Ideas in Space (3) - lost ants & spider monkeys

 In working with our clients, we developed the concept of "search patterns" within a landscape of ideas.  

We initially considered a spiral search pattern.  It's an approach that makes sense if you have an initial idea of where to start within a landscape and believe that what you're looking for is a finite distance from that starting point (e.g. where a lost and possibly wandering human was last located).  It's been identified in the behavior of ants looking for their nest, and used in robotics and diving training and other applications.




A spiral search pattern has a lot of interesting parallels with the "spiral model" of software development.  From wikipedia:

The spiral model is a software development process combining elements of both design and prototyping-in-stages, in an effort to combine advantages of top-down and bottom-up concepts. Also known as the spiral lifecycle model (or spiral development), it is a systems development method (SDM) used in information technology (IT). This model of development combines the features of the prototyping model and the waterfall model. The spiral model is intended for large, expensive and complicated projects.

One reason the spiral pattern seems useful in a software development methodology is that you're starting from a single point (the original design concept, which is likely to be fairly general or at least not completely optimized for the real world) and "searching" around that point to discover a more specific or optimized design.

But the spiral pattern is less useful if you don't know where to start within the landscape.  If you're really trying to transcend assumptions, shouldn't you be throwing out any preconceptions about where the best idea(s) are likely to be located?  Therefore a completely random pattern would make sense, something like the Brownian motion of a molecule within a gas.  Yes, but ... there's actually an interesting model that combines randomness with the ability to discover things about the landscape as you move through it.  It's called a Levy walk or Levy flight, after a French mathematician who defined the algorithm as a variation on Brownian motion.  Again from wikipedia:

Lévy flight, named after the French mathematician Paul Pierre Lévy, is a type of random walk in which the increments are distributed according to a "heavy-tailedprobability distribution. Specifically, the distribution used is a power law of the form y = x  where 1 < α < 3 and therefore has an infinite variance.

A Levy walk looks peculiar when mapped; basically there are a number of long "flights" through the space; each long flight tends to be followed by several short flights, sometimes called a "cluster".  It turns out that, like the ant example above, this distribution can be found in animal and human behavior.  It has been mapped to ocean predators or plankton feeders, spider monkeys searching for food in a forest, and in humans walking within a built environment, such as a city.

Intuitively, it seems like this might be a good pattern for looking for ideas within an "area of opportunity", a landscape which we think might be full of hidden opportunities.  It encourages us to jump to completely different locations, taking a long flight more or less at random, and then to look around that location to see if any of the surrounding ideas seem promising.  If not, we take flight again, to another part of the forest.