Daniel's been digging into the history of international law, and he keeps coming back to the same uncomfortable question. We've talked about the prohibition on collective punishment. But regulating war at all is a historically new idea. And one of the classic methods of collective punishment, war waged directly on civilians, was the siege. He points specifically to the Romans in ancient Israel and Jerusalem, and he wants to know how sieges were used throughout the ancient world, and how they persist today in direct contravention of international law.
And it's one of those topics where the ancient practice and the modern legal framework sit in genuine tension. The siege hasn't gone away. It's evolved, it's been rebranded, but the mechanism is the same. Surround a population, cut off food and water, wait.
The image that sticks with me is a walled city, people inside starving, and the army outside just... sitting there. No assault, no breach. Just time.
Time as a weapon. That's exactly what a siege is. The military encirclement of a fortified position to cut off supplies, reinforcements, and escape. Historically it ends in surrender, massacre, or both. And the thing Daniel is getting at—the reason it's collective punishment by definition—is that the siege makes no distinction. The soldier defending the wall and the infant in the house are starved by the same tactic. The besieger doesn't pick targets. The city is the target.
And the defender's choice to hold out doesn't change that calculus. Under modern law, the besieger can't say "well, they chose to keep fighting, so the civilians are on them." But historically, that distinction didn't exist. The city resisted, so the city paid.
Right. And that's the arc we're tracing today. How we went from "that's just how war works" to "that's a war crime." Because sieges are one of the oldest forms of warfare—older than pitched battles in some contexts—and yet international humanitarian law now treats starvation of civilians as prohibited. Geneva Convention Four, Additional Protocol One, the Rome Statute—they all say you can't do this. And yet.
And yet Sarajevo, Aleppo, Gaza. So let's start with the ancient world, because that's where Daniel pointed us. Rome, specifically. No empire did this more systematically.
The Roman siege was not just military necessity. It was deliberate terror. And the centerpiece, the one Daniel asked about, is the Siege of Jerusalem in seventy CE. This is Titus, son of the emperor Vespasian, sent to crush the Jewish revolt. Jerusalem was a fortress city, heavily defended, and the Romans didn't try to storm it. They built a wall around it.
A wall around a walled city.
Circumvallation. It's a wall facing inward, to keep the defenders in, and often a second wall facing outward—contravallation—to keep relief forces out. They literally built a prison around the city. Eight kilometers of fortifications, completed in three days according to Josephus. And then they waited.
Josephus is the source for most of what we know about this, right? The Jewish general who defected to the Roman side?
Yes. And his account is... it's hard to read. He describes the famine inside Jerusalem in detail that feels almost clinical. People eating leather, eating grass. He records mothers eating their children. And you can argue about whether Josephus exaggerated—he had his own agenda—but the archaeological record confirms the destruction layer. The city burned. The Second Temple was destroyed. Josephus puts the death toll at one point one million, mostly from starvation and disease before the final assault.
One point one million. In a single city.
And after the breach, the Romans didn't exactly show restraint. Mass crucifixions around the walls. Josephus says they were crucifying five hundred people a day until they ran out of wood. The survivors were enslaved or sent to the arenas. This wasn't incidental brutality. This was policy. Rome wanted the siege of Jerusalem to be a lesson to anyone else thinking about revolt.
The thing I keep coming back to is the deliberateness of it. A siege isn't a battle where things get out of hand. It's a slow, methodical process. Every day you wake up and choose to continue starving a city.
And the Romans were engineers of that process. They had a doctrine. Circumvallation, starvation as primary weapon, and then massacre or enslavement on surrender. It's the same pattern at Masada a few years later, though on a smaller scale. It's what they did at Carthage in one forty-six BCE—Scipio Aemilianus sealed the city, waited, then razed it and reportedly salted the earth. The siege was Rome's preferred tool for breaking a population's will permanently.
But Rome wasn't the first to do this. Daniel asked about the ancient world more broadly.
No, the siege is as old as cities. And the pattern is remarkably consistent across cultures and millennia. Take the Assyrian siege of Lachish in seven hundred and one BCE. This is Sennacherib's campaign against the kingdom of Judah. We know about it from the Assyrian reliefs at Nineveh—stone panels that show the siege in graphic detail. Battering rams, siege ramps, archers. And then the aftermath: prisoners impaled on stakes outside the city walls, lines of captives being marched into exile. The city was burned. The reliefs are basically propaganda—"this is what happens when you resist Assyria."
