Learn to do by doing

You can learn almost everything a small farm needs to know without leaving your couch. A person in a city apartment who has never touched soil can, in an afternoon, watch a stranger dig a drainage line and explain why it fails, watch another cut open a tomato that blight got to and name the hour they should have acted, watch a third walk through the first month of a first flock of hens, mistakes and all. None of it is a course. It is the primary material: people doing the work, in their own words, on camera. And the people watching do not stay passive about it. They copy, argue, try the thing on a balcony, get it wrong, and watch the video again.

This is the part of agriculture's learning system that nobody planned and everybody now uses. YouTube and TikTok have become the sector's unacknowledged knowledge infrastructure: the place knowledge is made mobile, translated from one field or one kitchen to another, and handed from the person who has done it to the person about to. Where a curriculum once decided what a novice should learn, a feed now offers it, and the novice is the one choosing. The sources are primary most of the time, the farmer or gardener with the road-tested trick telling it themselves.

The arrangement looks new, and the screens are new, but the shape is one of the oldest in farming: a person who has made the mistake showing the person who is about to. What that shape looked like when it lived in a room, what it looks like now that it lives in a feed, and why the future of it is also, in the oldest sense, its past.

More knowledge, less learning

Agriculture is usually described as a production system, and the description is not wrong so much as partial. Underneath the yields and the inputs sits a second system, and that one decides whether any of the first system's improvements ever take hold. The second system is learning.

Learning here does not mean school. It means the ordinary way a trade reproduces itself: watching, copying, asking, getting it wrong, being corrected, trying again in front of somebody who has already made that mistake. The important thing about the process is that it is social. It runs through people who know each other, and it needs to happen more than once.

For most of the last century, this system had visible form. Local clubs, extension services, fairs, colleges, and the neighbour who had learned the hard way. From a distance those looked like extras, a soft layer around the serious business of production. They were the infrastructure, in the same sense that a road is: nothing about the harvest depends on the road until the road is gone, and then everything does.

That is roughly where Canadian agriculture stands. It has more information available to it than at any point in its history, and a persistent, hard-to-name sense that it is not learning as fast as it should. The research is published. The tools exist. The practices do not spread evenly, and the same conversation about adoption happens every year with the same result. The problem is not knowledge. It is transmission.

The room where it used to happen

Barb Scott-Cole is a leader with 4-H Canada, and on the Future Herd podcast she described what the older version of that system looked like, starting with who was in the room.

"In the early, uh, 20th century, a lot of people never finished high school. They were lucky to, to get a few years of elementary school in."

Formal education was thin, so the learning happened somewhere else: "by having these small local groups where you could learn, 'cause people passionately wanted to learn, um, how to do things better and socialise as well." The two motives are in the same sentence, and that is the part worth keeping. The room was for skills and it was also for company, and the combination is most of the reason people kept coming back to it.

4-H kept a phrase for the method, and Scott-Cole uses it in the plain form. "it's that learn to do by doing." The motto sounds like a slogan until it is taken seriously as a description of how a trade is transmitted. Doing the thing, in front of somebody who can see what you are doing wrong, is the mechanism. Nothing has replaced it, including the very good instructional material written since.

What got dismantled

The Canadian version of that system was not destroyed in a single decision. It was defunded, reorganized, absorbed and allowed to drift out of reach, one budget cycle at a time.

Todd Ormann works at Olds College in Alberta. Speaking on the Future Herd podcast, he described an institution whose founding purpose is the cleanest available statement of the original idea. In 1913 it was, in his words, "a research farm to help teach basically Europeans how to farm". A college built to teach farming, in the countryside, with the farm itself as the classroom.

He is candid about what the modern version is up against. Post-secondary education does well at what he calls "chalk and talk where you put two to 300 kids in front of a room and one instructor", and it does that well because it is affordable. Teaching agriculture is not affordable that way, because it needs a lab, a field and a barn. The cost structure pushes in the opposite direction from the subject.

