Posted by andy Tue, 23 Aug 2005 14:09:00 GMT

Houston, we have a problem. Innovation in aerospace stalled out and started losing altitude around 1970. Instead of jet packs and vacations on Mars, we have foam falling off the 30-year old space shuttle. Our life today isn’t anything like the twenty-first century life we imagined in …

Houston, we have a problem. Innovation in aerospace stalled out and started losing altitude around 1970. Instead of jet packs and vacations on Mars, we have foam falling off the 30-year old space shuttle. Our life today isn’t anything like the twenty-first century life we imagined in 1970, extrapolating from the previous 35 years (1935-1970) of rapid change. It’s clear that innovation has slowed in many areas of industrial technology that are important for our daily lives.

This fact is going to be highlighted in a rumored article from Jonathan Huebner. You have probably already heard about it. Huebner counted major technological innovations, and found that the rate of innovation is declining, and the rate of innovation per person is rapidly declining. By this count, the rate of per-capita innovation peaked in 1873, and has now declined to the level that it was at in 1600.

Many do not agree. Clearly, a lot of discovery, both scientific and technological, is going on in the world. There have been a lot of rebuttals.

I personally believe that innovation is a fractal process. Each burst of innovation is likely to have an S curve of output (a bell curve of innovation rates, as noted by Huebner). So, any given space of innovation is bounded and runs out of steam. However, the bursts can be of any size, and start at unpredictable times. We are certainly coming to the end of the industrial revolution, but we may soon start an evolution revolution. Innovation has dropped off in aerospace, but it’s increasing in biotechnology. We need to keep changing our focus and placing our bets.

Evidence that innovation accelerates and decelerates comes from the oldest record of innovation we have- biological evolution. There was a huge burst of evolution / innovation at the time of the “Cambrian explosion” about 570 million years ago. This then tapered off into a very different kind of incremental evolution.

Whether you believe that the overall rate of innovation is decreasing, the fact that the rate of innovation does decrease and has decreased in many major industries has a practical applications for investors. We’re all investing in technology somehow – through our day job, through our education, or for some of us, through substantial portfolios of financial assets.

If you can figure out what industries will have increasing innovation, you can buy a diversified selection of participants, watch them grow and consolidate, and make a tremendous profit from the winners. If you catch one of the really big, Kondratieff-style, 50-year waves like oil, electricity, or computing, you can surf that wave for a lifetime.

What if we could figure out the difference between industries with increasing returns to innovation, and industries with decreasing returns to innovation? What if there were some essential indicators that would show the direction – increasing productivity driven by innovation, or decreasing productivity in spite of innovation?

I think these indicators exist.

Let’s consider one obvious factor – an increasing or decreasing size of the space of target solutions. You might be working on a problem for which each successful solution reduces the number of possible new solutions. This seems to be the case for drug discovery. There are a limited number of “targets” in a human body, and each time you find one, you reduce the number left. This leads to declining productivity of innovation.

On the other hand, you might be working on a problem for which a successful solution increases the number of possible solutions, or even the number of related problems, each with an increasing number of solutions. This is true if, instead of improving the health or efficiency of an existing system, you are actually constructing new things that have new uses. Each of those new uses opens up new opportunities for innovation. You make a microprocessor chip to go in a calculator, and then someone puts it in a personal computer, a telephone, etc. If you can use the personal computer to design new chips, you get another round of increasing returns.

The size of the solution space controls the return on innovation, even if innovation is rapid. I think of the impact of computing power on video games. The early video games, such as pong and space invaders, were fun. Modern video games have thousands of times more speed, memory, and resolution, and are only a little bit more fun. Computing capacity has diminishing returns if measured in terms of fun. This observation goes a long way toward derailing Ray Kurzweil’s hypothesis that, because computing power is rapidly increasing, we will all one day live in a computer-supported nirvana. That particular technology, no matter how good, has limits of applicability for humans

This is the problem with drug discovery. It doesn’t matter how much innovation gets thrown into the drug discover process. Vast amounts of new technology and ingenuity are being brought to bear. Yet, the drug discovery process gets less productive and more expensive every year.

Looking at biotechnology, we need to separate the dynamics of the inputs from the dynamics of the outputs. The input, biotechnology, is experiencing tremendous innovation, probably with increasing returns. The output, drug discovery, is experiencing little innovation and decreasing returns.

What about software? I think we have reached a point of saturation for most of the kinds of software that I work on. It’s like video games. Most of it is designed to improve the efficiency of humans and businesses, which gives it declining marginal returns. On the other hand, there are sure to be new domains where software essentially interacts with software. For instance, if we can make evolved software (genetic programming) we can apply that process to itself and get increasing returns.

By this measure, drug discovery is a loser. However, biotechnologies which actually construct new things, such as GMO’s, or synthetic biology, are likely to have increasing returns. I’m placing my bets with synthetic biology.