Impalement as a billboard.
And the deportation piece matters. The Assyrians didn't just kill—they displaced entire populations. It was collective punishment on a demographic scale. You lose your city, your land, your identity. The siege was the mechanism that made the deportation possible.
Lachish is in modern Israel, by the way. So Daniel's framing about ancient Israel specifically—it wasn't just Rome. The land bridge between empires got this treatment repeatedly.
And then there's Alexander. The Macedonian siege of Tyre in three hundred and thirty-two BCE is one of the more extraordinary engineering feats of the ancient world. Tyre was an island fortress off the coast of what's now Lebanon. Alexander built a causeway—a kilometer-long land bridge—to bring his siege towers to the walls. It took seven months. When the city finally fell, six thousand defenders were killed in the fighting, two thousand were crucified along the beach, and thirty thousand were sold into slavery.
Six thousand dead, two thousand crucified, thirty thousand enslaved. Those are the numbers that make you understand what a siege was. Not a military operation with civilian casualties as a side effect. The civilians were the point.
And then there's Troy. Mythological, yes, but culturally foundational. The Iliad is essentially a siege narrative. Ten years of encirclement, and when the city falls, it's total annihilation. The men are killed, the women enslaved, the city burned. The Greeks didn't have a category for "civilian" either.
The common thread across all of these—Assyria, Greece, Macedon, Rome—is that the city was a single target. No distinction between combatant and civilian. Surrender might spare some, but resistance meant annihilation. The siege was collective punishment by design, and nobody pretended otherwise.
And it was effective. That's the part that makes this conversation uncomfortable. Sieges work. They degrade the enemy's will to fight, they force surrender without a costly assault, and they punish the population that the enemy claims to represent. The Romans conquered the Mediterranean not by winning pitched battles—though they won plenty—but by being willing to sit outside a city for years until everyone inside was dead or broken.
So that's the ancient pattern. Total war on a city, methodically applied, no legal constraints, no moral hand-wringing. And then somewhere in the twentieth century, the law decided that pattern was unacceptable. Did that change anything?
It changed everything on paper. The legal revolution is real. Geneva Convention Four, nineteen forty-nine, Article twenty-three—it requires free passage of medical supplies and food for civilians. Additional Protocol One, nineteen seventy-seven, Article fifty-four—explicitly prohibits starvation of civilians as a method of warfare. The Rome Statute of the International Criminal Court, nineteen ninety-eight, Article eight—makes intentional starvation of civilians a war crime in international armed conflict.
So the law is clear. You cannot starve civilians to win a war.
Clear in text. And the International Committee of the Red Cross updated its guidance in twenty twenty-four specifically on sieges. It reaffirms that the besieging party must allow humanitarian relief and cannot use starvation as a weapon. Even if the defender is embedded in the civilian population, even if military necessity is claimed—the obligation to allow food and medical supplies remains.
And yet.
And yet Sarajevo. Nineteen ninety-two to nineteen ninety-six. The longest siege in modern history—fourteen hundred and twenty-five days. Serb forces surrounded the city, cut off utilities, and positioned snipers on the hills. People were shot fetching water. Eleven thousand five hundred forty-one civilians killed. The UN Security Council passed resolutions—eight twenty-four, eight thirty-six—declaring safe areas. They failed. The safe areas weren't safe.
The UN had peacekeepers there and couldn't stop it.
Wouldn't stop it, in some cases. Srebrenica was a UN safe area. The siege of Sarajevo showed that the legal framework existed, but enforcement depended on political will that wasn't there. And the snipers targeting civilians—that's not just a siege. That's a siege plus active terror. The encirclement creates the cage, and then you shoot the people inside.
Aleppo, twenty twelve to twenty sixteen. Syrian and Russian forces used starvation as a weapon. Barrel bombs on markets, hospitals. The UN tried to run aid convoys. One was attacked on September nineteenth, twenty sixteen. Twenty civilians killed, including the head of the Syrian Arab Red Crescent. The convoy was bombed from the air.