Then he describes the gap directly. The sector assumed for a generation that private companies would carry the research burden, and the assumption mostly held until it did not. "as we dismantled some of our research uh capabilities some of those things were not gonna fill the gap". The institutions that remain are not connected to each other the way the old system was, which is the kind of sentence a person says slowly.

What this leaves is an education system whose problem is not quality. Ormann says his graduates are employed: "We have an issue with finding students". The pipeline is thin at the intake end, in a country where "1.5 to 1.7 of the population is now still in agriculture". Fewer people farm, the teaching institutions compete for a smaller pool of young people, and the knowledge they carry has to be handed over faster than the pool is refilling.

Safe to fail

Scott-Cole is precise about the condition that makes transmission possible, and it is not enthusiasm. It is safety.

"create that condition where it's safe to learn and it's safe to fail".

The adults have to go first, and they have to be seen getting it wrong: "don't be afraid, afraid to say, you know, I really messed up on this, you know, when I was learning it, but this is a trick I learned." That is the method in one sentence. The trick is only useful because the failure is visible ahead of it, which is why the person demonstrating has to be somebody who has already made the mess.

Jennifer MacTavish is the Interim Executive Director of the Agricultural Adaptation Council, and in her own Future Herd conversation she made the same point from the other direction, about organizations rather than classrooms. She repeats a line from her own board chair, approvingly: "It's okay if you're making a mistake, it means you're doing something." Then the standard she holds herself to when a colleague brings her a problem: "You made a mistake and it's not life limiting and this is how we're gonna fix it."

Where that condition is missing, people do not learn. They hide. MacTavish has had the conversations with people who were never given that level of support, and the behaviour that results is easy to mistake for caution. It is a slow leak of capability, and it looks like a personality.

Who carries it forward

Learning that stops at one person is not infrastructure. It is a hobby. The test is whether what somebody learns gets taught again, and both guests describe the loop closing.

Scott-Cole's version is generational, and slightly impatient: "maybe by next generation we're gonna have adults that are able to teach it and come back and help you with that". Her proposal is concrete and cheap next to most of what the sector funds. "I would have an outdoor classroom in every school with a, a growing area to start plants." She is not arguing for agriculture as a school subject. She is arguing for agriculture as the setting where teaching happens, which is a different claim, because the setting is what produces the questions. She came away from watching a group of children work struck by "the questions they were asking, um, were ones that were really insightful", and questions of that kind are the one thing a curriculum cannot supply.

MacTavish's version is a habit rather than a program: "the theme of continuous improvement, continuous learning". In her organization that means building the review loop into how the work gets done, instead of treating learning as an event that happens at a conference and ends when the coffee runs out.

Her harder claim is about succession. The conversation she keeps having is with people staring down retirement and asking where their replacement is, and her answer is that the replacement is the thing they were supposed to be building all along: "the best legacy they have is a good succession plan". Then she names the point of all that training: "Mentor, they're, they're their replacements." The stutter is hers, and the word after it is the one she has to argue for, because the people who need training are also the people their teacher will be replaced by.

She does not draw the circle at her own organization. "But do they have to be in our organisation or can we mentor them into a role where they're just in our other department?" Her reasoning is the size of the sector: "Like our industry is so small, we don't lose them. When they go to another org, they just move to our other department, if you will." A sector that treats every departure as a loss cannot afford to train anybody, which is a strange way to run an industry that is short of people.

The feed is the room

The old room assumed the learner had to be somewhere. The new one does not. Alissa Overend is an Associate Professor at MacEwan University who studies how people actually come by food and the skills around it, and she states the change as a plain fact: "Obviously, you can learn a lot about gardening on YouTube and the internet, so we don't need a lot to be able to grow." A novice no longer needs a curriculum, a club, or a teacher who happens to live nearby. They need a phone, the name of a plant, and the willingness to watch, and the primary sources are right there, in the farmer's own hands and words.

