Adam Timlett
fractalagility is changing to naturalsearch
Updated: Sep 18, 2021
The Fractal Agility website is now using a new url 'www.naturalsearch.org' and the content will soon be changing to match this domain name.
The concept of 'fractal agility' has served me well as an umbrella term for my research, which originally was focused on organisational economics as well as biology. However, I have now reached the end of this initial exploratory stage and in the next few weeks I will be presenting the results of my research by unveiling a new way of looking at the concept of 'natural search'. I am currently in the process of writing an academic paper on natural search and next year will also begin writing a book on this subject which will summarise and extend the ideas I have worked on for the last few years.
Natural search and natural selection
While 'natural selection' is, of course, a very well known example of 'natural search' mathematically, it has long been unclear what sort of thing 'natural selection' is. For instance, over the last few decades computer scientists have pioneered the concept of 'genetic algorithms', but it is not clear how much such algorithms really have in common with natural selection. One major difference is that such algorithms are usually 'objective functions' trained against a mechanism to score them for their fitness. As the computer scientist Kenneth Stanley argues, it is less clear whether natural selection itself is fundamentally an 'objective function' since it is subject to constraints, but is never 'finished'. In a sense, it has also already acheived its 'objective' before it starts, as, by definition, the existing species must have already survived and reproduced. I believe that Kenneth Stanley's work on non-objective functions such as 'novelty seeking' functions points toward a new way to understand a broader class of search.
At the same time, the computer biologist Andreas Wagner has been working on a theory of how innovation is possible via evolutionary forces. He calls this 'innovability' and links the capacity to evolve to robustness of organism phenotypes to changes in genotype. I feel that this work is deeply connected to Stanley's work, though I'm not sure that they are aware of each other's work.
More recently, researchers have started to investigate the ability to simulate what they call 'natural search' by microorganisms. They define this simply as their search for 'targets' such as food, for instance, and seek to understand how that integrates with models of marine ecology, as one example. However, this definition could be broadened to include many more types of natural search, including natural selection.
Natural search as akin to natural experiments
We normally think of a 'search' as something deliberate, where we have a target in mind. But what if natural search is analogous to a 'natural experiment' in which the search can happen without any deliberate organisation, agency or even an objective? In this case many examples of natural search arise. One example is when a tweet goes viral and explores many combinations of tweet and context in which the tweet is viewed without any deliberative search by any agents. Even random lightning strikes can be considered a search without a given 'target', as even though there is no agency, many lightning strike 'targets' are naturally 'explored'. The key to such examples are that many states are explored, and that this exploration must be viable. Many forms of complexity such as Conway's 'Game of Life' can also be studied as a form of viable state exploration within the rules of the system. So the Game of Life can be said to also exhibit a search of a large number of possible distinct system states which occurs as a natural by-product of the simple system rules.
I will use naturalsearch.org to present my own ideas by leveraging this diverse research and drawing it together to redefine what we think of as 'natural search'. I will show that 'natural search' is a class of 'search' which is broader than 'objective search' and so different to typical genetic algorithms. However, 'natural search' encompasses 'natural selection'. I will go on to explain how the concept of natural search may then also help us to understand natural selection in a new light and how it is also relevant to other areas of biology. Eventually, I will also seek to show in more detail how various economic phenomena are also related to the concept of natural search. I will seek to link together economic phenomena and biological phenomena in a new way to better account for the way that innovation occurs in an economy, for example.
I hope you follow along and can participate in the talks and discussions that I plan to have alongside the paper and writing and blogging.
In the meantime, here are the three main pieces of research/books by others that I will be using to piece together my arguments which you might want to check out:
References
The Arrival of the Fittest. Solving Evolution's Greatest Puzzle
by Andreas Wagner
Why Greatness Cannot be Planned. The Myth of the Objective
by Kenneth O. Stanley and Joel Lehman
Hein, Andrew M., et al. "Natural search algorithms as a bridge between organisms, evolution, and ecology." Proceedings of the National Academy of Sciences 113.34 (2016): 9413-9420.