Fencing becomes software

There is a fence line in Alberta that is not there. A rancher opens an app on a phone, draws a boundary across a pasture, and commits the line. The cattle walk up to it and stay inside it. There is no post, no wire, nothing you could touch. The fence is a voice inside a collar that the animals have learned to hear.

The same field, the same herd, and two completely different situations. This fence is maintained, updated, priced, and can be switched off by somebody else. And it can be moved by someone who never has to stand out in the rain to put it up.

This is where the first three chapters have been going. The drone watched the field from orbit. The cameras and microphones went into the barn and started a conversation with the animals themselves. Both of those were, in the end, instruments of observation. Seeing is humble work, whatever the press releases say; a camera recommends. A fence commands. This chapter is where the technology reaches out and touches the animal, and it arrives first, of all places, as a subscription.

What barbed wire used to cost

Barbed wire is one of the oldest contracts in agriculture, and it is what the invisible fence replaces, in labour terms: you agree to work, every year, in kind, for the rest of your life. Wire rusts. Frost heaves the posts. Trees come down in August windstorms and take a quarter section of boundary with them. Livestock lean on things and stretch them. Every paddock division is a chore you have scheduled into your week whether the week wants it or not.

And the fence is not new to being revolutionary technology. Barbed wire was patented in the 1870s, and contemporaries argued about it exactly the way people now argue about collars: cheap, effective, and the instrument that ended the era of open grazing and started the era of the drawn boundary. Every generation of fencing has been somebody's unholy software update for the countryside. The novelty this time is not that the fence changes. It is that, for the first time, the fence is a service.

Rotational grazing makes that contract worse on purpose, and better on purpose. You move the herd often, in small pushes, so the grass gets rest and recovery; the practice is praised by agronomists and avoided by tired people, because it means walking out with a spool of polywire in the rain. This is the labour the collar was designed to eat.

Brianna Elliot researches livestock technology at Olds College's Technology Access Centre for Livestock Production in Alberta, and she describes the mechanics of the substitute from inside a study funded by NSERC, comparing virtual fence collars against conventional electric fencing: "You just draw a paddock on your screen that marks out the area in your pasture you want your cattle to enter. It'll take about half an hour to communicate that boundary to them and then they are locked in based on the GPS."

Half an hour, and the fence moves. For a certain kind of farm, a whole class of evening work has just been deleted, and the reason it was deleted is a line drawn on a screen by someone who, until the app opened, was expected to go and stand where the fence goes.

What the trials say

The companies making virtual fences have excellent marketing, good satellites now, and a founder story from New Zealand you can watch on YouTube all evening. The evidence trail that is not written by vendors is shorter, real, and still open. Here is what that evidence actually says.

The best synthesis so far is a 2024 review in the Journal of Animal Science that pooled thirteen pasture studies of cattle wearing virtual fence neckbands, beef and dairy, and found no significant welfare differences on most measured indicators. Some of the findings are genuinely surprising. Faecal cortisol, the stress hormone that accumulates in manure, came in lower for animals living under virtual fences than under physical ones. The ratio of shocks to warnings dropped steadily over the trials. The animals learned, and the more they learned, the less the fence touched them. The tone becomes the fence; the shock becomes a memory.

The same review holds up its own end of the mirror. Thirteen studies is thin, most ran for weeks rather than seasons, and the authors write that the evidence is "not yet sufficient" to guarantee the technology cannot affect the welfare of certain cattle. One dairy trial had to be stopped early, because the neckbands rubbed a few cows raw. A fence that exists only as a signal is still a device strapped to a living animal, and living animals wear things differently.

Canada is doing its share of the waiting, in public. The Olds trials compare eShepherd collars against conventional electric fencing inside a rotational grazing project, on purposely ordinary ground. In September 2026, the same centre became the first research facility in Canada to evaluate the newest generation of Halter collars: eighty cow-calf pairs on the college's land, plus thirty-five pairs and thirty-five yearlings running through a working partner ranch. The institution's own statement of the problem is blunter than anything a vendor would publish: the new design "has not yet been tested in Western Canada, where terrain variability, large pasture sizes, climatic extremes and inconsistent cellular coverage may influence system reliability and adoption outcomes."

And the researchers keep one hand on the brake. Elliot's verdict, asked about the labour savings everyone else calls revolution: "This is going to be the biggest benefit of this technology, particularly given the challenges in finding skilled agricultural labour. I don't think this is a fit for every operation by any means, but I do think it's a great tool that can have a pretty substantial impact on a producer's operation." Not for every operation. By any means. When a funded researcher can say that in a press release, the technology is still young enough to be telling the truth.

The first Canadian producer to live through the winter with the newest system, Mike Hittinger, ran about two hundred cows on eighteen hundred acres near Westlock, Alberta, and told The Western Producer that the early days were early enough that "the app wasn't even available in Canada yet." Once it was, he got the answer every tool promises and few deliver: "I can do it from wherever I am." He used it to feed swath and bale grazing through a winter without "wading through the snow" for the wire, without frozen posts, without the cattle pushing down the electric line. He also kept two caveats, the way producers do. The first was the training period: you cannot simply hand a herd to a voice in a collar; there are days at a physical fence while the lesson is learned. The second is the one that tells you where the technology still is and is not. Even with every collar working, Hittinger keeps his perimeter fence. Because when the weather truly arrives, wind, lightning, something with teeth, cattle will run, and a cow that has decided to leave does not know the fence is software.

The fence is a subscription

Now the other shoe, which in agriculture has always been the same shoe: the money. Virtual fencing is sold as hardware plus subscription. The reporting on the Canadian launch puts the collar at $97 apiece and the towers that carry the signal at about $6,000 each; the competing Gallagher collar runs three fifty to four hundred. Then the boundary itself is billed. That is the shape nobody has fully digested. Barbed wire was a capital expense you could steal from by cutting it, an ugly freedom. A grazing boundary that arrives as a monthly line item is a new kind of object: the concept of over here, invoiced.