The part that matters most is not the how-to. It is that the feed is safe to fail in, in a way the old room never reached. Scott-Cole's condition was that a learner could get something wrong without shame, in front of people who had already made the same mess. A feed goes one further: it lets you watch somebody else make the mess, in full, and be entertained while you absorb the lesson. The people who become popular in this world are often the ones willing to show the mistake and the fix in the same video, because the mistake is the thing nobody else will be seen making, and the mistake is the tuition. You inherit somebody else's, and you did not have to pay for it with your own crop.

What literacy is for

This is where the old system and the new tools meet.

A farmer who has just been handed a dashboard does not need training in the dashboard. Training is what the vendor supplies, and it covers features. What the farmer needs is the older thing: the ability to tell what a claim about their own operation can and cannot support.

Gerd Gigerenzer's work on risk literacy is a useful corrective here, because it treats judgement as compressed experience rather than as the opposite of thinking. A person who has farmed for twenty years knows things that they cannot fully articulate and that no dashboard contains. The question a new tool poses is whether it sharpens that judgement or replaces it with a number that has been stripped of its uncertainty. Read a yield prediction that way and the interesting part is not the estimate. It is the error range, and whether anybody will tell you what it is.

Gigerenzer calls the thing that twenty-year farmer carries a heuristic: a rule of thumb so compressed by experience that it no longer needs the reasoning it grew from, a judgement you reach before you can say how. His larger claim is that this is not a second-rate substitute for careful analysis. A person who has watched the same situation end a thousand ways decides fast because the situation has already taught them, and that fast rule is often more reliable than a fresh calculation that treats every case as if it had never happened before. The feed matters to this because it makes other people's experience portable. Watching a stranger catch a problem early, or miss it and pay for it, is a heuristic you acquire without paying the cost yourself, and that has always been how a craft travelled, just never at this distance.

The catch is the one Gigerenzer attaches to every rule of thumb. A heuristic is fitted to the place it grew in, and a clip of somebody's field in another province, on another soil, in another season, may not bring its conditions with it. Judgement still has to be grown here, on the reader's own ground. The feed can hand you the rule. Only the farm can tell you when it applies.

Put the question in the plainest form available. The system in front of you is either a tool, which you own, direct and can switch off, or a tenant, hosted on your operation and paying its rent in your data, or a landlord, a subscription you cannot farm without and cannot inspect. Sorting those three apart is a literacy question and not a technical one. It asks what you can look inside, what happens to your data, who can switch the thing off, and whether you could leave. None of that has anything to do with operating the software.

Reading the world

Paulo Freire spent his life on the difference between being taught and learning, and his criticism of what he called banking education names the failure mode directly. In the banking model, knowledge is deposited into a person who is expected to store it. The learner is a container. Freire's alternative was dialogue, and his phrase for the goal was reading the word and the world, which is the same claim Gigerenzer makes from the psychology side: understanding is something a person builds, out of their own situation, with somebody else in the room.

The political edge of that idea has not dulled. A sector that receives its knowledge as delivered product, from institutions that decide what counts, is being taught. A sector whose practitioners can question the delivery, check it against what they know and pass on what they find, is learning. The difference between those two conditions is not funding levels. It is whether people are treated as the authors of their own formation or as the destination for somebody else's conclusions.

Freire would have recognized the feed as a version of the thing he argued for, and he would have seen its limit. A novice who learns from the practitioners themselves, in their own words, and can answer back in the comments, is not being banked. They are choosing the sources and joining the conversation, which is reading the world rather than receiving it. But a feed has its own way of banking a person again. An algorithm that keeps serving more of what already held your attention is another delivery system, one that decides what counts on your behalf, and Freire's old test applies to it unchanged. The question was never which institution does the delivering. It is whether you are the author of your own formation or the destination for somebody else's conclusions, and now the somebody can be an algorithm instead of a curriculum.

