Skip to content

Genetic Architecture: Nature versus Nurture?

May 13, 2011

Susan at the Hanging Tree

This week’s reading was Architectural Genomics by Keith Besserud of Skidmore, Owings & Merrill’s Black Box Studio which describes their experiments using genetic algorithms to create building form and the results they achieved in shaping a multi-level tower to optimize sun exposure in the context of shade from the surrounding buildings.  Images are available here and in the paper linked to above.

Genetic algorithms use a heuristic search technique to mimic evolution by generating new “species” of objects and killing off the less successful members of each generation.  Search heuristics for complex optimization problems are often described in terms of natural processes.   Simulated annealing, for example, may be thought of as attempting to “shake” objects on integrated circuits into better and better placements and Ant colony optimization finds optimal connective paths based on the mechanisms ants use to reduce travel to the shortest path between their nest and a food source.  Each of the optimization problems addressed by these algorithms are “computationally hard”.  So hard that finding an exact solution is not feasible and the heuristics approach the problems by abstracting them, thinking about how the natural world approaches them, and crossing their fingers.

Optimizing architectural design, even ignoring aesthetics, is similarly hard.  There are too many parameters to optimize in a design problem and they are difficult to precisely quantify in relation to one another.   In thinking about using genetic algorithms on the built environment, however, it strikes me that this may not be the right natural heuristic to invoke.  Genetic codes define specie and the details of that specie’s design.  Whether you have a giraffe or a horse;  the one optimized to forage on ground cover and the other on tree leaves.  The closest analogy in the built environment might be the building “type” rather than form.  Is this a church or a skyscraper?  Certainly something that must be determined and potentially should be optimized within its context.  But in these experiments they already knew the type of building they wanted: a tower.  The optimization problem was to find the optimal form for that tower within its space and context.  A better analogy is that of how a plant grows.  The genetic code it starts out with tells it to seek the sun and it creates its form based upon that drive. A plant optimizes itself to its location, its orientation and surrounding environment, but it does not have to decide whether it is a dogwood or a maple.  From a single, simple, initial cellular starting point it knows how to create the appropriate sized leaves and branches and flowers and seeds and fit all those things into the right spots.

All of which leads me to speculate that a different optimization heuristic might better apply.  Perhaps there is a means of “growing” a building using  a preset genetic typology that defines its characteristics and how it will respond to its environment, and then leave its unique individual environment to determine the exact form it takes.  The algorithm would be progressive and generate form from the inside, developing in response to the surroundings.  It might be necessary to guide the growth by pruning, grafting or forcing branches into desired configurations. Call it the “Bonsai Tree” algorithm.

I don’t know how to do it, but it sure sounds like fun to try.

No comments yet

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

%d bloggers like this: