Street / Social Experiment

What this format actually is

You approach members of the public and their reactions are the content. One question asked a hundred times, a small social test, an offer, a request. The substance is the interaction with strangers.

The boundary with a Challenge is where the constraint sits. A challenge is a rule imposed on you. This is about what other people do, and you are the instrument rather than the subject.

Why it reaches so far

The appeal is that the answers are genuinely unknown, to you and to the viewer, and human beings are extremely interested in what other human beings will actually say. It also needs no context: a stranger being asked a question is instantly legible to someone who has never heard of you, which is the precondition for large short-form reach.

Done seriously, it is the closest thing on this list to primary research. One question asked of a hundred people, published as a distribution, is genuinely more informative than most commentary about what people think — and almost nobody does it that way, which is the opening.

The consent problem, which is the whole format

You are filming people who did not seek this out, and publishing them to an audience.

Get consent on camera and keep it. Not implied consent from them continuing to talk — an explicit, recorded "is it OK if this goes on YouTube?" after the interaction. Keep those clips even when you cut them, because both platforms and subjects ask later.

Know the law where you film. Filming in public is broadly permitted in the UK and US and is restricted or requires consent in much of Europe and Asia. Private property — shopping centres, stations, most of what looks public — is not public. This varies enough that "I filmed in public" is not a defence you can rely on without checking locally.

Do not humiliate people. The version of this format that gets reach quickly is the one where a stranger is made to look stupid. It is also the version that ends channels, poisons the well for everyone filming in that city, and is straightforwardly unkind to someone who was walking to work. If the video only works because the subject does not realise how they come across, do not publish it.

Delete on request, without argument. Someone changing their mind afterwards is entitled to. Treating that as negotiable is how creators end up in the news.

What it costs

Fourteen hours a week, and most of it is rejection. The overwhelming majority of approaches fail — people decline, walk past, or give nothing usable. Getting twenty usable responses can take a full day, and creators consistently underestimate this by an order of magnitude.

It also requires a specific temperament. Approaching strangers repeatedly and being refused is draining in a way that does not improve as much as you would hope, and no amount of planning substitutes for being someone who can do it warmly on the fiftieth attempt.

What separates the good from the exploitative

Ask something worth asking. "What's your salary" gets clicks. "What did you believe at twenty that you don't now" gets answers people remember. The second one is not harder to ask.

Lead with the best answer. Nobody watches a setup for a stranger they have not met. The strongest response goes first.

Publish the distribution, not just the highlight. If ninety people said the boring thing and ten said something surprising, showing only the ten is a lie about what people think. Saying "eighty-eight of a hundred said X" is more interesting anyway.

Be somewhere with something to say. The same question in two different neighbourhoods is a better video than the same question in one, and it is the closest this format gets to genuine insight.

Who it suits

Confident, warm people in a city with public space and a licence to film there. All three are required and the first is not learnable in any useful timeframe.

It does not suit anyone uncomfortable with repeated rejection, anyone unwilling to handle consent rigorously, or anyone in a place where filming strangers is restricted or culturally unwelcome.

Where to start

One question, a hundred people, published with the actual counts. That is the most defensible version of this format, the most useful, and the one nobody is making because it takes three days instead of three hours.

Ask consent on camera every time. Lead with the best answer. And pick a question you would be comfortable being asked yourself by a stranger with a camera — if you would not answer it, do not ask it.