Agents Of Fortune

I had one of my thunderbolt thoughts after a breakfast of too much coffee, and quickly went over to write it down: landscape is a stochastic art. The form of the design cannot be sure of any outcome, but is fundamentally open to chance. As an artist in landscape, you can set out a path and lawns to make certain experiences more likely, to make certain stories more likely. You can incentivize people to slow down and speed up; you can incentivize birds and dragonflies to land. But you are still far from control.The whole notion of landscape architecture, specifically, has a vexed relationship to this; as I’ve chewed over before, given the landscape architect’s standard workflow of generating instructions for laborers to follow in a defined construction process, the results of landscape architecture privilege, as much as they can, the stable and the stiff.

But in trying to follow the example of musicians or painters, who can fall into a relationship of confidence with their materials, the landscape artist finds themselves at loggerheads with their own material. Every branch on the trees they have ordered is open to the elements and open to hazard. A typical disclaimer from a plant seller: “We do not replace plants that are lost due to neglect, animals, winter losses, severe weather, or plants installed in soil or sun conditions other than those recommended.” The woeful failure rate of street trees highlights such a lack of certainty – no one would put up with structural concrete failing so frequently. A difficult dialogue with plants in particular, predictable at all only as broad aggregates and not as single actors, characterizes writing and thought around gardens; it tends to be shunted to the edges of landscape architecture. Plants are willfully ignorant of our plans for them.

It turns out that there is a much older idea of a stochastic art, coming out of Aristotelian thought: a stochastic art is one that cannot be sure of success, one that can try to perfect the artist’s aim without ever being sure it will hit the mark. It can be forgiven if the stochastic artist fails, because their materials are not entirely within their control. Some typical examples given: medicine, gymnastics, navigation, rhetoric, and, naturally, agriculture.

The more we think about it, the more we might see the necessarily stochastic aspect of any fine art - the willingness of anyone dialoging with the art to engage with it on anything like its own terms. If, on one hand, the arts of the 20th century recognized the dimension of willful and wayward reception, those of the 21st point to the ability to quantify the likely range of opinion and interpretation. That is, if the likely presence of failure is one key feature of the stochastic art, a more positive factor today is access to the volume of data necessary to better define the relationship between possible inputs and possible outcomes in complex systems. This need not mean feeling hedged in by the gauge of optimal impact. For instance, while discussion around Big Data in landscape architecture points toward generating yet more shaving-off of risk or error – routing paths to make optimal connections, or gauging the optimal placement of plants – it is just as possible to imagine the stochastic as a tool for exploring the possibility space of a set of parameters.

first noise islands
Most are more like Island 37 than Island 33 - but where's the fun in that?
second noise islands
It goes on and on.
error island
Here's Island 60: a very charming error that forgets to make the island at all.

To teach myself Grasshopper, I have been working on a process for quickly generating unique islands, not unlike the one I made to generate forests. These islands basically carve out segments of 3D-modeled Perlin noise – nothing advanced for the field. But they help me to understand how widely the results of one slate of parameters can range. And the errors generated along the way are their own sort of treasure.

It would be a blameworthy thing for a landscape designer to spend all of their time building gardens on the screen and never putting them into the world. As far as I can tell, aside from protoyping Voronoi’d benches, the best use for algorithmic tools is in envisioning possible outcomes of more open-ended processes. If such tools seem to have hit a wall in our field as a means to ecological simulation, as far as I can tell it remains to be explored how they can help us envision the results of iterative form-making over time. That is, they may be optimal for envisioning a process of landscape-making that is exactly not architecture, but a process, or a folk form, or a game.

(January 2022)