How Dogmas Begin

I bet you've seen something like this happen in a code review: Someone from the React side of the team posts a sloppy benchmark, and someone from the Vue side answers with a jab about the kind of developer who needs a virtual DOM to feel safe. By the third comment, nobody is discussing rendering performance anymore, they’re discussing what sort of person uses the other tool.

Funny thing is that neither framework or tool has changed. Only the communication about them did.

Dogmas in software begin the same way wars do in real-life: with bad communication between two groups that stop talking about the work and start talking about each other.

Tabs against spaces, monoliths against microservices, AI-assisted vs fuck AI. The pattern underneath is older than any framework, tool, or library, and you can see it cleanly if you strip the technology away.

Wars: A Brief Analogy

Imagine a society with two groups, Group A and Group B. They trade with each other, share a border, and get along well enough most of the time. One day an outlier from Group B harms someone from Group A. The story travels…

.. then Group A starts to believe Group B is dangerous due to that ONE incident, and when an outlier from Group A retaliates a few weeks later, Group B arrives at the same belief from the other direction.

Now each new incident confirms the story for both sides at once, as if the danger was there since forever, they just couldn't see it! Group A can only see its own violence as a response to the danger. Group B can only see its own violence as a response too.

Each side sees the other’s fear as intended aggression without knowing it's also fear.

Here is the trap: if you’re inside one of the groups, you can’t see that you’re alienating people from the other and creating assumptions, because the harm done by your outliers, which you can see and comprehend, reaches to you softened and explained, and the harm done by theirs reaches you raw with no explanation.

You can’t kill an idea and you can’t dissolve a group. Grouping is a biological construct that evolution built into us long before language.

So each side is left trying to eliminate the individuals who refuse to leave the other group, and each attempt confirms the other side’s worst story and recruits more people into it.

That’s how wars begin, and it’s also why they continue long after anyone remembers the first incident.

Now run that pattern through an algorithm built for engagement.

The Modern Hidden War

In 2018, Facebook changed its feed ranking to favor content that produced reactions, and internal documents leaked by former product manager Frances Haugen showed the company weighted the angry reaction at five times the value of a like.

Outrage rose to the top of the feed. Facebook’s own researchers estimated the company caught as little as 3 to 5 percent of hate speech on the platform.

The rumor inside a village in the stone ages took days to travel and lost detail on the way, a Facebook rumor reaches millions in minutes with a video attached. And these days you can't even tell if the video is real.

The consequences have names and dates.

In Myanmar, where Facebook WAS the internet for most of the country, the UN named the platform a useful instrument for spreading hate speech during the 2017 Rohingya genocide, in which over 10,000 people were killed. Meta admitted a year later it hadn’t done enough to keep the platform from being used to incite violence, and Rohingya survivors filed lawsuits seeking $150 billion.

In Ethiopia, hate campaigns on Facebook fed the civil war in Tigray between 2020 and 2022, and when Global Witness submitted test ads with open calls for genocide in Amharic, Facebook approved them within hours. A year after the war, a chemistry professor named Maereg Amare Abraha was murdered in front of his house following a defamation campaign on a Facebook page with 50,000 followers.

In Sri Lanka, a false rumor that a Muslim restaurant owner was sterilizing Buddhist customers spread through the platform and fed the 2018 riots, forcing the government to declare a state of emergency and block Facebook nationwide. The company apologized two years later.

In India, false kidnapping rumors on WhatsApp (which Meta owns) led to more than two dozen lynchings in 2018.

Facebook didn’t set out to cause any of this. The algorithm doesn’t read Burmese or Amharic, and it doesn’t need to, because it measures reactions and anger produces more of them than any other emotion.

European political parties told Facebook in 2019 that the ranking change forced them into divisive content or irrelevance.

Facebook didn’t invent group hatred. It removed the speed limit.

The cycle that takes generations to consume two villages ran through entire countries in a few years, with Group A and Group B each fed a stream of the other side’s worst outliers, each certain it was only responding.

The Moderate

So how do you fix a pattern that evolution wired into the human species

You create a third group.

Call it the moderators: membership is open to anyone, including you and me, and it cuts across every other group you already belong to.

The job has two parts.

The first part is learning this pattern well enough to explain it without notes, then making the explanation travel further than this post will.

Talk about it with friends inside your group, and with people in the groups your group distrusts. The cycle loses power when both sides can name it while it’s happening.

The second part is harder.

Call out alienation when it comes from your own side. Report the Facebook ad from your own political camp that dehumanizes the other one, and flag the post from your own community that trades in fear, even when you agree with its conclusion. When a dogma starts in the workplace about a framework or library, show how there might be 2 ways of thinking, and none of them is right or wrong in itself, talk about their tradeoffs.

The outliers in your group are recruiting enemies on your behalf, and each one you excuse becomes evidence in someone else’s feed. Spotting hatred in the other group takes no effort. Spotting it in your own group, and saying something about it, is what makes a difference.

Now here's a caveat: one person reporting one post won’t undo a ranking system tuned for outrage and a programming community that can't see anything else.

But the cycle itself starts small, with one story told to one receptive audience. An intervention that starts small matches the place where the problem starts.

And I hope, honestly, that we can build again a group all of us are part of.

If you liked this, you might like readplace.com, built for exactly this kind of reading.

Thanks for reading. If you have some feedback, reach out to me on LinkedIn, Reddit or by replying to this post.