Hi Dan B.; I think that computers are going to continue to evolve, and will continue to eat up larger bandwidth, and become more useful to more and more people as well.
On the other hand, I do know a surprisingly large number of people who are getting by with '486 or earlier hardware. I hang out with a lot of engineers, and being techno geeks, we throw our machines out after a year or two, and replace them with new metal.
But my ex got by fine with a '486 until the floppy broke. I refused to fix it until she bought a new machine. (I didn't want to her to call up the next time the 6-year old machine broke down.) So she got a decent $500 machine. She used her old machine for doing word processing, the new one might get hooked up to the net one of these days, but there is no way that she will ever be a computer game or internet addict like me.
The guy downstairs is busy writing the next great American children's novel (he also writes pornographic humor), and he is running on a '386. Park Place Motors, which is probably the ritziest car lot in Bellevue (they sell a lot to the Microsoft wealthy) uses a '286 and an ancient dot-matrix printer. I asked them when they were going to upgrade, and they said never. Y2K should be interesting... All these people use their computers for word processing, mostly.
My point is that computers do word processing. That's all that's needed from them. They've been doing an adequate job for 20 years. The real question is not whether or not computers are going to substantially improve in the future. The real question is will the public be willing to pay the premium required to get this year's technology. Or will they instead be inclined to save a few bucks and get slightly obsolete technology.
Calculator technology did not halt in its tracks 20 years ago. Modern calculators use less power, have more memory (i.e. enough to keep your phone list, for instance), have better displays, etc. There is no question that calculator technology continued to advance.
But what happened is that the public became largely satisfied with a vanilla calculator, rather than one on the cutting edge of technology. Consequently, the ASP for calculators dropped to the most efficient firm's cost of building them. That drove the companies that were not vertically integrated out of the business. TXN and HWP always made their own chips, and after all the calculator assemblers went out of business, TXN and HWP were still in production.
When a technology is new, it always underfulfills its consumers desires. As time goes on, it begins to fill a niche, and as that niche gets filled, the consumer becomes more and more content with the previous year's product. As this happens, more and more consumers choose to buy at the low end.
It all boils down to the age-old question of "what does the consumer want," or more realistically, "what will the consumer want in the year 2010." Currently, the trend is clear. Average Sales Prices continue to drop.
Sales are complicated things. There is no way of proving one's case with regard to what people will want to buy. (Case in point: Beanie Babies.) What I have argued (for 2 years now) is that from an engineering standpoint, these past 2 years, and the next 4 years or so have seen and will continue to see a substantial and unavoidable decrease in the marginal increase in performance for computers as the price is increased.
In other words, the price/performance curve now has a gap. Cheaper computers are giving substantially more bang for the buck compared to dearer computers, as compared to the ratio of 2 years ago. That performance/price gap is moving up through the performance curve, and will hit the server market in 2 years or so. The effect on the server sellers will be similar to the recent effects on the low-end box makers (which effects are only about half-way through.)
That decrease in marginal bang for the buck is going to suck most consumers into the low end, though who knows, maybe people like spending $2000 for a machine that will largely be obsolete in 24 months, and gives well under 300% more performance than a $500 machine. The low end is where the performance bargains are, not at the high end. This will be also the case in servers, starting in about 2 years.
-- Carl |
