How Big Tech Ruined Farming

How Big Tech Ruined Farming

Show Video

If you were in Las Vegas, in January 2023,  sitting in the audience of the opening keynote   of CES—perhaps the most influential tech  conference in the world—you likely had one   question on your mind: “Why is John Deere here?” After all, they’re the tractor company,   right? They make machines that push and  pull and move and dig, that’s their thing,   right? Well, not according to their CEO, John May. “We’ve quickly become one of the world’s leading   robotics and AI companies. Our solutions  leverage technology like computer vision,   advanced sensing and compute, machine  learning, and data analytics.” 

There’s one key word there: solutions. That’s a  word that gets thrown around a lot by companies   like Apple “…solution…” “…solution…” “…  solution…”; Microsoft “…we provide an   end-to-end tooling solution…” “…deceptively simple  solution…” “…the best end to end solution…”;   Google “…bespoke AI solutions…” “open  sourcing solutions…” “…a great solution…”.  This linguistic mimicry, their mere attendance  at an event like CES, it’s all… peculiar for a   tractor company. Unless, does John Deere  think it’s a tech company? Well… yes,  

they do. At least according to themselves, in  this LinkedIn post, sharing an article entitled   “John Deere: ‘We’re a Technology Company.’” And that assertion appears increasingly less   absurd. While the company lays off hundreds in  its manufacturing plants, it’s simultaneously  

staffing up its tech divisions. Of fifteen current  US job listings, twelve are in software, data, or   robotics—just three in manufacturing. The company  has been going through a metamorphosis from one   that makes machines to one that makes solutions. Apple, for instance, does not merely make   computers or phones or tablets. They  make integrated technology solutions,   blending software, hardware, and services into an  ecosystem that envelops ones digital experience.  

Correspondingly, John Deere no longer merely  makes tractors or combines or loaders, but   rather integrated production solutions—blending  software, hardware, and services into an ecosystem   that envelops a farmer’s day-to-day experience. At the extremes, the company appears entirely   disconnected from its original form as a  small Illinois storefront selling shovels and   pitchforks. Never could John Deere himself have  imagined that his company would eventually go on   to operate, for instance, a satellite network.  But waxing nostalgic about humble beginnings   would mask what the company now truly is: it’s not  an endearing family business, it’s not a scrappy   underdog, it’s a market-domineering behemoth. It  has the business of agriculture, especially in   America, in a stranglehold. And so might Apple,  with consumer electronics, but farming is not  

something you pick up or put on. It’s not one’s  digital experience, it’s one's entire experience:   it’s your job, it’s where you live, it’s what  your family does, it’s what your neighbors do,   it’s what your descendants do, it is an all  encompassing way of life whose future is now being   dictated by one tractor company that’s decided  it’s big tech. But John Deere’s power, their   influence, their ability to change the course  of history has been centuries in the making.  It started with this—the self-scouring  steel plow. Plows had existed in some form  

for millennia—so many millennia, in fact, that we  can’t even say how old they are—but by the 1800s,   they’d been refined and refined into this: a  single-piece cast-iron plow. These could be   pulled by an animal, and would efficiently loosen  the soil to bring nutrients to the surface before   planting a fresh crop. And these worked great, for  the time, except for here—the American midwest.   The soil of Illinois and its neighbors was thick,  moist, and rooty, in a fashion that would lead   it to clump on the plow, forcing farmers to  stop every once in a while to clean it off.  But John Deere had an idea: he  would manufacture the same plow,   but of polished steel. This cut straight through  the midwestern ground with far greater ease, and   the soil would shed right off rather than clump.  While he was not the first to invent this concept,   he was the first to start manufacturing  a steel plow en masse, and his production   steadily grew into the hundreds per month,  and a later thousands. This innovation played  

an instrumental part in spreading agriculture  across the region, and transformed John Deere   from a mere shop into a growing manufacturer. By the turn of the century, after the company   had passed through the generations of the family,  Deere had become a leading agricultural implements   manufacturer, but the industry landscape was  changing beneath them. Like the plow, tractors   as a concept, had long existed. Through the 19th  century it was typically animals like horse or   oxen that pulled plows and other implements,  yet around the world, across industries, animal   power was being replaced by steam power. Whereas  horse-drawn stagecoaches dominated the past, steam   trains were now the dominant form of long-distance  transport, so the logic carried that steam-powered   tractors could replace animal-power on farms.  And they certainly could, the technology existed,   but steam-powered tractors never became ubiquitous  due to their high up-front and operating cost. 

