Thursday, July 22, 2010

Thoughts on invention on reading Brian Arthur's book

Brian Arthur, in his book The Nature of Technology: What It Is and How It Evolves has a chapter on invention. In the chapter he is focusing on major inventions rather than the innovations that accrue in the course of improvements and extensions of existing technologies. Thus he is focusing on important inventions, not the "me too" things that are given so frequent patent protection.

He suggests that these inventions can arise from thinking about new ways to approach a given problem or from a focus on a phenomenon. I suppose that the latter case might be either focusing on ways to use a newly discovered phenomenon or new ways to use an existing phenomenon.

I am not an inventor in the sense of producing transformative inventions, but some nearly 50 years ago I was asked to investigate whether a laser could be used for measuring distance. The laser was a new technology at the time. In my report saying yes, I mentioned that if you pointed a laser at a target, a missile with the right optics could navigate to that target much as a ship can navigate to a lighthouse on a dark night. I suspect this is an example of a phenomenon (the laser) leading to an application (missile guidance).

I also developed a then new algorithm for design of neural networks with a colleague that was the result of spending time on the problem of how to better develop computers to recognize patterns.

It occurs to me that sometimes new techniques can arise when two people meet, one with a problem and one with a "phenomenon" that may solve that problem. I had lunch years ago with a colleague who was interested in the design of reverse osmosis plants. He had developed a computer program to predict the throughput of a plant based on a number of parameters. I was doing my doctorate in operations research at the time, and saw immediately how hill climbing techniques could be applied to the optimization of the parameters using my friends model. I traced the method for him on a napkin and he went off happily to write the program. The result was a major publication, which I hope contributed to the development of the technology. There is a similar story of J.W. Cooley and John Tukey meeting, with Cooley concerned with better ways of doing fast fourier transforms, and Tukey having a method which he wrote out on the back of an envelope. The result was the Colley-Tukey algorithm.

As a manager of a research program I reviewed a great many project proposals that sought to find a phenomenon to solve a specific problem. Many of these focused on finding an appropriate diagnostic reagent or vaccine.

I am somewhat concerned with Arthur's differentiation of science and technology. For example, Penzias and Wilson discovered the Background Cosmic Radiation using technology developed to receive signals from a satellite. They discovered the background radiation as they were seeking to eliminate noise from their instrument to make it more effective as a detector. Where was the dividing line between the science (since the background cosmic radiation is a key discovery in cosmology) and the technology?

In my small way I discovered approaches to measure the information in pattern recognition systems leading to new statistical models and a better understanding of the errors made by people in classifications.

This is one of a series of postings occasioned by Arthur's book:

No comments: