Most people name one problem and stop. So we rebuilt the moderator

Reinforced Probing: how the mechanic works

An open question runs into the ceiling of recall. A person retrieves the loudest item and stops — not because they have nothing else, but because the rest feels too small to bother a researcher with. Reinforced Probing is the mechanic Yasna built to get past that point. It uses no statements written by a researcher in advance: it uses what the study's own participants have already said.

1. The field feeds itself. As interviews come in, the platform analyses them continuously and extracts the viewpoints people express, condensing each into one short first-person sentence about a consequence — not “the city is dirty”, but “trash on the streets makes Berlin look run-down”.

2. The pool grows. Those statements accumulate in a live pool that keeps expanding while the study is still running, so later participants meet a richer set than earlier ones.

3. A random draw. When a participant runs out of spontaneous answers, the moderator draws a handful from the pool at random and shows them. Nobody sees a curated list, and no two interviews are identical.

4. Three questions, then normal depth. Which of these matches your experience? Which one does not? Which of the matching ones hits hardest, and why? The chosen theme is then explored exactly as a spontaneous one would be — a concrete occasion, and what it costs the person.

Because the pool is built from the study's own participants, the statements always arrive in the idiom of the actual population rather than a researcher's paraphrase — and in the language the interview is running in.

What it looks like in the conversation

An excerpt from a real interview in this study. The spontaneous part has already been exhausted: the participant named the economy, explained how it affects him, gave an example and said what frustrates him most. Then the pool arrives.

Moderator

Thanks for sharing. Here are a few comments other people have made about everyday problems in Berlin:

— “Construction sites and traffic make it hard to plan appointments.” — “Staff shortages at daycare centers prevent reliable childcare.” — “Trash on the streets makes Berlin look run-down.” — “Extreme heat robs people of sleep and takes a toll on them.” — “The immigration office processes residence permit applications unfairly and too slowly.”

Which of these statements best matches your experience?

Participant Number 1

Moderator Why do you think this statement best describes your everyday life in Berlin?

Participant Because there are construction sites everywhere and everything takes forever. Public transportation is also suffering as a result; everything seems underdeveloped.

Moderator Is there a statement among the others that doesn’t really reflect your experience?

Participant Daycare centers have improved. Trash is a huge problem. Dog poop and noise from motorcycles. It’s getting hotter, and the immigration office is really slow.

Look at the last answer. Asked what does not fit, the participant rejects one statement — and in the same breath volunteers four more problems of his own, including two that were on no list at all: dog mess and motorcycle noise. This is why the rejection question is not a courtesy. It separates the people a theme genuinely touches from those who merely nodded, and it keeps producing material of its own. Rejected statements never enter the data as that person's problem.

The study we tested it on

We needed a live subject rather than a lab exercise. In the summer of 2026 Berlin was in the middle of an election campaign: a quarter of a million posters, seventeen parties, promises phrased so broadly that it was hard to tell what the argument was even about. We decided to come at it from the other end and ask residents what they deal with every day.

The interview covered three subjects: what bothers you in everyday life in this city and what it costs you; what one small change would meaningfully improve your life here; and what you think the candidates are talking about. The election came last, so that a political frame would not decide the list of problems.

The conversation was run by an AI moderator rather than a questionnaire: it reads each answer and follows up on the spot — asks for a recent occasion, asks what exactly stings. Every interview runs in the participant's own language, and a hundred and fifty of them fit into a few days of fieldwork.

Exactly one thing differed between the two fields: in one of them the first subject ended with a handful of opinions from other residents. Everything else — questions, wording, order, who took part — was the same. From here on, grey is the ordinary interview and blue is the interview with Reinforced Probing.

First result: the silent half starts answering

In the ordinary field, 37 people out of 75 name exactly one problem and there the subject closes. With the mechanic running, 12 do.

The gain does not come from talkative people becoming more talkative. It comes from the silent half of the sample starting to answer.

And a different city comes out of the answers

The top places go to things a person physically runs into: dirt on the streets, the cost of living, unreliable transport, construction that never ends. And this is where the first difference between the two fields shows up — the grey track is systematically shorter than the blue one.

The bottom two rows are the control. Crime and migration do not move at all: the people for whom they are paramount name them spontaneously anyway. What grows is what feels awkward or unserious to lead with — dirt, prices, construction, drug use in public places.