But with the turn of the century came  gas-powered tractors: cheaper to buy,   cheaper to operate. Popularity exploded, and while  Deere was originally reluctant to stray beyond   their agricultural-implement core, they eventually  realized they had no choice if they wanted to   stay relevant. They tried to design their own,  and it was plenty capable, but it was just not   competitive. At about twice the price of that of  equivalent machines, their tractor never had any   shot of commercial success, and then its designer  died from pneumonia following a week of testing in   the wet and cold, so John Deere rather elected to  just simply buy their top competitor. With that,   the Waterloo Gasoline Engine Company was folded  into John Deere, and over the following years,   their tractor, the Waterloo Boy, enjoyed wild  success. Once again, like with the self-scouring   steel plow, Deere didn’t invent the technology,  but they popularized it—they identified the   opportunity, scaled up manufacturing, marketed  successfully, and helped transform the tractor   from a niche, novel technology into the  solution for moving power on the farm. 

Over the decades that followed, the company  transformed again from primarily an implement   company into a machine company—offering  combines and balers and planters and sprayers:   essentially anything you needed to turn a field  into a farm. Through much of the 1900s they were   always a significant, but underdog player in the  industry until the 60s and 70s when their primary   competitor, International Harvester, began  to falter. And with its collapse in the 80s,   John Deere took a firm lead in  the industry—becoming the go-to,   ubiquitous source of agricultural equipment in the  United States. But then, another monumental shift  

in the field. Like the tractor introduced  the mechanical era of farming, information   technology introduced farming to the digital age.  And again, John Deere had to adjust on the fly.  This started on June 23, 1995 when Rockwell  International Corporation, traditionally a US   defense manufacturer, unveiled its proprietary  Vision System—effectively firing the first shot   of the digital farming revolution. It seems  so simple now, but Rockwell’s Vision System   was poised to usher in a new age of efficiency by  using defense satellites to pinpoint and track a   tractor from above, which, in turn, would allow  a farmer to better monitor their field’s yield,   or when it came time to plant, better disperse  seeds and spray chemicals. Precision farming  

had arrived, and while Rockwell was first,  competitors such as Case Corporation, and Agco   Corporation were close on the company’s heels. So too was Deere, which took it one step further.   To the late-90s American farmer, GPS and precision  agriculture was a handy tool, but still a finicky,   expensive, and difficult-to-use luxury.  It helped, but it wasn’t required.  At least, not until Deere made it a practical  necessity. As they did when developing their   tractor, Deere looked further afield  for help. This started at Stanford,  

where the company collaborated with engineers  to develop an autonomous GPS-controlled tractor.   While it worked, it didn’t work well enough  to take to market. The problem was 1990s GPS   just wasn’t accurate or dependable enough.  So, more partnerships. Now Deere, along with   NavCom Technology and NASA’s Jet Propulsion Lab,  sought to figure out how to create a more reliable   positioning system to support not just yield maps  but autonomous guiding—the former a helpful tool,   the latter a potentially revolutionary product.  While autonomous guiding may have seemed a   lofty ambition, its use case was well grounded. Before the rise of precision farming, farming was,  

well, remarkably imprecise. Take, for instance,  actually planting a field. Now, laying down seed   is actually a rather complicated process with  a whole host of decisions to make and factors   to consider from when to seed to how to establish  then plant the field’s headlands and borders. But   regardless of such considerations—or what crop one  is even planting in the first place—each and every   farmer, since the dawn of the tractor, has dealt  with one major inefficiency: overlap. In farming,   overlapping is practically unavoidable—as  a tractor operator threads rows back and   forth across their field, it’s nearly impossible  for there not to be slivers of field—whether it   be where rows meet headlands, or just between  rows themselves—where the farmer doesn’t pass   over a small section twice. And considering the  alternative—what farmers call sparing—this makes   sense: if a farmer is to miss a small sliver  entirely while drilling, there will, of course   be no crops, if they miss it with pesticides  or fertilizer, the section’s yield will drop,   if they miss it during harvesting, well that’d  be an expensive and embarrassing mistake, too. 

So farmers overlap. But they try to do so as  little as possible. Experience helps with this, as   hours in the chair, along with a long-established  sense of pride in maintaining straight rows,   keeps overlap down. So too do generally normal,  rectangular fields, should a farmer have such a   luxury. And then there are tricks: spray foams  to mark areas already hit, guideposts along   fences for visual reference, thoughtfully laid out  tractor paths calibrated to align with the width   of the farmer’s equipment. But tricks only go so  far. One study has put numbers on the overlaps.  

