07 August 2007

The Art of Technology Prediction

It is relatively easy to make some relatively straightforward predictions about particular technologies with considerable accuracy.

We can predict that next generation computers will have more powerful microprocessors and the future memory devices will store more with quick recall times. We can predict what is possible with unmanned aircraft in thie military. We can predict winners and losers in some plausible peak oil scenarios. We can predict with considerable detail the impact of melting icecaps that raise sea levels in various amounts -- something devastating to Miami, Florida, but not directly relevant at all in Denver, Colorado. We can predict likely new classes of drugs to treat particular classes of illnesses.

The really hard part, is predicting how unrelated developments influence the choices people make and the technology that they end up using in the bigger picture.

Predicting that digital cameras would improve was a no brainer. Predicting that digital cameras would squeeze film cameras out of the market to a great extent was far harder to predict. that kind of prediction requires a particular kind of consumer level analysis of all relevant choices along with solid estimates of what retail prices are possible for particular services.

Will next generation MP3 players kill compact disks? Quite possibly, but it is a much trickier analysis.

The possibility of digitizing books is long standing, but the niches where this has become common place, like academic journals and legal reference works, was far harder to prediction. And the Google led effort to put the full text of the vast majority of the public domain online has dramatically impacted the future of the online book industry.

It is one thing to point out advances in car technology. It is another to be accurate in predicting the big winners in different modes like rail, road, and air. This is particularly true because methods of transportation like airplanes are as much impacted in flight times, and hence trip value, by security and billing constraints as by technological limitations. Similarly, the winner in the race between diesel, ethanol, gasoline, hybrid-electric, fuel cell, natural gas, hydrogen, and other conceivable motor vehicle engine systems is economically crucial, but hard to pick a winner for. And, how do changes in urban planning figure into the mix? In two decades we will have a new model, but what it will start to look like is hard to tell, in part because technology, far more so than science, is path dependent. When more than one technology was possible at one point, the one chosen is often largely a historical artifact with few technological features that explain it, particularly when compatibility is an issue.

1 comment:

Anonymous said...

The problem with technology predictions is that they're often economic predictions in disguise. High-definition television, for example, is an obvious incremental improvement, but it's up against the inertia of over-the-air broadcasts on the one hand, and the HD-DVD / Blu-ray format war on the other.

DVD is approximately 480 lines of resolution. High definition is 720 to 1,080 lines. The technology already exists to handle 2,000 lines (although storing and distributing it is still tricky), but I can't tell you if it will be commonplace in five years or twenty.