What a tool cannot teach

One thing follows from all of this. A tool can be installed in an afternoon and taught in an hour. Judgement accumulates over seasons, and it accumulates in people, which means every tool that arrives on a farm arrives with a teaching requirement attached that no vendor is set up to meet.

That is a practical problem before it is a philosophical one. A farm that builds its own tool has exactly one maintainer, and the open question left hanging over it is who that person will be in ten years. Maintenance turns out to be a teaching problem. The person who can repair the tool is the person somebody once sat down and showed how it works, including the parts that failed.

The Situation: one person, one tool, one month

The Situation for this chapter takes a month, and almost none of it is spent in front of a screen.

Choose one person and one tool. The person should be somebody who will actually use the thing: a neighbour, a hired hand, somebody in your family, a vendor at the same market table, a new member of your co-op. The tool should be one you already know how to use, ideally one you struggled with.

Then teach it the way that room taught: sitting together, in person, with the tool doing real work on a real task. Do not begin with what the tool can do. Begin with the job the person already does badly by hand, and show the tool doing that job.

Keep a page. Every time they get stuck, write down what they got stuck on, in their words and not yours. Those lines are the real manual, and the manual the vendor shipped is not. At the end of the month, ask them to teach it to one other person, and watch what they leave out. What they leave out is what you never really taught them.

Moves you can make

If you make software for farms, the manual is not the product. The person who can teach it is. Budget for that person, or accept that your product's real adoption rate is the number of farmers who happen to enjoy figuring things out alone.

If you fund this work, or regulate it, notice that the cheapest thing to fund is a room. Webinars scale and rooms do not, which is exactly why the rooms are the ones that get cut, and why the learning that only happens in a room is the learning that stops when the funding does.

If you farm, the person you teach this month is your maintenance plan. That is not a metaphor. It is the only answer the maintenance question has.

If you know a thing and own a phone, film the mistake and the fix in the same short video and put it where the next person can find it. You are somebody's teacher, the one the feed has been waiting for, and the person who watches you get it wrong and then right is the person who will teach the one after them.

Open questions

Whether the room is being remembered better than it was. Every generation believes that the way it learned was slower, more human and more effective, and there is no way to test that belief from inside it. The evidence here is testimony from people who were there, which is strong evidence about what the room did for them and weak evidence about whether the system was better than what has replaced it.

What happens to transmission when the screen runs one way. The feed lets a learner watch a stranger get it wrong, which the old room did too, but the old room also watched the learner back, and the correction travelled in both directions. Whether a feed can supply that second direction, the teacher watching the learner, is the question the screens have not yet answered.

And whether learning can be funded at all, in a system that pays for equipment, software and yields. Learning is diffuse, slow, and hard to attribute to a budget line. That is why it is the first thing cut, and why cutting it does not show up in any of the numbers until years later, when the pipelines are empty and nobody can say exactly when that started.

Each one teach one

Everything in this chapter reduces to one shape, and the shape has a name older than the institutions that forgot it. Each one teach one. The person who has learned something turns and hands it to the person beside them, who will turn and hand it on, and in the handing both of them change places. The student becomes the teacher because teaching is how you find out what you actually know, and the teacher becomes a student again the moment the next question lands.

Nothing in the last century improved on that shape. The room, the club, the neighbour across the fence, the stranger in the feed, are the same thing at different distances. When the shape was visible as a room, people could see it was worth keeping. Once it moved onto screens it looked like content, and content is not something anybody thinks to fund or fix.

The hopeful part is that this looks like repetition and is not. The future of learning on a farm is the past of it, with better reach. A novice who can watch any farmer in the world get it right and get it wrong has walked into a room with no walls, and everyone in that room is somebody's teacher, including the ones who do not know it yet. The last chapter of this book gathers the tools. The way they will actually spread is the oldest one there is. Each one, teach one.