Issue 4 / Scale

April 01, 2018
A photo of a "food server."

The OpenAg Food Server at the MIT Media Lab. Courtesy of the Open Agriculture Initiative, MIT Media Lab.

Farms Race

Nina Sparling

Advocates of “open-source agriculture” say they can build a better food system. Should we believe them?

Climate is anything but democratic. Agriculture is tied to place. Grapes thrive in the hot winds of Napa Valley while coffee requires the consistent humidity of the equatorial zone. Olive trees are abundant in Spain; cloudberries thrive across the Arctic zone, and nowhere else.

Climate change is making climate even less democratic. It amplifies the polarity and raises the stakes. Farmers in Jordan suffer from increasingly saline soils while Alsatian vintners fret over what warmer temperatures mean for their vines. California, subject to alternating periods of drought and flooding, produces over a third of the vegetables and more than half of the fruits and nuts that Americans eat.

The need to overcome these environmental challenges has only bolstered our existing, highly globalized food system. States import and export goods in attempt to feed populations and stabilize their own economies. The “food miles” add up; the emissions accumulate. Corporations reap incredible profits in the process.

And nevertheless, global hunger is rising. People are moving into cities at unprecedented rates. Farmers are aging out of the business worldwide. The corporations that control much of the food supply continue to consolidate.

Given these pressures, a set of entrepreneurs want to build a climate-agnostic form of agriculture. What if you could grow avocados in Jordan and olives in the Arctic? Deploying the latest developments in automation, machine learning, and computer vision, these high-tech food startups grow indoors, nourishing plants through hydroponic or aeroponic systems and rhythmic exposure to LED light.

As is standard in the commercial technology and agricultural industries, the companies don’t disclose their methods. From hardware to plant research to data standards, proprietary and defensible intellectual property forms the backbone of their businesses. But over the past few years, an initiative based at the MIT Media Lab in Cambridge, Massachusetts has started sowing the seeds of a different sort of indoor agricultural revolution—one built on open-source hardware and software.

Agriculture Wants To Be Free

The Open Agriculture—or “OpenAg”—ecosystem consists of three parts. The Open Agriculture Foundation protects licenses and manages a community of “#nerdfarmers,” as OpenAg practitioners are called. The Open Agriculture Initiative conducts research and partners with universities—currently Chiba University in Japan and Can Tho University in Vietnam. And Fenome, a startup, is attempting to commercialize OpenAg’s signature product: an open-source “personal food computer.”

The project started in earnest in 2015 when Caleb Harper hacked together the first personal food computer with a team of researchers. An architect by training who grew up tinkering with computers, Harper turned to indoor farming following a trip to Japan. There, he saw a plant factory: an immense facility growing tons and tons of greens. Something clicked: the prospect of growing large volumes of produce in a country that imports a significant percentage of its food supply looked like a way to democratize climate.

Back home in the United States, several companies were already at work on commercial indoor farms similar to what Harper had seen in Japan. But his proposition was different: he wanted to create a network of semi-automated indoor farms that ran on an open-source platform. The farms would range in scale from small units the size of a mini-fridge to warehouses outfitted with plant towers.

Over the next two years, Harper scaled his experiment into a global community of horticulturalists, mechanical engineers, plant scientists, teachers, and hackers committed to the idea of open agriculture. Like any open-source community, OpenAg enthusiasts swap hardware specs and software code on GitHub, and trade troubleshooting tips on an online forum.

But they’re not just hobbyists— they’re crusaders. Instead of highly centralized industrial farms run by Big Ag, they envision a radically decentralized network of customizable farms. They want to give everyone the power to grow anything anywhere.

Personal Food Computing

The most basic unit of food computing is the personal food computer, now in its 2.1 edition. It consists of a plastic container affixed to a metal frame outfitted with sensors, microcomputers, lights, and electrical wiring. The machine grows plants in a tub similar to what cafes use to collect dirty dishes. A floating styrofoam tray fits snugly inside and has evenly spaced holes where the plants grow.

The tub itself holds water treated with the nutrients and fertilizers best suited to the crop. Some #nerdfarmers outfit their food computers with aeroponics—a system that uses nutrient-dense mist rather than water. An outer shell made of clear acrylic encases the entire box, creating a controlled environment. An array of LEDs affixed to the top of the box delivers light to the plants.

