Farmers vs Machines. A 5 Layer Hierarchy
..and what it means for the future of farming.
Introduction
Machines have been taking over farm jobs for almost 200 years and we’ve welcomed it in New Zealand. We first welcomed it because we didn’t have enough labourers and then we welcomed it because we could farm with less people so produce cost less. Now we can get excited about machines because of their capability to assume what farmers have always had to control directly on the machine or program into the machine.
This article looks at this continuing trend through a five layer hierarchy of the skills machines need to operate at any given level of equivalent human complexity.
The Hierarchy of Machine Needs
Job Muscle
Level 1
Jobs that require muscle include tasks that humans have been conditioned to complete over many thousands of years; ever since some hunters became farmers. Our muscles are designed to lift, reach, pick up, put down, squeeze, turn, push and pull. Anything hand made relies to some degree on a mastery of this fundamental skills of being a human.
The bigger the job, the more muscles were required. Not all of the muscles had to be human however. Horses pulling ploughs could also be added in to the equation for bigger muscle power when sowing and harvesting crops.
For most of the 19th century, muscle size and definition equalled usefulness both in humans and animals. When machines came along they could do everything muscles could do except machines did it faster, cheaper and without a break. Sound familiar? Agricultural advances in the United States focussed on saving ‘muscle use’ and so New Zealand followed suit in pursuing directions that moved farmers closer to the ‘one-man operation’ promise.
The first moment for the machine arrived in New Zealand in the 19th century when Europeans brought agricultural machines with them. There wasn’t enough labour to get everything done that farmers needed, so machines filled the gap, rather than replaced the tasks. Since then, machines have made such brilliant companions on the farm that only 5.3% of New Zealanders work in agriculture now thanks to machines replacing muscles.
Job Control
Level 2
Machinery give farmers more control on the farm too. They can control how much feed to dispense to cattle and then thin it back to make it more palatable for sheep. Control of what machines do is a given so we don’t think about it. Instead, we think about growing our mastery of a machines capability to better achieve the outcomes we desire. For feeding stock from bale, minimising bale wastage reduces costs so having fine control over how a bale teases apart instead of simply unrolling it increases the control so an experienced machine operator can reduce the wastage.
Job Program
Level 3
Moving up the Hierarchy of Machine Needs pyramid, we do away with humans directly controlling the machines on the farm altogether. Driverless farm vehicles are move about a farm based on field coordinates programmed into the tractor’s system. The machine can then automatically go about its business of mowing, planting or ploughing.
Human programmed machines means farmers have freed up the time required to control a machine and can now use that time in more productive ways. If they’re sitting in the cab of the programmed tractor, they can now make calls, check emails or check stock auction prices. Additionally, programmed tasks eliminate missing or repeating a strip.
Job Design
Level 4
A human designed task sets the outcomes and lets the machine decide how to achieve that outcome based on gathered inputs. The farmer tells the machine to optimise for certain crop producing levels based on how they have interpreted weather information. Some of those inputs may be farm mapping drones measuring soil and moisture levels that determines what crop applications are required and where. Then the machine designs the plan to deliver water and feed and in what areas for approval by the farmer and with systems like what3words it is possible to accurately address the entire world in 3 metre x 3 metre sections now.
Job Choice
Level 5
This is the ultimate in human and machine direction and the last bastion where humans tasks are unlikely to be replaced by machines. A farmer may choose to plant corn or choose to farm sheep, however if in the 21st century machines mastered the job of choice, we might see entirely new farming practices. For example, we know that Google’s Alpha Go beat its human competitor by inventing a series of moves that initially looked like basic errors that ultimately were winning moves. Apply that to farming and if we let machines choose what the land should be used for, we may see an entirely different set of choices made. This may solve one of the fundamental stumbling blocks of regenerative agriculture which is the complexity it can introduce into farming practices. If that complexity was taken care of by the machines in the same way the complexity of the Internet OSI model layers 1–3 is hidden from Internet users, then maybe this could be possible.
Job Summary
The point of this article is that machines have been steadily replacing ‘inhumane’ tasks for over a century and we have encouraged it. Machines replaced tasks that were repetitive, low value and in many cases inhumane; anything classed as back-breaking for example.
By the time neural networks are ‘brainy’ enough to design better processes than humans we will welcome it they way we always have. When will that be?
Well the existing neural net weight is about 100 million which is around 1 million times less than the human brain. At current development speeds, the neural net will have caught up within 20 years.
Until then, we’re stuck with lower capability machines and humans will continue to be required at choice, design and program layer of the hierarchy of job needs.
What farm job tasks do expect machines to take over next?
If you liked this article, you’ll love the podcast Anatomy of Next on Superintelligence.
Enjoyed this article on the future of farming? Consider reading the book ‘100% Kiwi Business’ available at umprint.net
