The Autogenerated Gardens of the Anthropocene

To begin, let me suggest that the twentieth century involved both a suppression of nature and an obsessive recreation of it. In the physical world, we turned fields and forests into machines. America’s native grasslands have been replaced with acres of monocultural corn; Brazil’s Cerrado savanna is being overtaken with non-native soy and eucalyptus.¹ In our computational world, we wanted our machines to transcend their plodding, ponderous ways: we wanted them to see and think as we do, and be as vital as the living world.

But what does the living world look like after all of our interventions? We have higher agricultural yields – and heightened anxieties. We are increasingly reliant on monocultures to feed ourselves, but homogeneous fields are more susceptible to disease and drought, and lack the genetic diversity to adapt to a changing climate.

Many plant scientists have pinned their hopes on seed banks, which archive seeds and the genetic possibilities they contain. Scientists believe that these seeds can help us feed the world in a climate crisis when our current monocultures inevitably fail.²

A seed bank anticipates disaster – an end of the world – where our fields will need to be replanted anew. Existential threats to seed banks, then, are existential threats to our future food supply.

Let’s imagine one particular seed, born in the Fertile Crescent, inheriting over a thousand years of agricultural history. Born in Iraq; lived for many years in Abu Ghraib – the site of Iraq’s national seed bank; sent away from home in 1996 (a fortuitous move for the seed – Iraq’s seed bank was destroyed in 2003); taken in by the International Center for Agricultural Research in the Dry Areas (ICARDA) in Aleppo, Syria; years later displaced again, by conflict (ICARDA sent its seeds to the Svalbard Global Seed Vault, in Norway, before the Aleppo bank was also destroyed); today ICARDA is rebuilding in Lebanon and has welcomed some of its former seeds back home.³

(It should be noted that one of the most significant risks to seed banks, and their seeds, is the most banal: underfunding, the enemy of all public institutions in our time.)

The Svalbard Global Seed Vault, in the Arctic circle, is primarily funded by the Norwegian government and the Gates Foundation. It has archived over one million plant species – primarily staple crops, like wheat, rice, chickpeas, and maize (from Italy, to secure the possibility of polenta in the future).⁴

Architecturally, Svalbard and other well-funded seed banks have converged on particular themes. They often include subterranean elements that are pragmatic (allowing for seed preservation at optimal temperatures) and poetic (an instinctual sense that seeds and safes are meant to be underground). Svalbard’s seed storage rooms are embedded in the permafrost, and the entrance to the building emerges from the earth as an elongated slab of concrete. The Australian Plant Bank, in New South Wales, has a passageway that merges in and under the land to enter the building.⁵ These buildings often exhibit a commitment to public-facing science: the Australian building and the Millennium Seed Bank in Britain both have laboratories swathed in glass, inviting visitors to participate in the preservation of our agricultural future.⁶ At Svalbard, visitors are held at a remove and can view only selected seeds in a Snøhetta-designed visitor center.⁷

This is indicative of the Svalbard Global Seed Vault’s role as a “doomsday vault.” Other banks share their seeds with researchers and farmers, but Svalbard is a backup of other seed banks – the seed bank of last resort that can help ICARDA and others recover.⁸

In invoking backup as a metaphor, I am leaning on a computing metaphor that articulates each seed as a unit of information. But metaphorical loans can go both ways. In game design, a seed is an input (like 784555053457240316537164) that can be used to procedurally generate a built space or landscape instead of hand-making it.⁹ From this input, one algorithm (or a collaborative cascade of them) can cause data to produce a game world with a pseudo-random landscape of slopes and archipelagos, or flowers dotting a field; or buildings for a player to enter, and within that building, what each room contains.

Procedural generation subtly alters the role of a game designer: no longer an omnipotent, omnipresent god, but someone gently trimming and weeding an autogenerated garden. The architect/gardener articulates rules (generate a room, where a room has at least one door, and generate one window per exterior wall if the room is aboveground; generate a field, unevenly vegetated, placing trees such that no tree overlaps another tree) and the world unfolds from there.¹⁰

Relying on procedural generation requires surrendering control – this, in a world where perfect control is the norm.¹¹ In the real world, of course, perfect control is an illusion. And yet – our desire to achieve perfect control over crop yields has led to mass pesticide use and mass reliance on monocultures, and both of these sins may contribute to a future agricultural collapse.

The degradation and impoverishment of our natural world is unsettling when compared to our virtual landscapes. Game designers want heterogeneous fields and forests; they want to simulate an interdependent ecosystem of plants, thriving or thwarting each other for the simulated sunlight. Algorithms that simulate natural processes are turned towards the task of making a vital, lively, biodiverse world.

Perhaps our digital and physical worlds (where fields are increasingly regularized, industrialized, and controlled) converge. A comparison with the past is instructive: America no longer grows 90% of the fruit and vegetable species it was growing in 1950. China, similarly, has lost 90% of its rice varieties since 1900.¹² Establishing control has made us more fragile.

Perhaps these digital worlds will surpass our own one day, in a biodiversity singularity.

Some games will now use procedurally-generated landscapes to create an “infinity mode,” an endless field for human exploration and play.¹³ These games promise an unbounded future, where the world of the game never has to end.

The Svalbard Global Seed Vault was designed to withstand the end of the world, which may come sooner than we expect. In 2016, the anthropologist Eduardo Viveiros de Castro and philosopher Déborah Danowski observed “Although it began with us, it will end without us: the Anthropocene will only give way to a new geological epoch long after we have disappeared from the face of this Earth.”¹⁴

Meanwhile, our infinite virtual fields grow more naturalistic, decadent, and green.