And that's the other modern pattern. The siege isn't just passive encirclement anymore. It's encirclement plus bombardment. The besieger degrades the city from the outside while starving it from the inside. Aleppo was reduced to rubble. The eastern part of the city held out for four years. When it fell, the population was evacuated, but the city was destroyed.
And then Gaza. Daniel lives in Jerusalem. This isn't abstract for him. October ninth, twenty twenty-three, two days after the Hamas attack, Defense Minister Gallant declared a "complete siege." His words: no electricity, no food, no fuel, no water. The ICRC called it a violation of international humanitarian law. The International Court of Justice issued provisional measures on January twenty-sixth, twenty twenty-four, ordering Israel to allow humanitarian access.
And the legal argument from Israel's side is military necessity. Hamas embeds itself in civilian infrastructure. The siege is aimed at Hamas, not civilians. But the effect is identical. When you cut off food and water to two million people, you are starving civilians. The law doesn't care about your intent if the method is starvation.
This is the loophole problem. The prohibition applies to starvation of civilians as a method of warfare. But if you frame the siege as targeting combatants, and the civilian suffering as collateral, you can argue you're not violating the letter of the law. Even when the effect is the same as a Roman circumvallation.
Enforcement is the deeper problem. The ICC has never prosecuted a starvation case. Never. The Rome Statute made it a crime in nineteen ninety-eight. Twenty-eight years, zero prosecutions for starvation. The law exists, but the mechanism to enforce it against a major power or its allies is effectively nonexistent.
We have this strange situation. International law says sieges that starve civilians are war crimes. Every few years, a siege happens, civilians starve, humanitarian organizations condemn it, UN resolutions pass, and nothing happens to the people who ordered it.
The sieges keep happening because they work. That's the uncomfortable truth at the bottom of this. A siege degrades an enemy's will to fight, it forces surrender without a costly ground assault, and it punishes the population that the enemy claims to represent. It's brutally effective. The law says don't do it. But if you do it, and you win, and you're powerful enough, the consequences are... what, exactly?
A strongly worded statement from the ICRC.
Which matters normatively. It establishes the legal record. But if you're the family in the besieged city without food, the normative record doesn't feed your children.
Wait. Hold on. There's another dimension here that we haven't touched. The defender.
Go on.
We've been talking about the besieger as the villain. And under international law, the besieger bears primary responsibility. But the defender also makes choices that kill civilians. If you store ammunition in a hospital, if you refuse evacuation corridors, if you hoard food in military depots and let the poor starve first—those are choices. The law focuses almost entirely on the attacker, but the siege is a two-party transaction.
The human shields doctrine. Additional Protocol One, Article fifty-eight—parties to a conflict shall avoid locating military objectives within or near densly populated areas. It's a legal obligation on the defender. But it's rarely enforced, and it's politically toxic to even raise when civilians are dying.
Because it sounds like you're blaming the victim.
Sometimes it is blaming the victim. But not always. The siege of Leningrad, World War Two—the Soviet leadership made choices about food distribution that killed hundreds of thousands. The Nazi siege was the crime, but Stalin's decisions inside the city made it worse. The two things can be true at once.
Hilbert: Twenty-two percent.
What?
Hilbert: The food that reached Leningrad civilians during the worst months. Twenty-two percent of what the city needed. The rest went to the military and the party apparatus. I read the Soviet archives. Well—I didn't read them. I knew a man who did. Nikolai. He worked at the Hermitage before the war. After, he was in archives. He said the numbers were buried for fifty years.
Hilbert, you've been quiet back there.
Hilbert: I was a historical consultant for a video game studio. Two thousand and four. They wanted to build a siege simulator—realistic, historically accurate. Roman circumvallation, Assyrian battering rams, Byzantine Greek fire. My job was to advise on authentic siege tactics. I spent six months researching siege logistics. Food consumption rates, disease vectors, the order in which people die when the grain runs out.
A siege simulator.
Hilbert: The project was called "Wall." The studio was in Vancouver. They had a publisher lined up and everything. Then the legal team killed it. Said simulating starvation was "not fun" and would be a PR disaster. They were right, probably.
What was the game actually going to let you do?
Hilbert: Play as the besieger or the defender. Two campaigns. And the defender's strategy—I spent three months on this—the defender's strategy was always, always to hoard food and let the poor starve first. That's not a design choice. That's what defenders did. Every siege in history. The granaries get locked, the guards get posted, and the people with money eat while everyone else dies. The game was going to be morally uncomfortable because sieges are morally uncomfortable. And I think that's the part the law doesn't capture.
