Bayesian logic
Did Archimedes really die drawing circles in the sand?
Most people know the famous story:
Archimedes was calmly doing geometry when a Roman soldier told him to stop. Archimedes said “Don’t disturb my circles,” and the soldier killed him. Rome felt bad. Tragic genius, end of story.
But here’s the thing most people don’t know:
That story doesn’t come from eyewitnesses. It comes from Plutarch, writing about 200 years after Archimedes died.
Even worse — Plutarch gives multiple, inconsistent versions of how Archimedes died. In one, he’s drawing. In another, he’s holding instruments. In another, he’s just killed in the chaos. That alone tells us we are not reading a clean historical record — we’re reading moral storytelling.
Plutarch wasn’t just a biographer. He was writing inside the Roman Empire. His job was not to embarrass Rome. His job was to explain Rome in the most morally acceptable way possible.
So think about the incentives: Rome conquers Syracuse.
Archimedes — the most famous scientist on Earth — ends up dead.
That’s a PR disaster.
So what story do you tell? Not “Rome failed to protect him.”
Not “a breakdown in command.”
Not “he tried to escape and was killed.”
You tell the least bad version: “He was so devoted to knowledge that he ignored a soldier and was tragically killed.”
Then Rome can say: “We admired him.”
“We regret it.”
“It was a tragic accident.”
But here’s where Bayesian logic comes in.
In a city being violently captured: • People run
• People hide
• Important figures try to escape
• Soldiers panic
• Things go wrong
Even the fact that a soldier was executed immediately afterward (with no real investigation) points to scapegoating, not truth. That’s how institutions handle embarrassment: they find someone to blame.
So I’m not saying Archimedes escaped or faked his death.
I’m saying the probability that he tried to escape — and that Rome later cleaned up the story — is much higher than the neat legend we were handed.
History isn’t what happened.
It ’s what power decided was safest to remember.
1. What is the official claim?
Archimedes was calmly drawing figures in the sand.
A Roman soldier interrupted him.
Archimedes said “Don’t disturb my circles.”
The soldier killed him.
This is a morality tale, not a forensic account.
It makes Rome look tragic but noble
Make the conquest look civilized
Remove institutional blame
That already lowers its prior probability.
2. What are the priors?
Let’s define three broad hypotheses:
H₁: “He was randomly killed while distracted”
This is the textbook story.
But:
Requires a very specific coincidence
Requires Archimedes not to try to flee
Requires no one else to protect him
Requires a single soldier to ignore explicit orders
Low prior in a chaotic sack of a city.
H₂: “He was killed in the chaos”
This includes:
Crossfire
Mistaken identity
Mob violence
Or being hunted and cornered
This fits:
Warfare
Siege behavior
How famous people die in collapses
This has a high prior.
H₃: “He attempted to escape, possibly succeeded, possibly not”
This includes:
Trying to flee
Being smuggled out
Or vanishing amid chaos
Or dying during the attempt
This fits:
His personality
His incentives
The chaos
The absence of physical evidence
The political need for a scapegoat
This also has a high prior.
3. What does the scapegoat execution tell us?
This part is extremely telling.
A soldier was executed immediately, without investigation.
That is not justice.
That is damage control.
When institutions:
Lose something valuable
Or suffer embarrassment
Or face public anger
They produce:
A villain
That strongly updates against H₁
And strongly updates in favor of H₂ or H₃
4. Why H₃ may even be more likely than H₂
Archimedes:
Knew the city was falling
Knew he was a target
Knew he was too valuable to let be captured
And he had:
Engineers
Students
Loyal Syracusans
Access to tunnels, ships, and hiding places
The idea that he simply sat calmly and waited for a Roman to stab him is psychologically implausible.
The idea that he tried to escape is very plausible.
