Runners
By Ricardo Vega
Two in the morning, and the man in the driver's seat has been sober for nine weeks.
He's not a volunteer, exactly. He's not staff either. His phone buzzed at eleven forty with a pickup two miles away: four trays, one bakery rack, load window twenty minutes. He drove there, opened the back door of a strip mall he'd never had reason to enter, loaded a plastic crate into a car with a bad rear shock, and drove eleven minutes to a lit parking lot behind a church annex where two people were waiting to receive it. The whole thing took forty-five minutes. His phone logged every step of it without him doing anything.
He'll do three more this week. And somewhere in a folder, without either of them ever meeting, a shift manager now has a document proving eleven hundred pounds didn't go into a dumpster.
That's the machine. Let me take it apart.
Start with the provider, because that's where every surplus programme dies. I said earlier that liability isn't the wall — Good Samaritan protection has existed since 1996 and most managers half-know it. The wall is that binning takes thirty seconds and donating is unbounded. So the design constraint isn't persuasion. It's that donating has to become a thirty-second solved problem too.
Concretely: a designated spot. Not the bin, not the loading dock in general — a specific crate or a specific shelf in the walk-in, agreed once, and after that it's muscle memory. The manager doesn't call anyone. He doesn't negotiate a pickup window or vet a stranger. He puts food in the spot and taps once to say roughly how much is there. Done. Everything after that is somebody else's problem, which is the only version a person at the end of a ten-hour shift will actually do more than twice.
Then dispatch. The radius is not a design choice, it's arithmetic — the decay curve from the last piece sets it. Whoever can get there fastest inside a range short enough that the food arrives as food. Which means the system is not a fleet. It's a lot of people who happen to be near, with an empty back seat, at an hour when almost nothing else is moving. Eleven at night is the worst time to schedule a workforce and the best time to find one, because the people awake then are awake anyway.
Now the part I think nobody's built, and the reason this piece is named after them.
The runners are not incidental labour. Every surplus-food project I've looked at treats collection as a cost to be minimised — an unfortunate gap between the food and the need, filled by whoever can be guilted into a Prius. But the gap is the interesting part, because the thing it requires is exactly what a certain kind of person is desperately short of.
Structure. A reason the phone buzzes. A place you're expected. Forty-five minutes where somebody is counting on you and the task is unambiguous and finishable and unmistakably good. Not a job with an interview and a background check and a reference from an employer you burned. Not a favour someone's doing you. A shift you took, a route you ran, a record that says you did.
There's a whole population of people who need precisely that at precisely that hour, and who are currently being handed a coffee and a folding chair and told to keep coming back. That's not a criticism of those rooms — they save lives. But there's a gap between the meeting ending at nine and the rest of the night, and the gap is where people lose it. Filling it with something that has a clear beginning and end and somebody at the other side waiting is not a nice-to-have. For some people it's the whole ballgame.
Which gives the system a property I didn't design and only noticed later: the labour problem and the purpose problem solve each other. The food needs legs. Some people need somewhere to be at midnight and proof they were useful. Neither of those is charity toward the other. It's two shortages cancelling.
Then the ledger, which is the least romantic part and possibly the most important.
Everything that happens in this system produces a record as a by-product. Not as extra work — as exhaust. The manager taps once and there's a timestamped entry: what, how much, when. The runner accepts and there's a dispatch record. Delivery confirmed at the other end. Nobody writes anything down, nobody chases a signature, nobody files a report at the end of the month.
And that exhaust turns out to be worth more than the food, in a sense, because it's what makes the whole thing legible to institutions.
For the provider, a monthly report: pounds diverted from waste, disposal costs avoided, documentation adequate for a tax deduction if they want it. Suddenly the district manager isn't asking why a stranger was at the back door. He's looking at a number that makes his own numbers better, and the programme survives the manager who started it moving to another store — which is how most of these things die.
For the runner, verified hours. Time, place, duration, confirmed by two other parties. Whatever those hours need to prove — to a programme, to a sponsor, to a family that's stopped believing him, to himself on a bad Thursday — they're proved by something other than his word, which is the thing he has least of and needs most.
For the neighbourhood, a public number. This many pounds, this many nights, this many households. Which is the only currency a council meeting understands, and the thing the next piece depends on entirely.
And for the organisation, the honest one: proof of whether any of this works. Not testimony. Not a photograph of a grateful person, which I refuse to trade in. Pounds, routes, nights, and the ability to see when a route stops running and go find out why.
One thing I want to say plainly, because it's the difference between this and an app.
The runner is not a gig worker and the system must never drift into treating him as one. There's no rating, no ranking, no deactivation for a slow week — because a slow week means something in this population, and a system that punishes it automatically is a system that abandons people at the exact moment the whole point was to notice. Nine weeks is fragile. If someone misses three nights, the correct behaviour is for a human being to call, not for an algorithm to reassign.
That's not sentimentality bolted onto a logistics platform. It's the entire thesis of this series in miniature. The catching function is what we never built. Building a dispatch system that quietly drops people is rebuilding the thing that failed, with better tooling.
So: food that would have been binned, moved by people who needed to be moving, to a place I haven't described yet — and that place is the last piece, because where the pipe ends is the same question as who gets to decide, and that's where all of this has been heading.
We discover what people love — then we use technology to help them do more of it, and teach them to bring others along.
Join the Coalition