What I Almost Built

By Ricardo Vega

I want to describe an idea I had, in enough detail that you can feel why it seemed right, and then show you where it breaks. Not as a confession — as an argument. The flaw in it is not obvious, and I think a lot of people building things in this space are standing exactly where I was standing.

Start from the premise I still believe. The formal channel for reading a neighbourhood's will is broken: a tenth of a percent, skewed old, skewed owner, tilted toward opposition. If you want to route resources into a community, you need to know what that community actually wants, and the beige room cannot tell you.

So: go where the ninety-nine point nine percent are. They're on their phones. There exists, right now, an extraordinarily refined machine for reaching exactly the people in exactly the square mile you care about, and for measuring their response with precision no public hearing has ever approached. It's how everything else in your life gets sold to you. Nobody has pointed it at civic participation.

Here was my reasoning, step by step.

One: participation fails because the cost of participating is too high. An evening, a drive, a parking space, the nerve to speak into a microphone. Lower the cost to a thumb-tap and participation goes up by orders of magnitude.

Two: a tap is a real signal. Not a deep one, but real — and a shallow signal from five percent of a neighbourhood beats a deep signal from a tenth of a percent, because representativeness matters more than depth when you're trying to find the will of a group.

Three — and this is where it went wrong — a signal only exists if something provokes it. Nobody taps on neutral. So what provokes? Outrage. Disgust. The sense that something is about to happen to you. And the imagery available was right there: a person defecating in the fountain at the landmark down the road from you. Not to say those people are vile. To say if this goes unaddressed, this is what your neighbourhood becomes. Then the tap. Then: so what should we do about it?

Fourth step: harvest the responses, and now you know what that neighbourhood will tolerate, fund, and site. Vectorise the resources accordingly. Participation rate through the roof, real cross-section reached, actual will discovered. City hall could never.

I want to be clear that I didn't think of this as manipulation dressed as civics. I thought of it as an instrument. Society already holds people on the street in contempt — my organisation is named after that contempt, deliberately. I wasn't creating the disgust. I was borrowing something that already existed and pointing it at a productive question.

Here's the flaw, and it's not the one you'd expect from a lecture about ethics.

The instrument doesn't work.

If you provoke with disgust, you measure disgust. That's not a moral objection, it's a measurement one. The response you collect is not a reading of what the neighbourhood wants — it's a reading of what the neighbourhood wants ninety seconds after you showed them a photograph engineered to make them afraid. You've induced the state you're claiming to observe. Every data scientist reading this recognises it: I designed a survey where the framing dominates the response, and then planned to treat the output as ground truth.

And it compounds. Because I'd have taken that contaminated signal — a neighbourhood in a defensive crouch — and used it to decide where to site services. The reading would come back: put it far away, keep it contained, make it invisible. Then I'd have routed resources according to the will of a community I had personally frightened, and called it democracy because the participation rate was high.

That's the part that actually chilled me. Not that it was ugly. That it would have worked, produced numbers, and pointed the wrong way, and I'd have had a dashboard proving I was listening.

The second failure is slower. Suppose I ran that campaign across forty neighbourhoods. I've now spent real money placing imagery of people at their most degraded in front of hundreds of thousands of neighbours. Some fraction of those people are hardened permanently. Some fraction act on it — call it in, move someone along, vote no on the next thing. And the people who absorb that cost are sleeping outside in those same neighbourhoods tonight. I'd have manufactured the hostility my own programme then has to work against. The organisation buys its participation numbers with a debt paid by the people it exists to serve.

Third: it doesn't survive a screenshot. One journalist, one frame of the ad next to the nonprofit's name, and every provider relationship and every neighbourhood partnership evaporates. Something whose viability depends on nobody seeing how it was made is already telling you what it is.

So what's left, once you take that away?

The instinct was right. The lever was wrong. Low cost, high reach, honest question — keep that. Provocation by disgust at a population — throw it out.

And when I went looking for what to provoke with instead, it turned out the honest option was sitting there the whole time. Within a mile of where you're standing, tonight, this much edible food goes into a dumpster. Here's what it costs to throw away. What should happen to it?

That is genuinely infuriating. It doesn't need a photograph of anyone. The anger it produces points at a system, and the answers it collects are answers to the question I actually asked. Same reach, same thumb-tap, uncontaminated signal.

The test I use now: would I be comfortable if the neighbourhood saw exactly how I built the instrument? If yes, it's a survey. If no, it's an ad — and an ad measures nothing but itself.

I almost built the ad. I'm glad I wrote it down before I bought the media.