Water knows

Water starts with a slope, and so does reading it. After a hard rain the water has already left its account of the trip: it ran to the low corner, cut a channel down the wheel track, pooled where the soil went tight, and left the high ground dry enough to dust. The water made no decisions. It obeyed the ground, and the ground kept the record. Nobody drew that map. It was there before the fence posts were, and every farmer on the place knows it by heart: which corner floods first, which ditch runs in April and is gone by July, where the tile drain pops out along the fence line.

Every farm is a watershed in miniature, and water is the only thing on it that arrives from somewhere else and leaves for somebody else's land. You can hold it, move it, or let it go, and all three are decisions a neighbour downstream lives with. The farm never makes water. It chooses when and where the water goes, and that choice is older than any technology that has ever helped make it.

For most of a century the working answer was a calendar. Water went on a schedule, and the schedule came from habit, experience, and the person who sold the pipe. Rod King, of King Family vineyards in the Okanagan, described the end of that arrangement to a British Columbia adaptation project: "We can't do things on a schedule anymore." What replaces the calendar is a real question, and the answer so far is more interesting than either the marketing or the doubt. Water is where the farm's software story runs into ditches, pipelines, and permits.

Where the water already goes

The oldest water technology on a farm is a ditch, and the second oldest is a drain. Both are still doing most of the work. On heavy ground in Ontario, fields are underdrained with buried pipe so that spring melt and rain have somewhere to go, and the same pipe that carries water away in May can be fitted with a control chamber at the outlet so that the water table stays up through a dry spell. The practice has a plain name, controlled drainage, and a plain logic: the field already has plumbing, so the plumbing can be given a setting. Eastern Ontario and Quebec researchers have studied it for years, mostly for what it keeps out of the creeks, and the water it holds back for the crop is the same water the crop would otherwise have to be given.

Manitoba runs the same idea on a different landscape. The province is very flat, the water has nowhere to go in a hurry, and drainage rules are a standing argument there rather than a settled question. Drainage sounds like a boring word until you live downstream of somebody's decision. In Saskatchewan it stopped sounding boring years ago.

One of the largest water decisions in Canadian agriculture this decade is being taken in Saskatchewan, and it is not being taken by software. The Lake Diefenbaker irrigation project is budgeted at $1.15 billion and is moving into construction, and CBC reported in August 2024 that the feasibility study meant to test whether the project is a good use of public money was never completed or released. The reporting put the worry plainly: spending hundreds of millions of dollars to benefit just a handful of farmers. Robert Halliday, a water resource engineer who has studied the Saskatchewan river basin for decades, did not mince it. "It seems like the cart is before the horse here," he told CBC, and, "Faith in government is plummeting."

Two water decisions, then. One province is spending a billion dollars to move water onto more land, without a public answer to whether it should. Another is teaching fields to hold the water they already have, with a gate on a drain. Both are water technology. Neither is software, but both are systems: an arrangement of rules, permissions, and costs that decides what any later software is allowed to do. The gate on the drain and the canal across the prairie were settled by people making decisions about water long before anyone wrote a line of code, and those decisions set the terms for every probe and platform that came afterwards.

What a probe knows

The tools themselves are unglamorous, and they cost less than anyone expects. The most common one is a probe in the soil that reports how hard the roots have to work to pull water, which is a more useful number than how wet the soil is. Hortau, a precision-irrigation company with its Canadian base in Quebec, sells exactly that, soil tension probes that report back to a mobile web app, under a tagline any farmer would recognize as true: "Your eyes into the soil." The same company monitors flow meters and well depth, which is the beginning of an actual water account for a farm. British Columbia runs the public version: Farmwest, a provincial climate service that provides climate information to farmers and irrigators in British Columbia, including the evapotranspiration numbers that turn weather into an irrigation interval. In Saskatchewan, a farm equipment dealership, South Country Equipment, built Crop Intelligence, an app that turns weather station data into a moisture reading and a yield estimate for the season ahead, and now sells it beyond its own brand of iron.

Prince Edward Island built the version that fits its own crop. The province publishes an irrigation calculator intended to assist potato producers in scheduling supplemental irrigation, alongside a permitting process, soil health plans, and a drought contingency plan template for farms. Potato ground on the Island is sandy, the growing season has turned erratic, and the province's own summary of what happened next is unusually direct: changing climate patterns and increasingly erratic precipitation "has pushed many Island agricultural producers to irrigation or expansion of their current irrigation systems."