Across the study’s 17 locations and four years of  planting, the combine driller overlapped at 7.7%,   spin disk fertilizer at 9.5%, while the sprayer  overlapped at 15.7%, and the cultivator reached   19%. At every step of the process of growing  something, then, the farmer’s overlap is costing   them—8% of their seed is being wasted, 10% of  their expensive fertilizer is being overapplied,   nearly 16% in pesticide and herbicide is doing  more harm than good, and almost a fifth of their   field is being turned over by the cultivator  for no reason. This means more materials. It   also means more fuel, it means more time in the  field and in the chair, it means more hours put   on the machines, and therefore more hours in  the shop and fewer functional seasons. Such  

costs really add up, too, as a bad year will see  costs outpace income, while 10 year averages, in   the case of Kansas farms from 2010 to 2019, will  only net meager 11.8% profit margins. With such   touchy and tight finances, unnecessarily wasting  8 to 19% of one’s time and money on overlapping   is a massive inefficiency. One that Deere was  seeking to address at the dawn of a new century.  The answer was called Starfire which, by  correcting notoriously inaccurate GPS data   with ground location data, offered farmers field  mapping accurate to within 3 feet or 1 meter,   rather than the 10-to-30 foot or three-to-ten  meter accuracy of traditional GPS. With further   work on the product, by 2004, Starfire 2 provided  accuracy within 1.5 inches or 4.5 centimeters.   Through the collaboration with Stanford, NASA,  and Navcom who they eventually acquired, Starfire   positioned the company again on the cutting edge  of the precision farming revolution—not only   did their product provide superior accuracy  for yield and seed mapping, it was accurate   enough to address the fundamental inefficiency  of overlapping. Simply equip a machine with a  

Starfire receiver and a monitor then purchase  Deere’s Autotrac program and farmers could now   guide by precise lines laid out on a screen and  even let the autonomous feature take the wheel.  Today, through a combination of six uplink sites  on three continents, 46 reference sites around   the globe, and leased bandwidth from Inmarsat  satellites, the shovel and pitchfork company   is able to provide greater accuracy than the  public global positioning alternative, optimizing   every single thread and turn across a farmer’s  field. The influence of such guidance can’t be   understated. Only 10% of farmers used any sort of  auto steer and guidance system in 2004, but as of   2019, those numbers stood in the mid 50 to 60%  range, and on bigger, thousand-acre farms where   the economies of scale blunt the upfront cost and  the waste of overlap is only magnified, adoption   rates of such systems have reached over 80%. By applying the same playbook they did with the   plow and the tractor—embracing then perfecting  new technology through upfront R&D investment   while also acquiring sector leaders like  Navcom—Deere helped push farming into a   new epoch. But that epoch isn’t over, and Deere’s  only dug themselves further into the digital turn. 

Across the dozen American companies Deere has  acquired since 2007, only four are traditional   hardware manufacturers, the rest, broadly, are in  tech, and increasingly in artificial intelligence,   machine learning, and automation. In 2017, for  just north of $300 million, Deere purchased Blue   River Technology, who had recently been testing  their new product called See & Spray—what they   called the world’s first smart sprayer, which,  by feeding hundreds of thousands of plant images   through deep learning algorithms was capable of  identifying crops and weeds before then spraying   herbicide within a quarter-inch accuracy. Not  long after, Deere’s See & Spray Select entered   the market. Then, in 2022, See & Spray Ultimate  became available for factory installation on 2023   model 410R, 412R, and 612R Sprayers. With a camera  positioned along every meter of the carbon-fiber   spray boom, the product would reduce spray volume  by two thirds, saving money on herbicides and, by   extending trips between refills, saving time and  fuel. The benefits of the next step in precision  

agriculture also provided an environmental benefit  beyond the farmer too, as this product, the   company projected, would reduce the airborne drift  of chemicals by up to 87% and chemical run-off by   up to 93%. In this new era of AI and machine  learning precision farming, John Deere was not   the first, as a Dyson subsidiary entered the smart  spray space earlier. Nor is it alone, as AgZen,   a commercial outgrowth of MIT research, is pushing  into the space, too. Whether John Deere wins out   here, as they have so many times in the past with  new technological innovations, remains to be seen,   but given their history, it feels like a safe bet. Regardless of competition, though, this service,  

capable of plugging right into the broader  John Deere ecosystem, should be a boon for   the American farmer conscious of cost, yield,  and overall environmental impact of their work.   Or at least, that’s what it would seem. By standard metrics, farming in the US   has gotten better across the sector’s  digital revolution—we’re wasting less,   making better informed decisions,  and growing more than ever: just   look at average yields for corn, soy, and cotton. But consider the position of the American farmer.  

