The Same Machine Can Produce Myths and Summaries

What if a civilization and a model are solving the same problem?

What do a sacred myth and a state of the art text summarizer have in common? At first glance, almost nothing. One tries to explain why people become a mob, choose a victim, and then declare peace. The other tries to compress a document into a shorter version that preserves the important parts. Yet both are, in a deeper sense, machines for deciding what matters, what gets left out, and which version of reality will be treated as the truth.

That is the unsettling bridge between them. Humans do not simply observe the world and report it. We select, compress, and narrate. We turn messy events into stories with a center of gravity. Sometimes that selection reveals reality. Sometimes it distorts it. And sometimes, as with scapegoating, it creates a false coherence so persuasive that an entire group mistakes relief for truth.

The real question, then, is not just whether myths are false or summaries are accurate. It is this: what kind of compression preserves truth, and what kind of compression manufactures a lie that feels like truth?

---

Compression is not neutral

Summarization seems like a technical task. A good summary should keep the core facts, drop redundancy, and maintain fidelity. Metrics like ROUGE exist because we need a way to measure overlap between the summary and the source. Modern systems such as BART, PEGASUS, and GSum are judged by how well they approximate the meaning of the original text while reducing its length.

But every act of compression is also an act of power. To summarize is to declare which details are essential, which are background noise, and which can be erased. That sounds harmless when the task is an article or a news report. It becomes dangerous when applied to people, conflicts, or communities. Then the same mental operation can transform a complicated social reality into a story with one villain, one cause, and one clean emotional resolution.

This is where the connection becomes profound. A summary that strips away too much nuance does not merely become incomplete. It can become morally dangerous. In a mob, the crowd performs the worst possible summary: all the tension, envy, rivalry, and fear in the group are compressed into a single figure. The victim becomes the explanation for everything.

A good summary reduces information. A bad summary reduces responsibility.

That distinction matters because humans hunger for compression. We do not like unresolved complexity. We want the meeting to end, the argument to settle, the blame to land somewhere. We want the story to fit in our heads. The problem is that social reality is rarely neat enough to fit without distortion.

---

The scapegoat is a failed summary of a broken system

Imagine a room full of children and a pile of toys. One child picks up a toy, and suddenly another child wants that exact one, even though dozens of others are available. The object is not desired for its intrinsic qualities. It is desired because someone else desired it first. That is mimetic desire, the strange fact that people often want what other people want.

Now scale that pattern up to a workplace, a city, or a nation. People copy each other’s ambitions, resentments, and fears. Rivalries spread. Alliances shift. Everyone becomes a little more like everyone else, which means everyone is competing over the same social signals, status markers, and scarce symbols of prestige. The atmosphere thickens. Suspicion rises. The group is no longer made of individuals with separate aims. It starts to behave like a single nervous organism.

At that moment, a scapegoat offers the dream of compression. Instead of examining the tangled web of imitation and mutual escalation, the group can say, in effect: it was that person. The outsider, the dissenter, the odd one, the inconvenient one, the one already marked as different. The crowd’s chaos gets summarized into a single name.

This is why scapegoating feels so compelling. It is not only violent. It is cognitively efficient. It converts ambiguity into certainty. It converts a distributed process into a local cause. It offers the emotional satisfaction of a clean explanation, the way a bad summary can feel more useful than a faithful but messy one.

And then comes the most revealing part. Once the victim is gone, the group often feels calmer. That calm can seem like proof that the summary was correct. If the tension lifted after the person was expelled, surely the person was the problem. But this is exactly the trap. The relief was produced by coordination, not by truth.

The scapegoat is not the cause of social disorder. The scapegoat is the summary the disorder invents to survive itself.

This is why scapegoating is so durable. It works, briefly. It stabilizes a group without solving the underlying dynamics. It is a compression algorithm with terrible long term accuracy and immediate emotional payoff.

---

Myths, metrics, and the temptation to mistake fluency for fidelity

The most revealing parallel between myth and machine learning is not that both tell stories. It is that both can produce outputs that sound right because they are optimized to sound right.

A myth often does not preserve the experience of violence from the victim’s point of view. It preserves the emotional structure of the crowd. It tells a coherent tale in which the victim appears guilty, the group appears justified, and peace appears to confirm the verdict. The myth is not random nonsense. It is structured, persuasive, and socially functional. That is why it lasts.

Modern summarizers face a similar danger. A system may produce a fluent summary that has strong lexical overlap with the source text and therefore scores well on metrics like ROUGE. But a fluent, high scoring summary can still miss what mattered most. It might preserve the surface features of the document while flattening its causal spine. A meeting transcript about a company crisis can be summarized as “the team discussed several issues,” which is technically true, uselessly vague, and strategically misleading.

This is the hidden parallel: both myth and weak summarization can be excellent at preserving the wrong thing. Myth preserves the social legitimacy of the crowd. A sloppy summary preserves the visible fragments of language without preserving the real structure of meaning.

