In brief: The 2026 Grassroots Internships closed on Monday 26 October 2026 at Sir Syed University of Engineering and Technology (SSUET) in Karachi, where the two Karachi teams received their certificates and stipends. The Bangata team’s certificates are going to Tanzania. Twelve young people in three teams ran four-week experiments from the Grassroots Playbook in Karachi and near Arusha. The most useful thing they produced was a result that got smaller.
At the public showcase on 23 October, a member of the Beyond Differences team described the moment his team’s headline stopped flattering them.
They had added a range to their first result and a check to their second. Both made the number less impressive. “Leaving them out,” he said, “would have given a stranger a wrong picture of our work.”
That sentence is the programme. The result didn’t shrink. The method grew up.
How did the programme end?
On 26 October the two Karachi teams, Beyond Differences and Scene Makers, came to SSUET for the close. It was a fitting room for it: Beyond Differences had started its fieldwork at the university’s own Job Fair on 31 July.
Every participant receives a Certificate of Completion, signed by the two co-founders, Kubair Shirazee and Sahar Zaidi-Shirazee. It names the holder, the team and the programme. Each team is paid the stipend it was promised on completion, with a share of a discretionary bonus on top. The Bangata Knowledge Connectors could not be in Karachi, so their certificates are being sent to them in Tanzania.
Scores are not made public. Every team was recognised for its own special and extra endeavours.
The internships took no grant, no institutional money and no corporate sponsor. They were funded personally by Kubair Shirazee and his family.
How was the work judged?
Each team took one experiment from Kubair Shirazee’s Grassroots Playbook and ran it for four weeks, between 28 July and 24 August 2026, alongside school, exams or work. Under-18s took part with guardian permission and adult supervision.
Five reviewers scored the work on 18 September. Four are from outside the programme; the fifth is ChangeCraft’s engagement director. The founders took no part in the scoring.
The guide gives a hundred marks. Thirty go to how honestly a team handles its own evidence, and twenty to whether what it published matches what it was permitted to publish. An experiment could fail and still score well. One that looked good and could not show its working could not.
Why did the number get smaller?
Beyond Differences, five university students in Karachi, ran facilitated dialogues on the assumptions people inherit about other communities. They measured comfort in discussing ethnicity before and after each session, on a five-point scale. The two measured sessions were different groups of people.
| What it is | Figure |
|---|---|
| Week 3, the first measured session. Comfort rose from 3.29 to 4.43 across seven paired records | +1.14 |
| Week 4, with a tighter method and an outside facilitator. Comfort rose from 3.71 to 4.14 across seven completers | +0.43 |
| The same week 4 session with its single largest mover taken out | −0.17 |
Two caveats sit beside those figures, and they are separate. The week 3 result carries a range, from +0.86 to +1.14, because participant codes drifted between the before and after forms and five records had to be re-matched afterwards. The week 4 result rests largely on one person: without that participant’s response, the session shows a small fall.
The programme’s weekly review spotted what was driving the week 4 number and asked the team to show it. They did, in full. They kept that participant in their headline, reported the check beside it, and kept in the data the scores that went down. In their own words at the showcase, these are “small, self-reported scores without a comparison group.”
The week 4 session also tested something the numbers cannot. The team handed its written dialogue kit to a facilitator who had first come to them as a participant, and stood outside the room while he ran it. A method that only works while its authors are present is not yet a method.
What did the other two teams find?
Scene Makers were a team of two in Karachi. Their experiment was to swap a screen for a scene: offline games, mindfulness and art, to see what happened to focus and connection. Their first plan was a private padel court. Challenged on who that would leave out, they moved to a free public park and a home for older people.
They spent 16,175 rupees in cash over four sessions. Their participants still spent between 300 and 1,850 rupees each to travel to them. A free venue is not the same as a free session.
Attendance went from nine at the first session to 13 at the second and four at the last. Of the eight people who completed both forms, four reported no change in screen time, two an increase and two a decrease. For the three people who came back for the final session, loneliness scores rose, by one, five and two points. Their starting point for screen time was what participants recalled from July, and the team labelled it as recollection. They did not ask people why they stopped coming, so they did not guess.
The Bangata Knowledge Connectors, five students at a girls’ secondary school in Bangata Village near Arusha, worked with Savannah Vision Tanzania. Before knocking on a single door they went to the Village Executive Officer for permission. He was out on 3 August, so they came back the next morning and left with it.
They mapped what 12 households knew and what they lacked, then introduced neighbours who had never compared notes. One shopkeeper told them that since the introduction, her debtors had started paying in advance. That is one person’s account, and the team presented it as one. They spent nothing in cash. They took no baseline, because exams came first, and they said so plainly: “We skipped the formal baseline, and it cost us.”
What did the guests ask?
The second half of the showcase belonged to the guests. Scene Makers and Beyond Differences answered for themselves. The Bangata students were sitting exams, so Savannah Vision Tanzania answered on their behalf.
Not one question asked whether an experiment worked. Guests asked why one participant came back for the final session without being asked. How attendance could have been kept up. What could continue after the four weeks, and who would carry it. Whether four weeks was long enough at all. And what each team would never do again.
Attendance, continuity, ownership. Every one of those is about what happens after the facilitator leaves. We have started calling it the week five question: not “did it work?” but “who is still doing it when nobody is watching?”
The answers are the teams’ to give, and they gave them on film. Beyond Differences plan to run the revised version of their kit themselves, then hand it to other facilitators. Scene Makers plan to carry on, starting with the mindfulness session their participants valued most. Savannah Vision Tanzania saw nothing in the way of the Bangata students continuing, beyond fitting it around the school timetable.
What would we change for a future cohort?
The teams’ own lessons come first, because they are better than ours. Measure from day one rather than from memory. Ask people why they stop coming. Put the request for criticism inside the session, before people leave. Recruit a team, and participants, who live near each other.
Ours follow from theirs.
The scoring guide reached the teams in their fourth week, not before they began. A team cannot aim at a standard it has not seen. Next time it goes out on day one.
Only one team measured before and after each session. One rebuilt a starting point from recollection, and one took none. A baseline cannot be produced afterwards, and a programme that lets two teams in three reach week four without a measured one has a design gap of its own. A measure taken in week one is the first change on the table.
And four weeks is long enough to learn a method, not to prove a community changed. The programme claims capability, not community impact. That is the claim the evidence supports, so it is the only one we make.
So here is our question, and we would like an answer rather than agreement: if your last result had been worked out again without the person who moved it most, would you still have printed it?
The teams, their evidence and the scoring are on the 2026 Grassroots Internships page. To mentor a team or host an experiment in a future cohort, get in touch.
Field note, October 2026. Part of the record of Peace Through Prosperity’s work since 2011.