On Saturday afternoon, Halima placed timestamps on the wall of a print shop room that a friend had allowed them to use as a temporary desk. The papers held dates, times, amounts, and transaction types. Juma brought his first statement, the group-deposit export, and Kelvin’s liquidity-stabilization attachment. Their third participant, Ms. Atieno, was a volunteer analyst who helped teachers’ groups examine records. She was not an investigator for the company.
“Today we are not looking for all the money,” Atieno said. “We are looking for what the timestamps allow us to say without crossing the line.”
Juma pinned up the first paper. “This is my deposit of 1,050,000.”
“And the first withdrawal?” Atieno asked.
“230,000.”
Halima placed it beneath the date. “And this is a new member’s deposit, seven minutes earlier.”
Musa was not in the room. He had sent a message asking for an update, but Halima said they would not begin with his anger. This meeting was for records.
Atieno drew a line from one deposit to one withdrawal, then erased it.
“We cannot connect them simply because they are close in time,” she said. “We can write that the withdrawal followed the deposit by seven minutes. We cannot say it was the money that paid the withdrawal without the internal ledger.”
Juma said, “But at first I used that withdrawal as proof that the system was paying from profit.”
“You had a record of a withdrawal,” Atieno replied. “You did not have a record of its source.”
The words were short, but they touched the foundation of what Juma had told people. He had shown them the amount he withdrew and said the platform appeared to work. He had not said where the payout came from, because he had never asked himself.
Halima arranged the deposit papers in time order. Colored marks separated Juma’s family, people in Kelvin’s group, and accounts whose names were unknown.
“The first possibility,” Atieno said, “is external returns that we cannot see. The second is that the early payouts depended on new members’ inflows. The third is that our records are insufficient to distinguish them.”
“Which one is best supported?” Juma asked.
“At this stage,” Atieno said, “the third is the data truth. But when the timestamps are arranged, there is a strong correlation between new inflows and early payouts. That is an inference, not a forensic conclusion.”
Halima wrote across a large page: CORRELATION ≠ PROOF OF SOURCE. Beneath it she wrote: SEEK FORMAL REVIEW.
Juma opened Kelvin’s attachment. The document called liquidity stabilization mentioned matching new inflows while selected withdrawals were delayed. There was no payout-pool ledger or data on external revenue. Juma asked whether “matching” meant every withdrawal had been paid by someone else’s deposit.
Atieno shook her head. “We don’t know. The word could mean an internal system, liquidity management, or something else. We need the company’s terms and account ledger.”
Halima placed each of Juma’s withdrawals on its own line. The first came after a new member’s deposit. The second happened on a day when the group received three deposits. The third arrived after Kelvin’s message that the system was stable. None had an external-return statement beside it.
“I withdrew three times,” Juma said.
“Yes,” Atieno replied. “And that is a fact about you. It is not a guarantee for anyone else.”
Halima turned the screen toward him. “Here is your first message to the group: I’ve tried it; the system pays. People can start small.”
Juma read it and fell silent.
“What did you mean?” Halima asked.
“I meant I had seen a withdrawal.”
“What did people read it as?”
“As if I had verified the safety.”
Atieno told him, “This is where your responsibility meets the source anomaly. We cannot say you created the payout pool. We can say you used an early result as stronger proof than you could defend.”
Juma took the pen. “Write that.”
He wrote: Early withdrawals observed. Source of payouts not established. My prior statement overstated what withdrawals proved. Halima looked at him as if she wanted to correct the words, but did not.
Musa sent another message: Any update? Juma replied that the records showed a pattern of inflows and payouts, but they still had no proof of the source. Musa said they had to pressure the company. Halima replied herself: We are building a bundle that can stand even without an angry word.
Atieno asked them to keep the unknowns in a separate column. They wrote: whether “selected users” were chosen by an algorithm, whether referrals received credits that were not cash, whether K-Plus was a vendor or a different entity, and whether the terms changed after onboarding.
Juma marked the last question. “Kelvin sent the attachment before I joined. I saw it only after entering.”
“Then it goes into the chronology,” Halima said. “It does not make you know everything, but it shows there were red flags that were not followed.”
As afternoon moved toward evening, Atieno cross-checked the amounts. The new member’s deposit before Juma’s withdrawal was very close to the amount withdrawn, but she did not say “same source.” She wrote: temporal and amount proximity; causal link unverified.
Juma felt the words stripping away his protection. He had hidden behind the truth that he “withdrew money.” Now that truth was smaller than the story he had given people.
“We are not saying where all the money went,” Atieno said. “We are saying what the timestamps allow us to say.”
At the bottom of the page, Halima drew a red circle around Juma’s first withdrawal. Beneath it were the new member’s deposit date, the similar amount, and the seven-minute difference.
Juma asked whether this meant he had brought people into a system using new inflows.
Atieno looked at him. “It means you now have a reason to ask that question formally.”
Halima marked the line while Atieno insisted the mark was an invitation to investigate, not a verdict.
Juma’s first withdrawal followed a new member’s deposit by seven minutes, with an amount that was very close.
---