This started as a normal problem. A Nigerian journalist I know (freelance, local focus, nothing flashy) noticed that sources were drying up. It wasn’t all at once or dramatic, just fewer replies, longer delays. More “I’m busy right now” that never resolved.

At first, he assumed it was the stories themselves. Maybe people were scared of the subject matter. Maybe attention had shifted. Maybe he needed to publish louder, faster, more often. But none of that helped. The people who stopped responding weren’t being threatened or arrested. In most cases, nothing had happened at all. They were just uneasy.

The mistake he didn’t know he was making

He thought the risk was publication, but it wasn’t. The risk was contact. Same phone number he used for family, same email he used for invoices, same messaging apps everyone else used. Even when conversations were encrypted, they were still visible. Not the content, the existence of the relationship: who talks to whom, how often, from where.

That’s enough to spook people who have something to lose.

If you’re doing anything sensitive - journalism, activism, whistleblowing, research - this is the part people underestimate. Most people aren’t afraid of what you’ll say, they’re afraid of being seen talking to you.

What he changed (and what you can do, too)

None of this requires advanced tech skills. It just requires changing priorities.

1. He made himself slightly inconvenient

This sounds counterproductive, but it isn’t. Fast replies, casual DMs, “just text me” - those are comforting for you, but not for the other person.

He started using fewer channels, replying more slowly, and keeping conversations narrowly scoped. It sent a signal: “this isn’t casual, and it isn’t sloppy.”

You can do this. Stop giving out your “main” account by default. Separate casual contact from sensitive contact, and let silence be normal instead of apologizing for it. People interpret friction as care when the stakes are real.

2. He stopped thinking about encryption and started thinking about patterns

He realized something uncomfortable. Even a perfectly secure message can still put someone at risk if it creates a visible pattern. Same person, same time of day, same location, same app.

You can avoid always contacting people the same way. Don’t force everyone into the same communication setup. Let some conversations be short, boring, and forgettable. The goal is unremarkability, not invisibility.

3. He designed interactions that were hard to screw up

This matters a lot. Most people will not read a guide, remember instructions, or maintain perfect habits under stress. So instead of telling sources to “be careful”, he changed how he interacted with them. This means giving fewer choices, clear expectations, and no technical jargon.

If someone had to think too hard, the setup was wrong.

You should assume people will make mistakes. So, remove optional complexity and prefer “safe by default” over “safe if followed correctly”. Good security should survive user error.

What didn’t work (worth knowing)

Some things he tried were useless:

• Publishing more aggressively
• Being transparent about his process
• Telling people “it’s safe, trust me”
• Advising caution without changing structure

None of that reduced fear. Fear doesn’t respond to reassurance, it responds to felt safety.

The part people don’t like hearing

Protecting other people is harder than protecting yourself. You don’t control their devices, their habits, or their risk tolerance.

If your work depends on others, your setup has to be designed around their worst day, not your best one. Once he internalized that, something shifted. People started reaching out again, slowly at first, on their own terms.

The takeaway

Censorship doesn’t usually arrive with sirens. It arrives when participation feels risky, even if nothing bad has happened yet.

The fix isn’t being louder or braver. It’s making contact feel boring, contained, and survivable. If you’re losing voices in your community, collaborators, or sources, ask yourself “What does talking to me look like from the outside?”

That answer matters more than your threat model.