Donald Killorn, who runs the PEI Federation of Agriculture, has spent the last few years inside that shift. On the Future Herd he described the season that forced it, a "once in a lifetime drought last year in Prince Edward Island that cost our economy $500 million," and what he and his members were asking a data scientist to build: which irrigation investment reduces risk the most, field by field. He also described the week he was living through while we talked, delivering intelligence to his first farms as "the temperature has soared to 35 degrees." The $500 million figure is his own, and the production records run the same way: potato output on the Island fell sharply in 2025, the largest provincial decline in the country.

Add those instruments up and something is still missing. A farm with soil probes, a weather station, and a calculator is not a farm that understands its own water. It is a farm with a better schedule. Every one of those instruments answers a version of the same question, when and how much, and none of them knows where the water came from, where it goes when it leaves the field, or who else has a claim on it. The Okanagan factsheet that carried Rod King's line made the same admission in a single sentence, and it is the sentence worth keeping: "Your first and best tools for making irrigation decisions are your eyes and a shovel."

The dry year is a negotiation

Water is the one input on a farm that cannot be bought in a bag, and in a short year it is not a purchase at all. It is an argument, and Canada ran a live version of it in 2024.

That spring, in southern Alberta, the province's major water users signed voluntary agreements to share the water in the Oldman and South Saskatchewan basins if the season turned severe. It turned severe. The agreements were activated on May 9, 2024, and licence holders who had not signed were asked to conserve to match the signatories. The province had been explicit about the alternative: a licence is a right to take water in a particular order, and when there is not enough water to go around, earlier licences come first. Sharing by agreement is what a basin does instead of finding out what seniority means. The agreements were deactivated in October, having got the season through.

Alberta is also the province that spends real money on moving water, and the money goes into pipes, not platforms. The province's own irrigation strategy describes districts replacing open channel canals with buried pipelines, reducing water conveyance losses and improving water delivery efficiencies, with a target of replacing every technically feasible canal by 2035. That is the state of Canadian water intelligence right now: a canal turned into a pipe, funded by a province, a lender, and the districts themselves, with the savings measured in water that stays in the river.

Jamie Reaume, an Ontario agricultural policy consultant, described how the same question lands differently depending on the region, and the Alberta answer he sketches is the opposite of expansion. Is irrigation needed in Alberta, he asks, and the answer he gives is yes. "But do they have the resources and the water for it? No. Do they have the infrastructure? No." The comparison is not between provinces that have water and provinces that do not. It is between provinces that have already built the plumbing and provinces that cannot, which is the same sentence as the last chapter's drone question with a river running through it.

The frontier is mostly concrete

The claim worth testing is that the next frontier is not merely automated irrigation, but systems that understand the whole farm's water movement. Nobody has built that yet, anywhere this research could find. What exists is a sensing layer that is cheap and largely public, a plumbing layer that is expensive and largely public, and an allocation layer that is legal and negotiated. A whole-farm water account, the thing that would know that the pond is full, the well is low, the tile is running, and the forecast changed, is a diagram in a strategy document and a research ambition, not a product you can buy.

That gap has a shape, and the shape repeats the open-source split from the last chapter, with one important difference. The valve is already a commons. OpenSprinkler, a GPL-licensed firmware with open hardware, runs irrigation controllers without a cloud account, and open crop-water models descended from the UN's FAO work are free to anyone who wants to run scenarios. Free reference weather data exists too, though the fullest version of it is American. The parts of the water stack that can be built and owned by a person with a laptop are real, and they are the parts that decide when a valve opens.

What stays closed is everything the water has to travel through and everyone the water has to be shared with. A buried main is not a download. A licence is not a repository. Which puts water in a different position from the drone and the collar: on this farm input, the open questions are not mainly technical. A probe answers a question about a valve. It has nothing to say about a pipe, a licence, or a river.

The commons that irrigation always was

There is a body of work that explains why the Alberta agreements look the way they do, and it is not about farming software. Elinor Ostrom spent decades studying how communities govern shared resources without either privatizing them or wrecking them, and her field cases were irrigation systems: farmers sharing a canal in Nepal, water courts in Spain settling disputes by custom, groundwater basins in California allocating a shared aquifer among users. Her finding was that durable arrangements look alike in unglamorous ways. The people who use the water help make the rules. The rules fit the local situation rather than a template. Someone the users trust monitors what is actually being taken. Misuse gets a graduated response instead of a lawsuit, and there is a cheap, fast place to settle an argument.

With that list in hand, the Alberta notice stops looking like bureaucracy. Signatories agreeing to conserve before the province imposes seniority, activation dates, deactivation dates, a meeting to review how it went: that is a commons being governed, live, by the people who hold the licences, under the shadow of a rule that would be worse for everyone if the agreement failed. Ostrom called those arrangements polycentric, meaning that the authority is layered rather than held in one place. Farmers, districts, the province, and the basin each have a piece.

