In 2013, a survey asked Americans whether they believed lizard people were secretly controlling the government.
About 4% said yes.
It’s easy to dismiss that as a joke. A statistical oddity.
It’s more useful to treat it as a signal.
Because that 4% doesn’t go away. It shows up in every system that relies on human judgement. Not as a fixed number, but as a pattern. A small, persistent layer of distortion.
And when the system changes, that layer doesn’t shrink.
It scales.
The Constant We Prefer Not to See
Every large dataset involving humans carries noise.
People who:
- misunderstand the question
- answer carelessly
- agree reflexively
- or genuinely hold fringe beliefs
Survey researchers have known this for decades. It shows up as measurement error, satisficing, expressive responding. Different names for the same underlying fact.
You never get a perfectly rational sample.
You get a distribution, and the edges are never empty.
In small systems, this sits quietly in the background.
In large, connected systems, it becomes visible.
That shift from invisible to visible is where most misdiagnosis begins.
We assume something new has emerged.
Often, it hasn’t.
When Belief Becomes Identity
There’s a related finding that seems counterintuitive at first.
If you ask people directly, “Are you a narcissist?”, a surprising number of those who are will say yes.
Not reluctantly. Quite openly.
Because for them, it isn’t a flaw. It’s a signal.
It says:
- I’m different
- I don’t follow the rules others do
- I see myself as above the norm
This matters, not because conspiracy thinking equals narcissism. It doesn’t.
It matters because it shows how some traits and beliefs stop behaving like information and start behaving like identity.
Once that shift happens, accuracy becomes secondary.
What matters is what the belief says about you.
Social Media Didn’t Create This
It connected it.
Take any large population:
- a small percentage holds implausible beliefs
- a larger group is uncertain
- the majority dismisses them
Before networks, that minority stays scattered.
After networks, it finds itself.
Forms groups. Develops language. Reinforces internally.
What was once isolated becomes coherent.
Not because it is true, but because it is shared.
And once something is shared, it starts to feel real.
The Incentive Problem
Now add the structure of modern platforms.
Content is ranked by engagement. Not by accuracy, not by calibration.
What spreads tends to be:
- emotionally charged
- identity-reinforcing
- oppositional
Absurd or extreme claims do well because they generate reaction.
Reaction drives visibility. Visibility creates perceived legitimacy.
The loop is simple:
visibility ? perceived legitimacy ? identity reinforcement ? resistance to correction
A small group can start to look like a movement.
Not because it grew dramatically, but because it became visible in the right way.
Why Correction Backfires
If a belief is just information, you can challenge it with better information.
If it’s tied to identity, challenge feels like attack.
So the response shifts:
- contradiction becomes proof of suppression
- disagreement confirms the belief
- exclusion strengthens belonging
This is well documented in motivated reasoning.
People don’t just defend beliefs. They defend the version of themselves those beliefs support.
At that point, the belief becomes self-sealing.
Not logically. Socially.
What You’re Actually Seeing
It looks like:
- more conspiracy theories
- more irrationality
- more fragmentation
But underneath, it’s three interacting forces:
- a constant level of human noise
- identity expressed through belief
- systems that amplify what engages
Change the system, and the expression changes.
The underlying pattern does not.
What To Do With This
You can’t remove the noise. Any system that involves humans carries it.
So the question shifts from elimination to management.
1. Stop treating visibility as evidence
Most people misread scale.
A belief that appears everywhere in your feed can still sit at the edges of the population.
What’s changed is not how many people believe it, but how efficiently it’s surfaced.
If you don’t correct for that, you overreact.
- You design for the loud minority
- You misprice risk
- You misread demand
The practical move is simple:
Separate volume of signal from distribution in reality
They are no longer the same thing.
2. Learn to recognise identity claims disguised as beliefs
Not every statement is trying to describe the world.
Some are trying to position the speaker within it.
These have a different signature:
- high certainty, low proportional evidence
- language of “seeing what others don’t”
- resistance framed as validation
Arguing facts against identity rarely works.
The more useful move is to recognise when you’re not in an information exchange at all.
And decide whether engagement serves any purpose.
3. Design for distortion, not for ideal behaviour
Most systems are still built on an assumption of rational users.
That assumption is wrong.
If 3 to 5% of inputs will be distorted, then:
- surveys need filtering, not blind aggregation
- feedback loops need weighting, not equality
- decision systems need buffers, not immediacy
If you don’t design for distortion, distortion becomes your signal.
4. Be careful what you amplify, even when you oppose it
There’s a structural trap here.
Outrage spreads the thing it’s reacting to.
By arguing against fringe ideas at scale, you can:
- increase their visibility
- signal that they matter
- accelerate their adoption
Not every idea benefits from being engaged publicly.
Some benefit from being contained.
5. Treat certainty as a risk signal
In complex systems, strong certainty is rarely a sign of accuracy.
It’s often a sign of:
- identity investment
- incomplete information
- or social reinforcement
The more certain a claim feels, especially when it flatters the holder, the more carefully it should be examined.
That applies to others.
It also applies to you.
Why this matters
If you misread the system, you end up solving the wrong problem.
You try to correct beliefs instead of managing amplification.
You argue facts where identity is doing the work.
You design for rational actors in systems that reward reaction.
The result is predictable.
Small distortions start to look like cultural shifts.
They aren’t.
But if you build as if they are, they become one.
The Design Problem
If a small percentage of any population will hold implausible beliefs, the question changes.
Not how to eliminate them.
But:
- how to stop them scaling
- how to prevent identity from locking them in
- how to design systems that reward calibration over reaction
Most current systems do the opposite.
They turn small distortions into visible movements.
The mistake is assuming we are watching the rise of something new.
What we are actually watching is exposure.
A constant made visible.
An identity made performative.
A system that confuses engagement with truth.
Once you see that, the content becomes less surprising.
The harder question is what we choose to build around it.
