Technology as a Search Problem

One of the ongoing interests in my research revolves around comprehending the nature of technological progress. This involves exploring how new technology emerges and evolves over time, and the various factors that influence it. I have examined this topic from multiple perspectives, including the operations and legacy of Bell Labs, the progression of specific technologies, and the frequency of multiple inventions along with the time it takes for new inventions to materialize.

I have recently formulated the idea that developing a new technology can be conceptualized as a search process. Essentially, a given technology can be viewed as a point within a high-dimensional "design space" or "technology space," which encompasses all conceivable technologies, encompassing every potential arrangement of components, materials, natural phenomena, and so forth.

Inventors must traverse this space, filtering out the ineffective combinations to discover the viable ones. This search process becomes challenging due to the sheer enormity of technology space, akin to enumerating the conceivable combinations of a significant number of elements. Economist Matt Clancy highlights that Wikipedia catalogued 272 named equations that illustrate various regularities in the world - from the Navier-Stokes equations describing fluid motion to the Exner equation detailing sediment deposition in rivers.

This implies an exponentially vast number of potential technologies that could be conceived. Clancy's rough estimation suggests that the number of possible combinations of these regularities is 2 raised to the power of 272, surpassing estimates of atoms in the observable universe. This colossal number, however, slightly underestimates the sheer volume of possible technologies because it considers technologies employing the same principles as indistinguishable, irrespective of their distinct manifestation or magnitude of application.

Comparatively, the most commonly cited estimate for atoms in the universe is approximately 10^80, or roughly 2^265. Furthermore, it's probable that there are significantly more regularities that could be harnessed by technology. Brian Arthur and Wolfgang Polak's paper, "The Evolution of Technology within a Simple Computer Model" supports this notion.

In their study, they discovered that within the universe of logic circuits fabricated from three fundamental components - the NAND gate and the values 1 and 0, the possible variations in circuits for a small number of inputs and outputs is immense. For instance, an 8-bit adder represents one among over 10^177,554 possible configurations with 16 inputs and 9 outputs.

Therefore, the search space for developing a novel technology is vast. The process of technological creation essentially entails gathering information to distinguish viable from non-viable options, thereby narrowing down the extensive range of technological possibilities and pinpointing a solution that fulfills a specific objective.

For example, in the 19th century, numerous individuals were fascinated by constructing a mechanical flying machine, but it remained unclear what exact constituents would result in a successful design. However, knowledge of the laws of physics and the existence of birds like the eagle narrowed down the search space considerably, providing a general idea of the essential components - lifting surfaces, propulsion system, and control mechanisms - and the potential arrangement thereof.

The Wright brothers' triumph can largely be attributed to their adeptness in gathering relevant information, enabling them to swiftly converge on a functioning airplane design compared to their contemporaries. My previous exploration into technological search involved examining Brian Arthur and Wolfgang Polak's simulation of evolving different logic circuits in search for workable technologies amidst an enormous sea of possibilities.

I appreciated the foundational simulation approach of randomly generating technologies to gauge the ease or difficulty of discovering workable ones. Subsequently, I aimed to employ simulation to delve deeper into the dynamics of technological search. Nonetheless, while Arthur utilized digital circuits - various arrangements of logic gates such as NAND, OR, AND, etc., capable of executing tasks like addition - I sought a more general model capable of encapsulating a wider range of technological evolution scenarios.