The defender also makes choices that kill civilians.
Hilbert: The law treats the defender as a passive victim. But the defender is a government, or a military command, making allocation decisions under scarcity. They decide who eats. They decide whether to let civilians leave—and a lot of them didn't, because civilians are a bargaining chip. "Let us go or we all die together." That's a choice. It's not the same choice as the besieger's, but it's not nothing.
The doctrine of human shields gets at some of this, but it's underdeveloped in the law. And politically, nobody wants to touch it. If you're a state being besieged, you don't want the international community examining your food distribution choices. If you're a non-state actor embeded in a civilian population, you definitely don't want that scrutiny.
Hilbert: The game was going to have a mechanic for it. A slider, basically. How much food goes to the military, how much to the civilians. And if you put it too far toward the military, the civilians would riot. But if you put it too far toward the civilians, the military would lose combat effectiveness. The player was going to have to sit there and decide who starves.
The legal team said no.
Hilbert: The legal team said "this is a lawsuit waiting to happen." The publisher wanted a game about castles and trebuchets. We gave them a game about moral triage during famine. They were not the same product.
What happened to the research?
Hilbert: It's in a box. Somewhere. Nikolai's notes are in there too. He photocopied documents from the Hermitage archive before he died. Not sure it was legal, actually. He just... took them. Said they belonged to the dead and the dead weren't using them.
That's a uncomfortable point to sit with. The law focuses on the besieger because the besieger initiates the siege. But once the siege is underway, both sides are making decisions that determine who lives and who dies. And the law has almost nothing to say about the defender's choices.
It's not that the law is silent. Common Article Three of the Geneva Conventions applies to all parties. The prohibition on collective punishment applies to non-state actors too. But the enforcement machinery—the ICC, the UN mechanisms—they're structured around state responsibility. And in an asymmetric conflict, where the defender is a non-state group embeded in a civilian population, the legal framework strains.
Hilbert: The Assyrians didn't have this problem. They just impaled everyone and moved on.
There's a clarity to ancient warfare that modern law has made impossible.
That's not a bad thing. The clarity came from not caring about civilians at all. The legal framework exists because we decided that was unacceptable. The problem is that the framework is aspirational without enforcement. And urban warfare is increasing. The UN projects that sixty-eight percent of the world's population will live in cities by twenty fifty. More cities, more urban combat, more sieges.
We're going to see more of these, not fewer.
Almost certainly. The ICRC's twenty twenty-four guidance tries to close the loopholes—it's more specific about what the besieging party must allow, more explicit that military necessity doesn't override the prohibition on starvation. But guidance isn't law, and even the law isn't enforced. The ICC has never prosecuted a starvation case. Until that changes, the gap between the norm and the reality is going to persist.
What would it actually take to make the prohibition real? A major power prosecuted at the ICC? That's not happening. The US isn't a party. Russia and China have vetos on the Security Council.
The most realistic path is probably regional courts and universal jurisdiction. Some European countries have prosecuted war crimes under universal jurisdiction—Germany's tried Syrian officials, for example. But starvation specifically? No. And the political barriers are enormous.
The open question we're left with is: if sieges are illegal but effective, and if both besieger and defender bear some responsibility for civilian suffering, what does accountability actually look like? The law has an answer on paper. In practice, it doesn't.
The misconception most people have is that sieges are a medieval tactic that doesn't happen anymore. Sarajevo, Aleppo, Gaza—these are modern sieges. The other misconception is that international law clearly prohibits them, so they don't happen. The law has loopholes and weak enforcement. The prohibition exists. The practice continues. Those two facts coexist.
The third misconception—that the besieger is solely responsible for civilian suffering. The defender's choices matter too. The law recognizes this in principle but barely touches it in practice.
This has been My Weird Prompts. Our producer is Hilbert Flumingtop, who apparently has six months of siege logistics research in a box somewhere and has never mentioned it until today.
If you want to hear more about the legal frameworks that try to constrain war, we did an episode on the legal status of disputed territories—Episode thirteen oh two. And as always, send your weird prompts to show at my weird prompts dot com.
We'll be back soon.
