grid Compass 1101 laptop with "bubble memory". It still hasn't happened. However, in 2020, a Blu-ray disc should be a fast, cheap and reliable way of delivering or backing up 100GB of data. This will appeal to the few billion people who still won't have the high-speed broadband or ubiquitous Wi-Fi that people already expect in major Western cities. As author William Gibson is said to have remarked, "the future is unevenly distributed".
Either way, I've watched computer storage media progress through cassette tapes, floppy discs, CDs and DVDs to Blu-ray discs, and formats rarely disappear as quickly as people expect.
The problem is that while PC specifications can change rapidly, people don't. The market is ultimately governed not by what's technically possible but by what ordinary people want to do with their machines, and how much they're willing to pay.
The PC I'm using now has much more power than the one I had running Windows Me/2000 in 2000 or running Microsoft Windows 3 in 1990, but it's still doing the same things: word processing, spreadsheets, email and messaging, games. Of course, it's doing them faster, in higher resolution, with added video, but the basic operations remain the same. Revolutionary developments such as touch screens, virtual reality, augmented reality, artificial intelligence, speech recognition and "mind control" -- all of which I tried in the 1980s -- still haven't changed the PC world, though touch screens have become popular on smartphones.
In the end, my relatively modest expectations for the desktop PC's progress reflect my feeling that the technologies that really change things tend to appear on new devices. That was true of the Nintendo Game Boy when it was launched in Japan in 1989, the Palm handheld organiser introduced in 1997, the Nintendo Wii from 2006, and this year's Apple iPad. Today's desktop PC hasn't really changed that much since the IBM PC set the standard in 1981, it's just that we now have lots of limited-use devices for different purposes, instead of one general-purpose PC.