Pili’s print shop closed at eight, but the lights in the back room stayed on. The front shutter was pulled halfway down, the large vinyl printer had gone quiet after the final job, and the worktable was buried under two laptops, an extension cable, cold tea, and a large A3 sheet on which Pili had written in marker:
**KAMERA HAISAHAU — SPREAD MAP**
Nuru walked in carrying her folder, with the lighting invoice still sitting heavily in the back of her mind.
“Do not tell me you collected the whole internet,” she said.
Pili shook her head. “That is exactly what we agreed not to do. We only track nodes we can verify.”
“Good.”
Pili opened a spreadsheet.
The columns were simple:
TIME FOUND.
ACCOUNT.
LINK.
ORIGINAL / REPOST / PAID.
SOURCE VISIBLE?
SPONSOR MARKER?
ASSET CODE?
STATUS.
SCREENSHOT FILE.
Nuru pulled up a chair.
“When did you start?”
“Last night. But today I removed a lot of material we cannot verify.”
“Like what?”
“Screenshots cropped so tightly there is no account name. Forwarded clips with no source. Comments saying ‘I saw it on another page’ without a link.”
Nuru smiled tiredly. “Evidence assistant.”
“Laugh all you want. This is my second job now.”
They opened the first node: the original Mwangaza Street post. Status: deleted.
Second node: sponsor paid card. Status: live.
Third: partner account carrying a vertical edit. Status: live.
Fourth: a reaction creator using the clip in commentary about marriage. Status: live, source unclear.
“How do we classify the reaction account?” Nuru asked.
“Downstream organic. We cannot prove the studio sent it.”
“Then it does not belong in the official distribution chain.”
Pili changed the tag to `ORGANIC / SOURCE UNVERIFIED`.
“And the sponsor?”
“This is the important one.”
Pili opened a screenshot where an asset code appeared beneath the creative preview.
`MSM-BK-SMB-042A`
Nuru read it aloud.
“Internal code?”
“Probably. And I found something similar here.”
Another account had used the same clip to promote a different workshop. The code looked almost identical except for the suffix `042B`.
Nuru frowned. “That suggests more than one export.”
“Or version IDs. We do not claim more than we know.”
“Good.”
They screen-recorded the route to every page. Pili did not use scraping software, stolen credentials, or private logins. She stayed on public pages and links already shared with them. Each screenshot was saved under a filename matching the spreadsheet row and timestamp.
After an hour, Nuru rubbed her eyes.
“This is exactly how doom-scrolling starts.”
“That is why we have a rule,” Pili said. “A new link must answer a question.”
“Which question?”
“Does it show source? Paid marker? Timestamp? Asset code? Does it strengthen the sequence? If not, we do not waste ten minutes on it.”
Nuru copied the rule into the swatch notebook.
Her phone rang. It was Rafiq from the corporate client.
“Nuru, the board has not made a final decision. Send your written statement before ten tomorrow.”
“I will.”
“And the replacement lead?”
“Saidi is ready.”
“Good.”
After the call, Pili asked, “So the booking is still alive?”
“On oxygen.”
“And the equipment installment?”
“Tomorrow night.”
Pili pushed the cold tea toward her. “Then we finish this map tonight.”
Nuru nodded. “But we do not map in anger. If we reach something we do not know, we write that we do not know.”
They returned to the timestamps.
Pili had the original post at 18:42 on the interview day. The first sponsor screenshot Nuru personally saw came two days later. But visible metadata inside the fictional partner ad library showed creative activation at 09:10 on the second day.
“When was the viral spike?” Nuru asked.
Pili opened a screenshot of a public celebratory story the studio had briefly posted before deleting it. The story claimed the clip’s views had “exploded” after noon.
“Spike is after twelve,” Pili said.
“And paid activation?”
“09:10.”
Nuru became still.
The difference mattered. The studio might later describe the sponsor placement as a reaction to virality. But the timestamp showed that paid use had started before the large surge they celebrated publicly.
“Do not jump to the conclusion that they planned everything before the interview,” Nuru said.
“I was not going to.”
“We do not know that.”
“What we do know is that the paid placement predates the viral spike.”
“And that alone is enough to ask how the workflow worked.”
Pili created a new row in capital letters:
PAID NODE — VERIFIED TIMESTAMP.
Nuru looked at the paper map beginning to form. Original post at the center. One line toward sponsor placement. One toward a partner page. A dotted line toward the reaction clip. Another line toward a workshop promo. The internet was no longer one giant cloud of panic. It was a group of nodes with different strengths of evidence.
“This is the difference,” Pili said, “between saying my clip is everywhere and saying these are the places I can prove.”
Nuru nodded.
“When I go back to the studio, I do not ask them to control every repost. I ask them about these IDs and the sponsor deployment.”
“And partner channels.”
“And partner channels.”
Pili printed the current spreadsheet. The printer came alive with its heavy mechanical rhythm. Nuru took a red marker and circled only three nodes: studio original, sponsor placement, partner promo. The rest she left in the gray zone.
“Why only those?” Pili asked.
“Because they have a traceable relationship to the asset. The reaction clips may help show harm, but they are not necessarily proof of studio distribution.”
“You sound like Amina already, and you have not even met her.”
“Who is Amina?”
“A consumer-rights mediator who comes to the print shop sometimes. I asked her about a generic consent problem without giving her your name.”
Nuru looked up sharply.
“What exactly did you say?”
“That creator consent can spread further than a participant understood, and I asked what evidence is usually useful. No names, no screenshots.”
Nuru relaxed slightly.
“Okay. Moral boundary intact.”
“Yes, boss.”
They returned to the map. Pili found a second screenshot where the same sponsor creative appeared under a slightly different crop. The caption had changed, but the tracking pattern remained consistent.
“Same master, different treatment,” Nuru said.
“Which means we should not talk as if there was one single video file.”
“Exactly. There is a source clip, then derivatives.”
Pili added a note under the asset-code column: possible derivative family; verify with studio.
A forwarded message arrived from one of Nuru’s acquaintances claiming another account was “obviously paid.” Pili opened it, studied the page, then shook her head.
“No paid marker. No sponsor label. No visible code.”
“Then unknown.”
“People online would call it obvious.”
“People online do not have to defend the spreadsheet.”
Pili smiled. “Fair.”
Near eleven, they had reduced a chaotic folder of clips and messages into a small evidence set they could explain without exaggerating. Every row contained not only what it showed, but also what it did not show.
Nuru read the entries aloud one last time.
Original post: proves studio publication; now deleted.
Sponsor card: proves paid placement tied to campaign.
Partner promo: likely derivative; studio relationship still to verify.
Reaction clip: proves downstream spread and altered context; does not prove studio sent it.
Pili saved the spreadsheet in two folders and printed a second copy.
“What happens if some posts disappear tomorrow?” she asked.
“We still have timestamps and screenshots. But disappearance itself does not prove bad intent.”
“You really will not give yourself one dramatic conclusion?”
“I have enough drama. I need facts.”
Pili leaned back, tired but pleased.
“Then this is usable.”
Nuru looked again at the sponsor row.
The timestamp was unmistakable.
The first paid placement had begun before the surge in views later described as the reason the campaign picked up the clip.
Pili tapped the screen with one finger and said, “Look at the time—the earliest paid placement was already live before the viral spike.”
---