Farmville
The Autonomous Future of Agriculture
Editor’s Note: This is a piece I wrote some months ago for a publication that ended up ghosting me after we worked together on a final draft. First time that’s happened. I’ve been too immersed in our ongoing constitutional collapse to figure that all out and get this posted.
But even during times of crisis, technological progress can still continue, as it did during cataclysms like the Great Depression and World War II, and it’s worth reminding ourselves of that. As you will see, this story is also closely connected to our current policy on immigration.
Between when I wrote this and now, John Deere did a big unveiling of its autonomous vehicles at the 2025 Consumer Electronics Show. See an update about that here.—RWT
There has been a lot of talk in the tech world for the last decade about the need to innovate with atoms instead of bits. But what about dirt?
The technology industry tends to be dominated by college-educated urbanites who think in terms of what can be done with a microprocessor. There has been a recent push to apply more innovation to the gritty world of heavy industry. But less attention has been paid to technological progress in agriculture. After all, if heavy industry seems like the world of our 20th Century past, agriculture seems like the world of the 19th Century.
Yet new technology is already transforming how our food is grown, and without much fanfare, agritech has been advancing at least as quickly as other more attention-grabbing innovations.
Let’s put it this way: The self-driving tractor will be available for purchase by the general public before the self-driving car—and just in time, perhaps, to head off a serious crisis in agriculture.
The Internet of Cows
Drones might be used increasingly in war and for entertainment—but probably their most practical use is on the farm. The need is obvious: farms are big and sprawling, and getting information on the state of crops and soil requires a lot of traveling back and forth. Why not do it with drones, instead?
But drones aren’t just being used for crop surveillance. They are also being used to spray crops with pesticides and fungicides and even distribute seeds, potentially replacing more expensive ground-based rigs. They can also be used to spray trees with pesticides and move over uneven terrain.
Drones are beginning to be used in cattle ranching, where they can herd cattle, replacing dogs and men on horses or all-terrain vehicles.
Or consider the Internet of Cows. You may have noticed that cows usually have yellow tags pierced through their ears, bovine jewelry that bears numbers for identifying the animals and recording important information about them. For years, farmers and ranchers have been able to buy a version that also contains a GPS tracker, powered by a small solar panel on the tag, which allows a rancher to track his herd remotely. It’s a curious application of the Internet of Things: Open up your laptop and see where your cows are.
The next step—which has been talked about for years but is still in the early stages—is “virtual fencing.” Couple the GPS to a device along the same lines as the invisible fence already used for dogs, which can deliver a shock to move cattle away from a barrier set only through GPS co-ordinates. Some are trying to incorporate this into the ear tags—while others are doing it with very large collars. This saves ranchers the labor of sending workers to move cattle, as well as enabling them to move cattle more frequently to allow pastures to regrow more quickly. It also saves the expense of putting in barbed-wire fences—something the World Wildlife Fund is promoting, because it removes barriers to the natural migration of wildlife.
Pixelated Weeds
Technology for sensing and automation also holds out the promise of “precision agriculture.”
The Internet of Cows, for example, also includes monitors in the stomachs of cows that help to track their digestion and health.
Rather than watering an entire field the same amount or delivering the same amount of fertilizer—an approach that is optimized for the average plant, not for every plant—precision agriculture uses sensors, drones, and GPS mapping to determine variations in things like moisture and nutrient levels and automatically vary the response of farm machines for each area of the field.
Helped by its acquisition of Sunnyvale, California startup Blue River Technology in 2017, farm-equipment giant John Deere has been applying machine learning and machine vision to precision agriculture. For example, Deere’s Chief Technology Officer Jahmy Hindman describes an automated sprayer that uses an array of cameras and machine learning for “discriminating pixels that have weeds in them from pixels that don’t have weeds in them, and then we only spray the areas where the pixels have weeds in them.”
The Self-Tilling Farm
All of this is leading up to the big breakthrough in agritech, the self-driving tractor.
Years ago, Case IH unveiled an Autonomous Concept Vehicle that was striking for the absence of a cab for a human driver. But notice the phrase “concept vehicle.” In actual use, IH has offered only “supervised autonomy” that still requires a human driver.
Livermore, California-based Monarch Tractor has been offering small electric tractors that are autonomous or remotely operated. But it is a startup that has only shipped 500 tractors to date, and it recently announced that it is retrenching and focusing on software for non-agricultural applications of autonomy. The decision to produce electric vehicles limited them to small tractors, and most of their customers were California wineries that have suffered a serious downturn due to a post-pandemic decline in wine consumption.
But battery technology has not advanced far enough to make electric tractors practicable for the house-sized behemoths that are used to work the vast fields of the Great Plains.
Probably the leading position in autonomous tractors has been clamed by Deere & Co., famous for its giant green-and-yellow tractors. Deere is not a startup and is decidedly not based in Silicon Valley. It was founded in rural Illinois in 1837 by a blacksmith who invented the self-scouring steel plow. Yet Deere & Co. is trying its best to propel itself rapidly into the 21st Century. According to Hindman, Deere now has more software engineers than mechanical engineers. The company has already been testing out fully autonomous tractors on selected farms and plans to offer them for sale in 2025.