Two printed circuit boards form “the brain.” A Raspberry Pi microcomputer hosts the software that allows a farmer/user to control the system from a computer, while an Arduino microcontroller links to the temperature, humidity, pH, and CO2 sensors that collect data about the environment.

Let’s say you want to grow arugula from seedlings. You load a “climate recipe” onto the Raspberry Pi that consists of many lines of code. The personal food computer interprets this code, and adjusts the lighting, humidity, temperature, pH, nutrient levels, and CO2 to the appropriate levels:

"air_temperature": [

	{"start_time": 0, "end_time": 17, "value": 21.1},

	{"start_time": 17, "end_time": 24, "value": 15.6}],
“water_potential_hydrogen”: [
	{"start_time": 0, "end_time": 24, "value": 6}],
“air_flush”: [
	{"start_time": 0, "end_time": 24, "value": 3}],
“nutrient_flora_duo_a”: [
	{"start_time": 0, "end_time": 24, "value": 10}],
“nutrient_flora_duo_b”: [
	{"start_time": 0, "end_time": 24, "value": 10}],
“light_intensity_red”: [
	{"start_time": 0, "end_time": 17, "value": 1},

	{"start_time": 17, "end_time": 24, "value": 0}],
“light_intensity_white”: [
	{"start_time": 0, "end_time": 17, "value": 1},

	{"start_time": 17, "end_time": 24, "value": 0}],
“light_intensity_blue”: [
	{"start_time": 0, "end_time": 17, "value": 1},

	{"start_time": 17, "end_time": 24, "value": 0}]

Arugula is only one example. The machine could grow anything, although like the rest of the indoor farming community #nerdfarmers tend to focus on crops with short growth cycles and high value propositions. The less plant you have to grow to get the end product, the more efficient the system. And efficiency is important because the machine isn’t cheap: in all, a personal food computer costs between $3000 and $5000 to build.

The #nerdfarmers I spoke to see personal food computers as a way to bring crop cultivation back to where people live and reduce the distance between farm and table. Everyone I spoke to lives in a city, and by and large, they recognize the current limitations of OpenAg. We aren’t going to see amber fields of grain enclosed in a warehouse-sized “food data center” anytime soon.

That’s because the technology is still fairly primitive. Dan Nelson was among the first people to build a personal food computer from the published plans. He works out of AgTech X, a coworking space in Brooklyn for entrepreneurs in agricultural technology. He described the GitHub instructions as “terrible.” And he isn’t much more optimistic about the machine’s latest version: “I think it’s totally inaccessible and totally unaffordable for the overwhelming majority of people who aren’t doing it in a research context,” he said.

It’s inefficient to grow anything except leafy greens. The economics are wildly bad. Still, #nerdfarmers see the personal food computer as an important first step: not a wholesale solution to a broken food system, but a useful teaching tool, and a modest opening salvo in the battle to wrest control of agriculture from Big Ag.

Through the forum, Nelson met Peter Webb, who lives in St. Louis. Webb almost discovered the OpenAg community too late. He wanted to build a machine that grew food. So he scoured the internet for guidelines, parts, and advice. His research took him into the depths of the hydroponic cannabis industry: most of the material he found online focused on equipment for growing weed. The parts were priced accordingly—up to $1000 for LED lights, several hundred dollars for grow tents, $100 for reservoirs.

Spending thousands of dollars to grow weed might be worth it. But spending that amount of money to grow a tomato plant, as Webb put it, is simply not sustainable. Working from his bedroom, without institutional support or corporate research dollars, he put months into creating a prototype for an open-source build-your-own-hydroponic-farm kit without any knowledge of the MIT project. He listed ten of them for sale on eBay, but none sold.

Not long after, Webb found out about OpenAg, and he burst into tears. “Within two months I went from being totally alone, a crazy kid in a closet to having 1000 people on the forum, to having a community of people.”

Webb connected with Harper, and he now leads a team that is developing designs for a cheaper, leaner food computer. The current prototype costs only about $300 to build, and uses lower-cost materials: mylar and PVC piping instead of metal frames and sheets of acrylic. Moreover, it’s simpler. The personal food computer has 152 unique parts; Webb’s version has thirty-one. This also means fewer aspects of the machine are automated, which means more work for the user.