Water is also further along the same road as the fence. A drone can be built in a shed by someone with a soldering iron and a grudge. Almost nothing about the farm's water is buildable by one person, because almost all of it is shared with the neighbours before it ever reaches the field.

The system everything else runs through

Water is, on most farms, the most consequential system there is. Yield is decided by it, but so is almost everything else: whether the soil holds together in a heavy rain, whether the nutrients stay where they were placed, whether a herd has shade and a drink in July, whether the tile runs in April and the well holds in August, whether the neighbour's field floods because of something that happened on this one. A season can be lost to a single week of water at the wrong time, which is why every established farm already has an answer for it, even when the answer is a pump, a ditch, and forty years of noticing.

Past generations in this country could take the system mostly for granted. Rain came, the creek ran, the well held, and a licence was paperwork that lived in a drawer. The last few years have made that assumption expensive: a drought in Prince Edward Island that cost the province's economy a half billion dollars by its farm federation's accounting, water sharing agreements activated on the prairies for the first time, and a billion dollars committed to moving water onto more land. Water is moving from the category of things a farm inherits into the category of things a farm manages, on purpose, with instruments, and with rules that can no longer stay in the drawer.

This is where the principles of agroecology stop being philosophical and start being practical. Keep the water on the land instead of rushing it off. Build organic matter so the soil itself holds more of it. Keep the ground covered when the rain comes. Grow roots deeper than the plough pan. Control a drain rather than maximizing it. Every one of those is a water decision, and each of them reduces what the farm has to buy, pump, or ask for. A programmable farm takes its name from the belief that a place can be arranged on purpose, and water is where that arrangement is tested first, because water is the system that comes free and leaves on terms nobody negotiated with the farm. It is also the system that decides who gets to farm at all, since the operations that survive a bad water year are the ones with the capacity to plan for it. A farm that can read its own water, hold what it has, and show the working to a basin, a lender, or a neighbour is a farm with options in a decade that will not hand them out for nothing.

Open questions

In a short year, who decides how the water is shared, and by what rule: seniority of licence, agreement among users, price, or something the basin has not tried yet? And when the rule and the map disagree, which one wins?

Does reading water better mean using less water, or spending the same water more precisely? A farm that irrigates exactly enough, on time, every time, might be a farm that has made a shrinking allowance comfortable.

What standing does the water that leaves have? A creek, a lake, a fish, a town's well, or the farm downstream are all downstream of somebody's decision, and only some of them sit at the table when the rules are negotiated.

And a question for the chapters where the fork gets drawn in full: if the pipe, the licence, and the reservoir stay closed while the probes and the models stay open, does the open layer matter, or does it just make the closed system easier to operate?

Moves you can make

Draw your water map. One day of water through your kitchen, one season through your garden, one year on a farm if you have one: where it enters, where it leaves, and where it stalls. Mark the place upstream or downstream that you never thought about, because the map usually has one.

Read your own slope. After the next hard rain, walk the ground and watch where the water went, then change one small thing and watch again: a mulch patch, a berm, a rain barrel under a downspout, a cleared culvert, a repaired drain. You are doing what the probes do, with the tools the factsheet recommends.

Find the rule where you live. Somebody decides who gets water when there is not enough of it: a licence registry, an irrigation district, a conservation authority, a municipality, a utility, a watershed board. Read one clause of it plainly, and notice how much of your water life was decided before you arrived.

The situation: the map and the rule

Two parts, one evening, a piece of paper and a phone.

Part one, the map. Draw where water enters and leaves one place you know: a kitchen, a garden, a farm. Kitchen version: the tap, the kettle, the drain, the dishwasher, the downspout, the street. Garden version: the rain, the hose, the lowest corner, the storm drain on the block. Farm version: the well, the ditch, the tile outlet, the dugout, the creek. Include the water you do not control, which is most of it, and mark the two places where the water moves fastest.

Part two, the rule. Find out who decides what happens to that water when there is not enough of it. In a city that might be a utility, a bylaw, or a water restriction notice. On a farm it is a licence, a permit, a district allocation, or a basin agreement. Write down the decider, and one sentence from the actual rule, in your own words. Then compare what the map wants with what the rule allows. The gap between those two things is where the ownership question lives.

Run it in a room and the maps will not match, which is the point. Everyone's water arrives from somewhere they have never seen, which is a fact about cities and farms alike, and everyone's rule was written by somebody with a different map in mind.

Water leaves the farm, and so does everything else the farm does not sell. In the next chapter, the leftovers get their own economy: the pomace, the whey, the spent grain, and the businesses built on what one operation used to throw away.