That phrase is the book's subject in miniature. A software fence can be moved in seconds, which barbed wire cannot, and that is the entire point. But that same quality means it can be moved without you, repriced without asking, paused when the server is down, and discontinued on a schedule set in a city you do not live in. Nobody owns "here" any more; you subscribe to it. The rancher who cannot reach a boundary in an app, in a storm, with a dead tower, has a fence that is somebody else's problem wearing the costume of their own.

The price, though, is the interesting part: six-thousand-dollar towers, ninety-seven dollar collars, a subscription per head. This is the arithmetic of infrastructure, and infrastructure is the classic object of a co-operative. Chapter two of this book met the farmers' organisation that put the gateway on its own roof for exactly this reason. Nothing about the fence is beyond that shape. The reason the fence arrives as a vendor subscription rather than a shared tower is not technology. It is what the market builds by default, and defaults can be declined.

This is the moment in the book where the fork in the road stops being a metaphor, and you will get the full argument about it in the chapters ahead. What belongs here, in the pasture, is one concrete observation. Every technology in this book so far could, in principle, be made small and open by somebody with a soldering iron and a grudge. The fence cannot. The collar is a device touching an animal, certified and insured and validated for behaviour in ways a soil sensor is not. The signal needs towers. The satellite needs the constellation. Which brings us to the open-source question for the fourth time, and this time the answer is different.

The open fence is not there

The open agricultural hardware movement, which will get its own chapters soon, has done remarkable work on everything in this book except the things that act. There is an open-source virtual fencing project called OpenFence, and it is exactly what the community makes when it wants a thing it cannot buy: a GPL-licensed, solar-powered collar prototype, plain documentation, and not much deployment; its own repository calls it a work in progress, and it has been quietly sitting there. There is a sub-$100 open GPS collar out of academic range science that tracks cows across public grazing allotments beautifully, which is to say: it is a fence with no command. It can tell you where the animal is. It cannot tell the animal where the fence is. And there are excellent open-source platforms for planning grazing, mapping paddocks, and scheduling moves, all the way up to the moment the plan has to become a signal.

Observation, open. Command, closed. If you want one sentence that explains where power sits in agricultural technology right now, it is that one. The parts of the stack that can fail safely have been captured by commons and cooperatives. The parts that can hurt something, or that insurance companies and regulators have wrapped around the manufacturers who will warrant them, are subscription hardware. The drone can fall out of the sky and nobody is liable. The fence cannot fall.

So the software fence is a genuine gift and a genuine hostage, in the same delivery truck, and this book has been promising not to pretend those are different chapters. The herd gets rest and recovery that a tired rancher could never schedule. The rancher gets the evenings back. The rancher also gets, for the first time, a farm boundary that has a vendor, a terms of service, and a road map. What the animals get, exactly, is still being measured in Alberta, which is as close as anyone can point right now.

Open questions

• What is a boundary, legally and practically, when it lives in a company's database and moves at a tap? Who is liable when a virtual fence lets a herd into a highway, and who was liable when barbed wire did?
• Is a technology that only works with towers, satellites, subscriptions, and cellular coverage a tool for remote places, or a filter on who gets to live remotely?
• What do the animals make of it: a fence that cannot be sniffed, tested at the nose, remembered as a landmark? The trials measure cortisol. Nobody has measured consent, because consent is not a thing the market pays for.
• Where do the movement trails go? Twenty-four hours a day of position data on every cow, held by a company, has an obvious use for the farmer and less obvious uses for everyone else. That question belongs to the data chapters; the trail is one of the first things this book's technology has ever kept about an animal.

Moves you can make

• If you keep animals and want to see this working without buying anything: Olds College's centre publishes what it is testing; a farm tour or a field day costs a day out and no capital.
• Price a subscription against a chore. The correct comparison for the fence is not free wire; it is the evening you no longer spend in the rain. If the subscription costs less than your evenings, the analysis is the same one the ranchers are making.
• Before the dependency question becomes academic, ask the vendor the questions the price cannot answer: what happens to my boundary in an outage, what does this cost in year five, and what do I have to buy again if you leave the country.
• Learn the failure mode of anything you depend on that you did not build. If you cannot explain how your fence works when the signal does not, you own a ritual, not a tool.

The situation: write the manual page

Somewhere between the drone and the dinner table, the things we own have started shipping with terms of service. This one is a two-part game with a pen.

Part one, on paper. Somewhere on your property, in your building, or in your kitchen, there is a thing that used to do a job by physics and now does it with software: a thermostat, a tractor with a subscription, a car that bills you by the month to turn on its heated seats. Write the manual page the vendor did not ship: the one you wanted. Fine print encouraged: warranty void if thought about. Feature arrives in next update, price changes at next moon. Your boundary is backed by our cloud; the cloud has its own weather.

Part two, out loud, with whoever will sit still. Swap pages. Then answer one question together about each object: after the manual page, would you still buy this, and what would the vendor have to promise on paper, in the warranty, in law, for the answer to change to yes. Write the sentence you would need. It will not fit on a manual page, which is the point, and which is also, incidentally, the topic of the rest of this book.

The next fence line

The water, next. Because a fence that moves is one half of a farm's geography; the other half is what runs through it, and there is a version of this chapter's sentence, about a boundary becoming software, that is already true of irrigation valves in fields that used to flood. The herd, the wire, the water. A valve that opens because the soil, the forecast, and the creek, all three, have said so; the drought you can schedule around; the flood you cannot. But that is the next chapter, and it starts where every farm starts, with what the land is made of.