The rollout of autonomous tractors largely parallels the rollout of self-driving cars.
As with self-driving cars, autonomy has been implemented partially for years. Automakers long ago began offering “driver assist” functions like lane assist, automatic emergency braking, and parallel parking. Similarly, Deere’s efforts in autonomy began in the 1990s, particularly with its acquisition of GPS navigation company Navcom in 1999, which allowed tractors to use precise positioning data even in remote locations. Elements of autonomy have been rolled out in stages. AutoTrac, for example, includes the ability for large tractors to make precise automated turns at the ends of rows, removing from the human driver the tasks of precisely aligning the tractor and raising and lowering a tiller or disengaging and engaging seed planters. To this, Deere has added the ability to set GPS guidance lines while tilling a field and follow them precisely during later passes of planting, spraying, and harvesting. Add to this automated tracking between harvesters and grain carts, the truck-like vehicles into which they deposit their grain while still moving. See the comical efforts of British TV-host-turned-farmer Jeremy Clarkson to understand why this is so valuable.
Also as with self-driving cars, the rollout of autonomy has been cautious and limited to avoid catastrophic outcomes that would give the new technology a bad name.
With self-driving cars—particularly in their current, limited application as robotaxis—the catastrophic outcome is that the vehicle will strike a pedestrian or cyclist. In this respect, self-driving tractors are a more promising application because there’s not much to bump into in the middle of a corn field. This is why there are few regulatory barriers to making them available for sale to the general public.
The catastrophic outcome for autonomous tractors is ruining a crop. A tractor moving at 15 miles an hour is crossing the field at 22 feet per second. If it veers off course for even a few seconds, it could trample multiple rows of an established crop. Even a few such mistakes could wipe out thousands of dollars of a farmer’s harvest, making autonomy a liability rather than a savings.
This is why Deere’s rollout of full autonomy will initially focus on tilling fields before the crop has been planted, which limits the damage of a mistake and leaves time to correct it. But Deere is targeting a fully autonomous process for all stages of planting and harvesting corn and soybeans by 2030.
Such deadlines have slipped before, just as they have done with self-driving cars. But whenever it happens, it won’t be a moment too soon.
How Ya Gonna Keep ‘Em Down on the Farm?
Robotaxis are a convenience for city-dwellers. Autonomous tractors may soon be a necessity for farmers.
The fundamental driver of the push for automation and autonomy in agriculture is an acute shortage of agricultural labor, as yet another generation of farmers’ kids are moving to the cities, drawn by shorter hours, more stable pay, and less arduous conditions. This is not just in America. It is a global phenomenon. The average age of an American farmer is about 58. It’s 60 in Kenya and 67 in Japan.
There are already reports of established farms forced to shut down operations because they can’t find enough workers and of fruits and vegetables left to rot in the field because there aren’t enough workers to harvest them.
Immigrant labor could make up for this shortage, but legal immigration and guest worker programs have long been restricted—and now much more so. The Department of Labor estimates that “nearly half of the nation’s approximately 2 million farm workers lack legal status…as well as many dairy and meatpacking workers.” The Wisconsin Examiner reports that “workers from Mexico and other Latin American countries make up an estimated 70% of the labor force on Wisconsin dairy farms.” The president of the Idaho Dairymen’s Association predicts that promised mass deportation sweeps would have catastrophic results—not just higher prices, but actual shortages. “I don’t think there would be milk. I just don’t think we could get it done.” It is a grim irony that rural areas voted overwhelmingly for Donald Trump in 2024, yet farmers are now panicking that he will follow through on his central campaign promise.
Farm automation companies are already using this labor shortage as their main selling point. It is certainly a compelling economic driver: the need to run a farm with far fewer workers than ever before.
There is a long history behind this. The great industrialization of America in the 19th and early 20th Centuries was also a great de-agriculturalization. In 1800, nearly three quarters of Americans worked on the farm. Today, only 2% do. This is the next stage. In the coming years, those numbers could collapse again, as a dwindling number of farmers do the same work, but with even more machines and even fewer workers.
Yet is not just the number of workers that may change. It’s also the character of the work. Consider this tantalizing anecdote.
A farm in Sonoma County had 27 tractor driver positions open and posted the positions on various job boards. The farm didn’t get a single applicant for weeks.
A few weeks later, the owners onboarded an autonomous tractor system for the farm and updated the job listing to say they were looking for an agtech operator position. In the preferred qualifications they listed “video game experience.” The applications came rolling in.
Some readers may remember that during the early years of Facebook there was a mania for an online game called FarmVille. The game finally shut down a few years ago as its audience dwindled and its underlying technology became obsolete. But it accustomed a whole generation of very online young people to the idea of tending a virtual farm on their phones, notebooks, or laptops.
That may end up being a curious preview of the future of American farming.
A wonderful non-Trump story. It's always fascinating to find out the many wags free human minds can continue to improve our lives. Thanks!
Fascinating peek into the redesign of a very old industry!