Halima turned off every camera before the student-radio program began.
“No livestream. Audio only. Callers do not identify themselves.”
Wema sat behind the second microphone with four rules written in front of her: no names, no screenshots, no judging whether someone deserved aid, no private family detail without consent.
The show was not titled *Wema Versus the Scholarship Office*. It was called **Student Aid: What Should a System Have to Explain?**
The first caller said she had been flagged after a cousin bought her graduation shoes.
“Were you given that reason in writing?” Wema asked.
“No. The review just came afterward.”
“Then we say the timing followed the event. We do not call it cause.”
Another caller had traveled to a funeral. Another had attended a birthday dinner paid for by roommates. Another had received a second-hand phone from a brother.
Halima built a whiteboard taxonomy: one-off gift, emergency travel, shared purchase, required equipment, social image.
After each call, Wema asked the same questions: Did you know the criterion? Could you appeal? Were you asked for context?
Mostly, the answers were no.
But she refused to claim that the callers represented the whole campus. They were a self-selected group with problems strong enough to call a radio station.
During the break, Halima said, “You sound like you want to apologize for every sentence.”
“I do not want to turn harmed people into content.”
“Fine. But do not kill the story.”
“The story is that the process does not explain itself.”
“Then say that.”
Back on air, Wema did.
“What we can say so far is that one-time events are being interpreted as signals of long-term financial capacity, and students report little opportunity to give context before consequences. We do not know the entire model. We do not know the office’s intent.”
A caller demanded that they name Karanja.
“We have evidence about governance questions,” Wema said. “We do not have evidence that he personally submitted any one signal.”
The final caller spoke quietly.
“I was not reported by a person.”
“How do you know?” Halima asked.
“My data-access record says source type ‘public media ingestion.’ There is no peer identifier.”
Wema sat straighter.
“Do not send us the full record,” she said. “Ask for a formal explanation. If you later choose to share a redacted result, that is your choice.”
The caller laughed bitterly. “That is the problem. The system picked up my picture by itself.”
The studio went still.
After the broadcast, they reviewed the raw audio before publishing replay clips. They removed locations, courses, and details that could re-identify callers. Some voices were not used at all because callers had withdrawn permission.
The producer complained that the edit would lose traffic.
“Privacy matters more when the story is about privacy,” Halima said.
Wema built a tally but refused to produce percentages from the small, self-selected sample. She wrote five policy questions instead: Is public media a source? How are peer signals verified? Are merchant categories contextualized? Can a student see the evidence? Does a functioning appeal panel exist?
The post received fewer clicks than a scandal montage would have. Yet students from different faculties began contacting them with decision letters and volunteering for interviews.
Wema replied with one instruction: *Do not send full records. Tell us the aid type and whether you have a written decision. We will discuss consent first.*
Before leaving, she wrote the last caller’s words in the corner of the whiteboard.
Before leaving the studio, Halima locked the raw audio in a restricted archive and created a separate editing copy. Callers who had withdrawn permission were flagged so their voices could not accidentally appear in a replay. Wema wrote a note above the tally: *Radio callers are not a campus sample.* The sentence looked almost comically cautious beside the dramatic stories they had heard, but it shaped the next decision. Instead of running another call-in show, they would move through faculties and service points, looking for people who had written decisions and different kinds of context. The radio had given them questions. Field reporting would have to test whether those questions survived outside the studio.
**The system picked up my picture by itself.**