Think of a news article about a scandal. A bad summary might include names, dates, and a generic outcome, while omitting the key mechanism, such as incentives, conflicts of interest, or procedural failure. The reader comes away informed enough to repeat the headlines but not informed enough to understand the system. That is a kind of epistemic scapegoating. The complexity is sacrificed so that a simplified narrative can stand in its place.

Here is the deeper lesson: compression always reveals a theory of importance. If your summary says one thing is central, you are revealing a model of the whole. If that model is too narrow, you will unintentionally imitate the mob, which also thinks it has identified the one central cause.

---

A better model: truthful compression versus coercive compression

Not all compression is bad. In fact, civilization depends on it. Without summarization, we drown in details. Without narrative, we cannot share lessons. Without abstraction, we cannot think at scale. The problem is not compression itself. The problem is the absence of a standard for distinguishing truthful compression from coercive compression.

A useful framework is to ask four questions about any summary, story, or explanation:

1. What was kept? Which facts, relationships, or tensions survived the reduction?

2. What was removed? Were the omitted details merely repetitive, or were they the parts that changed the meaning?

3. Who benefits from this compression? Does the simplified story clarify reality, or does it protect an institution, a faction, or a crowd from accountability?

4. What does the compression encourage people to do next? Does it lead to wiser action, or does it invite blame, conformity, and premature closure?

This framework applies to AI systems, journalism, office culture, and politics. A good summary helps you see the structure of the problem more clearly than the original did. A bad summary gives you the emotional feeling of clarity while concealing the structure itself.

Consider an example from work. Suppose a team is falling behind on a product launch. The coercive summary says, “The problem is Alex. Alex keeps slowing things down.” That story is easy to tell, easy to repeat, and easy to rally around. But a truthful compression might be: “The team is overcommitted, requirements changed late, coordination is poor, and Alex is absorbing the resulting friction because the role is ambiguous.” The second version is harder to say and harder to use as a slogan. But it points to the real system.

The same distinction shows up in public life. When a social crisis is summarized as the fault of a single group, the story becomes emotionally satisfying and strategically useless. When it is summarized as the interaction of incentives, fear, signal amplification, and institutional failure, the story becomes more demanding but more truthful.

The best summaries do not make reality smaller in the emotional sense. They make it smaller in the informational sense without shrinking its causal depth.

---

The hidden ethical lesson: defend the victim, defend reality

If scapegoating is a false summary of social disorder, then one of the deepest ethical acts is to refuse premature explanation. That means resisting the urge to let the most vulnerable person become the container for everyone else’s anxiety.

This is not just a moral injunction. It is an epistemic one. When groups coordinate around a victim, they stop asking uncomfortable questions about themselves. They no longer need to examine envy, imitation, status competition, fear, or failed institutions. The victim absorbs the explanation, and explanation is what humans most crave.

That is why the role of a genuine defender is so important. A good defense attorney, in the broad sense, does not merely protect an accused person. They interrupt the social machine that wants a quick answer. They force the group to confront complexity before it turns into consensus.

The same role exists in good editing, good science, and good engineering. A strong editor does not compress reality until it is neat. A strong scientist does not force data into a flattering pattern. A strong engineer does not optimize for a metric while losing the real user problem. In each case, the defender’s job is to protect the system from the tyranny of the easiest explanation.

This is why the comparison between sacred narrative and summarization is so useful. It reminds us that truth requires structure, but structure can be abused. Every culture needs forms that reduce complexity. But when those forms become instruments of blame rather than understanding, they turn into ritualized lies.

If you want a practical test, ask yourself whether a conclusion leaves room for shared responsibility. The moment a narrative makes one person, one group, or one cause bear all the weight, alarm bells should ring. Reality is usually distributed. Harm is usually systemic. And the urge to simplify is often the first sign that a crowd is searching for a victim.

---

Key Takeaways

• Every summary is a theory of importance. If you cannot justify what got left out, you may be mistaking compression for understanding.
• Scapegoating is a coercive summary. It reduces a complex social process to a single blamed person, which feels clarifying but usually hides the real cause.
• Fluency is not fidelity. A story can sound coherent, or a model can score well, while still preserving the wrong structure.
• Ask who benefits from simplification. The more a summary protects a group from accountability, the less trustworthy it is.
• Defend complexity before it hardens into blame. Slowing down the rush to a single culprit is one of the most important intellectual and moral habits you can build.

---

Conclusion: the real battle is over what gets to count as the whole story

We like to think the opposite of myth is data, and the opposite of mob violence is justice. But the deeper opposition is between truthful compression and self serving compression. One reduces complexity while preserving causal reality. The other reduces complexity by deleting anything that threatens the crowd’s preferred story.

That is why the connection between scapegoating and summarization matters so much. Both are acts of selection under pressure. Both decide what survives. And both can either illuminate a system or conceal it.

So the next time you see a neat explanation, whether in a rumor, a headline, a meeting recap, or an algorithmic summary, pause before admiring its simplicity. Ask what kind of peace it produces. If the answer is relief through blame, it may be the oldest lie in human life, dressed up in modern clothes.

The hard work of intelligence is not merely making things shorter. It is making them shorter without turning a world into a victim.