In the past, being an all green farm—that is  running strictly John Deere equipment—was a point   of pride. Today, though, it’s increasingly feeling  like an expensive necessity without alternatives.   New-found hyper efficiency comes with a cost, or  really, a whole host of costs. Say an Illinois soy   farmer is sizing up purchasing a See & Spray  attachment—well, first they’ll need to have a   fairly new sprayer to begin with, which if they  don’t have, will be in the ballpark of $50,000.  

Then add on another $25,000 for equipment and  install, which can only be done at an authorized   dealership. Still, given that soybean pesticides  have reached an all-time high this decade coming   in at $77 per acre in 2022, and given that  this farmer owns the median sized farm for the   state at 4,500 acres, considering scale, such an  upfront investment may well be worth it. Without   See and Spray, pesticides would cost $350,000,  with the product, assuming it cuts spraying   down to a third, the farmer would only need about  $117,000 in pesticides—so with upfront costs,   savings total about $62,000. But then another  cost: See and Spray subscriptions cost $4 per   acre so cut out another $18,000 and the  economics become slightly less appealing.   And then there’s the less tangible costs. Fundamentally, precision agriculture is  

changing what it means to be a farmer. What was  once an occupation defined by individual autonomy,   problem-solving, and improvisation is now  increasingly beholden to monitors, screens,   software subscriptions, and the availability of  manufacturer-authorized technicians. Undoubtedly,   this is an issue of nostalgia, but it  permeates in costly and frustrating ways too. 

Consider the solar storms that pushed northern  lights as far south as the American midwest. While   it might’ve been a once in a lifetime experience  for the farmers who stayed up late to see it,   it caused far more costly problems when the storm  knocked out their navigational systems. Just at   the moment farmers needed to be out planting corn  their precision navigation systems failed them. 

While solar storms are few and far between,  issues with software programs and machinery   that’s now more complicated than ever are far too  common and far too difficult to get figured out   for your life-long farmer. Rather than hauling  a tractor back to the barn to fix a hydraulic   leak and get back out on the field that same  day, when new-era hardware fails, there’s a   good chance a farmer will be out of their depth  if it’s on the technology-side of the machine,   which as the far more finicky side, it likely  is. So rather than fixing it and getting back   out on the field, the farmer’s left waiting for  an authorized technician who will be expensive   to pay and costly on time, as it’s unlikely  they’ll be available at the drop of a hat.  And all that’s without considering whether the  farmer has the tools and information to make the   fix in the first place—which is also a matter of  contention. Across the past decade, John Deere has   found itself in the middle of a battle over the  right to repair. For those savvy enough or bold   enough to fix their own issues, they often need  access to the diagnostic software to begin with,   which is something Deere’s been slow to hand over.  Their stated reason to keep software restricted is  

a matter of liability and responsibility.  If they hand over the keys, they figure,   their machines might get used and altered in  ways they shouldn’t. But for an increasingly   boisterous farming community, this withholding  of key information is simply another way to make   sure that farmer is also on the hook for costly  repair bills that make their way back to Deere,   thus providing the company yet another revenue  stream. Regardless as to who is really telling   the truth, what’s undeniable is that the  farmer is as financially squeezed as ever,   and with the rise of big tech in farming, they are  increasingly being moved out of the driver’s seat. 

But Deere is facing their own financial  pressure—competition is rising, so they   have to adapt to maintain their relevance. Over  the past decade, venture capital money has poured   into startups that insist they can disrupt the  world of food production. Whereas in 2013 there   were 42 funds focused on the AgriFood space,  today there are almost 300 reaching a peak   of $53 billion of investment in 2021. A simple  thesis is presented to potential investors—the   global share of land dedicated to agriculture is  peaking as more and more of the world urbanizes,   yet simultaneously, the global population is  expected to continue increasing for at least   half a century more. Therefore, it is objectively  true that we will have to produce more from less,   and these startups believe the way to do  that is through technological innovation.  Different companies have different solutions  to this problem—some are focused on “controlled   environment agriculture,” growing indoors to  eliminate the threats and resource-losses from the   outdoors; others are leveraging big data analytics  and machine learning to remove the inefficiencies   of guesswork; while still others are working to  increase outdoor production yields and lower labor   cost through various forms of autonomy. Building  on early successes in precision agriculture,  

