8 Comments

Great article. I wonder though if we shouldn’t consider the short-term implications, namely (a) the tenets of intentional design of systems / processes / organizations that enable those combinations to be efficiently and effectively explored and (b) specifically the importance of a different paradigm for the exploration, which is neither individual humans or small groups or individual machines, but rather the large-scale network thereof. That’s the main tenet of the design of “augmented collective intelligence” I discuss at www.supermind.design based on the work that MIT’s Center for Collective Intelligence (and others) have done over the years. It is possible that we can push back the “efficient frontier” for quite a few more years.

Expand full comment

Another great article!

Reminded me of the book "The Nature of Technology", specifically an experiment from Brian Arthur and Wolfgang Polak described therein: to see if the system could evolve technologies—logic circuits—by combination from existing ones to fulfill niches on the list, and to study this evolution as it happens. The results were quite remarkable...worth the read

Expand full comment
author

Great book! I should read the article though.

Expand full comment
Feb 18, 2022·edited Feb 18, 2022

Arthur, W. Brian, and Wolfgang Polak. “The Evolution of Technology Within a Simple Computer Model.” Complexity, 11, 5. 2006.

https://doi.org/10.1002/cplx.20130

an example of "standing on the shoulders of Giants"

Expand full comment

If there is a normal distribution combinations in period 1, then you find a more productive combination or two, don't the new combinations re-set the distribution in period 2 and open up new combinations? How does that affect the distribution in period 2?

Expand full comment

Jones is a common name. Who is it?

Expand full comment
author

Sorry about that, the link to the paper seems to have gotten lost. It's Chad Jones, paper is here: https://web.stanford.edu/~chadj/weitzman_meets_kortum.pdf

Expand full comment

The combinatorial-explosion argument for slowdown in technological progress strikes me as quite odd because we use heuristics rather than exhaustive searches in these types of situations.

I suppose that an interesting way to rescue the implication that we'll eventually get overwhelmed is to suppose that our search-heuristics get exponentially better in pruning the space, but the search space grows factorially. If this were so, innovation would start out slow, get easier, and eventually become almost impossible (factorials grow faster than exponentials).

Expand full comment