And themes appear that the ordinary field does not contain at all

Five subjects exist almost exclusively in the interviews with the technique: conflict between road users on pavements and cycle lanes, sanitation as distinct from litter, graffiti, the price of a ticket, dying shopping streets. Without the technique, conflict between road users comes up once and in passing. With it — six times, and always as a scene.

Bicycle–scooter–pedestrian chaos. You walk between them and everyone is aggressive. Narrow pavements, squeezed between bikes and restaurant tables. *Das Fahrrad-Roller-Fußgänger-Chaos […] Schmaler Bürgersteig, zwischen Fahrrad und Restaurants* - Man, 41

I walk past my own bin area: it stinks, and the rats are helping themselves. *Ich gehe an meinem Müllpark vorbei und es stinkt und die Ratten bedienen sich* - Man, 58

A wider view changes what people ask the city for

The second question was identical in both fields: imagine the city could change one thing for you — not everything, one thing, and a small one.

Requests for more police, more cameras and harsher penalties nearly disappear. And this is not because the second field happened to gather calmer people: crime was named as a personal problem by exactly the same number in both, 14 and 14.

Someone who has named a single problem answers the question about change inside that problem's frame: unsafe means police. Someone who has talked through four themes has to choose between them — and the choice shifts towards what feels achievable.

The level of abstraction drops with it

Right after the request we asked what would be different in the person's ordinary week. The two answers side by side show it better than any table.

A request the size of a city object is testable: the person immediately says what changes in their week. A wish addressed to society cannot be tested — and the account of the week falls apart into “we’d have to wait and see” and “there would be hope”.

Now put that next to what people heard from the candidates

The third subject asked what, in the person's impression, the candidates were talking about. The two lists are close to mirror images.

Dirt — the city's number one problem — is almost absent from what people remember of the campaign. Housing is the reverse: recalled four times more often than it is lived. Eight themes from our list were attributed to the candidates by nobody at all: aggression in public space, crowding, sanitation, noise, tourism, graffiti, shared scooters, dying shopping streets. The first four of those are named as a personal problem by 12 to 15 per cent.

What was visible from the street

We did not set out to study the campaign. But once the data was in, we compared it with the reviews of election posters written by Berlin journalists. Of the slogans that made it into those reviews, seven out of thirty-one carry no subject at all: “Someone has to be sensible”, “Berlin will get better”, “Dangerously honest”. Three are about schools — a theme not a single one of our hundred and fifty participants named as a personal problem. Dirt appeared exactly once, as a joke by a small party that did not enter parliament.

Against that background one party behaved differently. Almost its entire street communication was reduced to a single field: what living in this city costs.

It is the only campaign message our participants retell in their own words. Five days after our fieldwork closed, that party came first in a Berlin election for the first time in its history.

We did not measure any link between these facts: the study contains no voting question, and one election in one city proves nothing. But the asymmetry — that exactly one clearly stated field reached residents out of the whole campaign — we saw in the data before we knew the result.

And a check: the technique does not inflate discontent

Any technique that shows a person other people's opinions carries an obvious risk — are we winding them up? The third subject lets us test that, because we never touched the campaign ourselves.

The distributions are all but identical — and that is with the same people naming half again as many problems and asking the city for different things. The effect stays inside the account of personal experience and does not spill over into judgements about the outside world. For comparison: within the field with the technique, collected in three waves across three weeks of campaigning, the “something other than daily life” share fell from 68 to 36 per cent. The calendar moves opinion about a campaign far more than the technique does.

What this means in practice

Reinforced Probing can be added to a study without breaking comparability with earlier waves: measures of attitude towards outside objects do not shift, while the data about people's own problems gets fuller and more granular. More themes, more niches, more grounded requests — and less risk that a whole layer of everyday life stays off the record simply because it is not the done thing to mention.

Two things worth knowing in advance. Requests do not become more concrete in form — what changes is what people ask for, not how they phrase it. And the link between personal pain and request weakens: the share of people for whom “nothing in my week would change” rises from 7 to 12 out of 75.

Want your own subject taken apart this way — write to us: client@yasna.ai

Fieldwork by Yasna, September 2026, using AI-moderated interviews. 150 Berlin residents aged 18–80 in two independent fields of 75, balanced by gender and age. Quotes are translated from German, with the original in brackets. Slogans as published by The Local, taz and Tagesspiegel.