John Deere has committed to developing a fully  autonomous production system for corn and soybean   by 2030—that means every step from plowing  through planting through harvest without direct   human involvement. And that’s remarkably  believable. Corn and soybean are planted   in straight rows with relatively high distance  between each plant, meaning there’s already the   predictability and margin for error that makes  it easiest for autonomous systems to succeed.  The combination of innovations like  precision agriculture, indoor growing,   autonomous production, and more will yield amazing  benefits for us all: in sum, they create a food   production system that is less expensive and less  resource intensive. But they come at a cost—a very   literal, incredibly significant, upfront cost. The economics of paying an exorbitant amount for  

a fully autonomous wheat production system work  out first for the absolute largest farms. Just   as with See & Spray, every innovation promises to  improve efficiency by a certain, small percent,   so the larger the overall operation, the  more valuable that small percent can be,   and therefore the more likely the upfront cost  is worth it. So innovations can be worth it,   but only if you grow a ton of food. This  has been true for a while—there have been  

greater and greater economies of scale  in agriculture—which has contributed to   a long-term trend of consolidation. Over the past  25 years, the average size of an American farm has   grown by 7% even as total farmland has declined  8%—as small farms face increasing cost-pressure   by more-efficient big-ag operations, they  either shut down or sell their land to big ag.  And as big tech encroaches into farming,  innovation is accelerating, which is great   by itself, but this leaves the small family farm  behind. A layperson’s perception of farming,  

as a mom and pop living in a homestead in Kansas,  working the fields around, answering to no boss   but themselves, is becoming a cinematic fiction. Increasingly, those living in regions dominated   by agriculture work not for themselves, but  for landowners holding hundreds of thousands   of acres. Often, the owners of this land  live time zones away, meaning profits from   production are not spent at the local diner  or car dealership, but rather distributed to   a multitude of investors and left to sit in  mutual funds. This contributes to a further  

gutting of the economy of rural America—one of the  rare ways to build a business outside of cities is   becoming an increasing impossibility, and rural  resources are being extracted for urban gain.  This is, in many ways, inevitable. Tech  innovation, across essentially any industry,   has primarily benefited larger corporations  and incentivized consolidation. Farming,   being so far from urban areas and so culturally  isolated from Silicon Valley, has long been   shielded from these forces. Yet today, John Deere  and others have recognized the upside of bringing   these two worlds together. But this progress will  hurt. The death of the family farm is upon us,   and the autonomous tractor sits just beyond  the horizon, waiting to unleash its destructive   ability to incrementally optimize yields. The reason why John Deere’s See & Spray  

technology is so powerful is that it’s able to  interpret visual information and make decisions   on what a given plant needs in an instant, at  a huge scale. What makes this possible behind   the scenes is a neural network—the software was  fed millions of images and taught to interpret   them as a human would. Neural networks are  fascinating as they’re essentially recreating   a brain in a digital environment, and they’re  also some of the most wide scale instances of   machine-learning right now, so it’s worthwhile to  understand how they work. And for that, there’s   our sponsor Brilliant.org. Their class on neural  networks starts by teaching very basic principles  

through interactive problem-solving exercises  and straightforward, intuitive explanations,   and then goes on to bring these concepts together  into practical applications. I really think they   understand how to make learning more effective  and more engaging, so they’re perfect if you’re   the person who really likes to learn a lot  efficiently. In addition to their classes   on AI topics like neural networks and large  language models, they have dozens more in math,   computer science, data analysis, and other STEM  subjects, each thoughtfully designed. And to make   it all practical, they break everything up into  small chunks that can be completed in those small   moments of your day, either on your computer or  on your phone, making it a perfect way to replace   your scrolling with something that’ll better your  brain. Whether you’re in school and want to brush   up on core principles or in the workforce and want  to better understand fundamental math and science   topics that affect everything, Brilliant.org is  really the best place to learn. And you don’t   have to trust me, you can see for yourself since  you can try everything Brilliant has to offer,   for free, for a full thirty days when you go to  Brilliant.org/Wendover. Plus, you’ll also get  

20% off an annual premium subscription and  help support the channel when you sign up.

2024-06-30 16:26

Show Video

Other news