Big Data Farming

You need efficient hardware to grow food indoors, but both #nerdfarmers and agtech entrepreneurs recognize that the true value lies in the software and the data. “No company is going to get millions of dollars for anything other than proprietary algorithms and data,” Webb said.

But the indoor farming industry is still young—by and large, it lacks standards, said Allison Kopf, the founder of Agrilyst, a software company for indoor growers. The pressure that venture capital investors put on startups to protect their intellectual property is counterproductive, she thinks. “It would be hugely beneficial to the industry as a whole if everyone could sit down and say how do we develop these standards from climate devices to software.”

From hardware to software, the industry needs a playbook, and OpenAg could help build that. The open-source community could develop a common set of standards, practices, and source code that startups might use to build their businesses.

One example is climate recipes, the source code of OpenAg. In the past year, the MIT team has honed in on how to write these recipes, but the community hasn’t seen much of that research yet. Harper has posted about partnerships with AI firms and doubled down on phenotypic research—which involves learning how to put environmental stress on basil plants in order to produce particular traits. But the dream of the Open Phenome project—a public database full of climate recipes that anyone can download onto their personal food computer—is some ways off.

Data is another weak point. The data that personal food computers collect, Webb said, doesn’t easily translate into meaningful plant science metrics. It’s fine to measure CO2 levels, temperature, and humidity, but without a clear understanding of how those influence the biomass of plants, the information isn’t all that valuable. A climate recipe should ultimately clarify the details of output and yield. “There is this disconnect between open field plant science and this controlled environment agriculture,” Webb said.

He often thinks about Monsanto, where his dad worked: although anything but open-source, the company has mastered the art of data collection. Its extensive laboratory research on corn and soy has given it an enormous competitive advantage.

For digitized agriculture to fulfill its potential, it has to follow Monsanto’s lead on data. It must standardize how it gathers information, and develop a process for converting that information into knowledge that improves the growing process. If and when OpenAg hacks its way to a platform where users can post and exchange climate recipes, the open-source model could prove a valuable tool for commercial ventures—much as Linux, Apache, and countless other projects have for the tech industry.

Open-Source Capitalism

For all its idealism about disrupting Big Ag, OpenAg may ultimately help strengthen it—or reproduce it in a different form. Just because open-source projects give their source code away for free doesn’t mean they aren’t immensely useful for commercial enterprises. Today’s tech giants rely heavily on open-source projects.

The question, then, isn’t whether OpenAg will benefit business, but which businesses will benefit. Will traditional Big Ag companies like Monsanto embrace OpenAg research and development? Or is it more likely that we’ll see tech companies move into the space?

It’s too soon to know, but the signs seem to point towards tech. Already, the indoor farming industry has modeled itself more after tech companies than after agricultural ones. SoftBank’s Vision Fund—the world’s largest corporate venture fund—made the single biggest investment in agtech on record in Plenty, a vertical farming company based in South San Francisco. Jeff Bezos’s personal venture fund also invested in Plenty—a significant gesture, perhaps, given Amazon’s recent acquisition of Whole Foods.

It’s possible to picture OpenAg becoming big business in the near future. Imagine an Amazon warehouse that not only stocked toilet paper and televisions, but grew broccoli and lettuce—and sent you any of the above in the mail.

“A suite of production technologies can simultaneously embody radical ideals of cooperation, freedom, and social change and be an ever more widely embraced model of capitalist software production,” write Samir Chopra and Scott Dexter in their paper, “The Political Economy of Open Source Software.” The same is true of OpenAg: developing shared networks of agricultural knowledge can embody utopian ideals even if it serves to retrench industry.

When farmers, like software developers, have the right to modify and alter their code, everyone benefits. The #nerdfarmers vision of a radically decentralized digital food system isn’t likely to prevail over the deeper pockets of the closed-source community, but the work still matters. Linux is better than Windows, and farmers have long fought to preserve their rights to “open-source” methods from sharing seeds to cultivating common lands. As we race towards a digital future of food, the ability to freely adopt and adapt technology amounts to a small yet fertile plot for resistance.

Nina Sparling is a writer based in New York whose work has appeared in The New Food Economy,, and The Rumpus.

This piece appears in Logic's issue 4, "Scale". To order the issue, head on over to our store. To receive future issues, subscribe.