Gliders

Why do we do the things we do? Why do bureaucracies not do things? Why do large companies do things three years too late? These are the questions that we seek to answer through emergence.

Emergence

Emergence is the process in which lower-level objects, with their own rules and behaviors, interact with one another to unknowingly create an abstracted, higher level pattern.

In his book Creation: Life and How to Make It, Steve Grand, creator of the 1993 computer game Creatures, discusses how he used emergence to create bottom-up digital organisms. Grand argues that the top-down approach (e.g. brute force search,) which had heretofore dominated the AI field, could never represent real intelligence; a computer beats a rabbit in chess, but when thrown in a pond, which swims to safety?

The fallacy inherent in early AI techniques, Grand states, is that they are “modeled on the outward appearance of mental processes, rather than the structures that give rise to them.” We regard our brains as top-down, serial, procedural machines because that’s how our stream of consciousness perceives them. In reality, our brains are a “parallel, relational, and bottom-up” system of neurons in which a serialized stream of consciousness emerges.

Conway’s Life

To illustrate this model of emergence, Grand recalls a mid-80s art exhibit at SeaTac airport -- a wall of light bulbs programmed to represent John Horton Conway’s Life.

In Conway’s Life, each light bulb (or bit) has two possible states: on or off. The wall of lights is set to a beginning pattern by the user and then refreshes itself in cycles. With each cycle, a given light will reverse its state (from on to off or vice versa) depending on the state of its adjacent lights. From this simple rule emerge complex behaviors.  

First, there are still configurations, in which an array of lights is balanced and will not change its state upon refresh.

Then there are oscillating configurations in which an array will switch back and forth between two states.

And then there are emergent behaviors. In his book, Grand talks about the glider -- an array whose pattern actually glides across the wall of light bulbs.

Here’s the glider in isolation.

And here is a series of gliders moving across the wall.

The glider pattern in Life is a great example of emergence. It’s comprised of lower-level elements (the light bulbs) that, in their own behavior and state checks, are driving an abstracted, emergent behavior (the glider) of which they are completely unaware.

Steve Grand would take this idea of emergent behavior and use it in Creatures, creating a lower-level of sensory inputs to drive the abstracted level of intelligence present in his digital creatures.

Applications of Emergence

The cool thing about emergence is that, once you’re aware of it, you start to see it everywhere. Nature. Cities. Companies. All are systems in which lower-level components interact in a way that produces higher-level phenomena.

The hard things about these large systems is that their inherent complexity makes it near impossible to predict the emergent behavior that will emerge. To our minds, it looks like chaos.

As we saw in the popular approach taken to AI, our immediate response to chaos is to try and categorize it using the directive, top-down method our minds are used to. This is probably clearest in large companies, in which management dictates the strategy that the organization will take and mandates it -- top-down -- to the rest of the organization.

Emergence presents a different approach to management. One in which innovation and strategy are fostered from the bottom-up. Companies that take this approach (Google, Amazon, et al) focus not on top-down strategy, but, instead, on how they can influence the behavior of individuals. In the context of Conway’s Life, these companies don’t focus on the entire wall of lights, they focus on the behavior that drives the state of each light bulb, and how that behavior leads to the company’s emergent strategy (e.g. innovation.)

The behavior that drives these lower-level states is a company’s culture. Here’s Edgar Schein (sourced via Stratechery) on the importance of culture in fostering a company’s emergent behavior:

“Perhaps the most intriguing aspect of culture as a concept is that it points us to phenomena that are below the surface, that are powerful in their impact but invisible and to a considerable degree unconscious. In that sense, culture is to a group what personality or character is to an individual. We can see the behavior that results, but often we cannot see the forces underneath that cause certain kinds of behavior. Yet, just as our personality and character guide and constrain our behavior, so does culture guide and constrain the behavior of members of a group through the shared norms that are held in that group.”


In culture, a leader gets an opportunity to structure the “shared norms” that will guide behavior and, hopefully, drive the organization towards its objectives. Of course, therein lies the rub. Everything between the culture and the results it produces is chaos; it’s hard to see the destination from the path tacked today. All that one can do is set the right values for their organization, and see which way it glides.