Israel's response is disproportionate

The fuller answer

12 sectionsJump to one

The word is doing two different jobs

This is the most misused word in the whole argument. Learn what it means and you have an advantage in every exchange you will have on this subject, not because you will know more law but because the person opposite you is using a legal term to make a point that has nothing to do with the law. They have not noticed.

In ordinary speech, disproportionate means the losses were lopsided. Far more of them died than of us, so the response went too far. That is what the word means at a dinner table, and as a moral reaction to a death toll it is intelligible.

It is not the legal test, and never has been, in any treaty, in any period.

Follow the ordinary meaning through and it collapses. If lopsided casualties made a war unlawful, the side that lost more people would be the wronged party in every conflict in history. Automatically. Whoever started it, however either side fought. That makes the better defended side guilty by construction: a country that builds shelters, evacuates its towns and intercepts incoming fire becomes more culpable the better its civil defence works. Nobody believes that when it is put to them directly, and it is why the drafters wrote a different test.

What Additional Protocol I actually says

Article 51(5)(b) of the 1977 Additional Protocol I lists, among the attacks to be considered indiscriminate:

“An attack which may be expected to cause incidental loss of civilian life, injury to civilians, damage to civilian objects, or a combination thereof, which would be excessive in relation to the concrete and direct military advantage anticipated.”

Article 57 turns the same test into a duty on the people planning the strike. They shall “refrain from deciding to launch” any attack expected to cause harm on that description, and under Article 57(2)(b) an attack “shall be cancelled or suspended if it becomes apparent” that it would.

The Rome Statute words it differently. Article 8(2)(b)(iv) makes it a war crime to launch an attack knowing it will cause harm “clearly excessive in relation to the concrete and direct overall military advantage anticipated”. Two additions: clearly, and overall. The threshold for a conviction sits above the conduct rule, which is ordinary in criminal law and is not a loophole.

Now read either sentence again and look for the enemy’s dead. They are not in it. They are not one side of the scale.

Three features of the test, each of which surprises people

It is forward looking. It asks what the commander expected before the strike, not what was found in the rubble afterwards. Chatham House states it flatly: proportionality requires an ex ante assessment. So a strike that killed many civilians can have been lawful, and a strike that killed none can have been unlawful. It is a test on the decision, not on the result.

It is per attack, not per war. No treaty rule totals up a campaign and asks whether the sum was proportionate. Attacks are sometimes grouped where the advantage from one depends on the others, which lawyers call the attack as a whole. That is a tactical grouping, not a war.

Say the exception yourself, because a briefed opponent will produce it. In Kupreskic, on 14 January 2000, an ICTY trial chamber wrote that where attacks are repeated and all or most of them fall in the grey area between indisputable legality and unlawfulness, it might be warranted to conclude that their cumulative effect is not in keeping with international law. That is a court reasoning about a pattern, and it is the best thing anybody has against the per-attack rule. What it is not is a licence to argue from a total. It builds the pattern out of individual attacks and then asks about the sum, so it needs the attacks named just as much as Article 51(5)(b) does.

It weighs civilian harm against military advantage. Not against enemy casualties, not against the harm the other side inflicted, and not against what was done on the first day. The other side of the scale is not their dead. It is what the target was worth militarily.

The reasonable commander, and the file you have not read

Who decides what counts as excessive? The standard the tribunals converged on is the reasonable military commander. A committee reviewing the NATO bombing of Yugoslavia reported in June 2000 that the determination of relative values must be that of such a commander, and added something more useful still: a human rights lawyer and an experienced combat commander are unlikely to assign the same relative values. The ICTY refined it in Galić into a question about “a reasonably well-informed person in the circumstances of the actual perpetrator making reasonable use of the information available to him or her”.

Notice what applying that would take. What the target was, what the intelligence said was inside it, which munition was chosen and why, what warning was given. Almost none of that is public in any conflict, and where it surfaces it is usually years later, through litigation or a leak.

Which is why the confident verdict from a studio, in either direction, is not a legal opinion. It is a reaction to the one variable anybody outside can see, and that variable is the input the test does not take.

Two questions, and the ratio answers the second one

Everything above says a ratio is not the legal test. Keep saying it, but do not let it slide into a claim that the figures tell you nothing. They tell you nothing about the law. They tell you a great deal about how a war was fought, and those are separate questions.

On the modelling published by Action on Armed Violence in January 2026, a body arguing that Israeli claims about civilian harm do not hold up, the ratio of civilians to combatants killed in Gaza runs between five to one and eight to one. That earns its keep set beside other campaigns fought against an embedded enemy in dense urban terrain, which are the only fair comparators. The figures, and the Mosul and Raqqa campaigns, are in the entry on the death toll.

Say it like this: “The ratio doesn’t answer the legal question. If we’re judging a war by it anyway, compare it with Mosul and Raqqa and use one standard for both.”

The damage figures, and the one thing they are evidence of

This is the most dangerous material on the page. Used as most advocates use it, it recites the charge against Israel and calls it a defence. A briefed opponent will say exactly that, in one move. Read the concession first, and never say the figures without it.

Concede this before you touch a number. Article 147 of the Fourth Geneva Convention makes it a grave breach to destroy property extensively where military necessity does not justify it. That is a war crime about property, and it stands whether or not anybody was inside. Emptiness does not make a building a lawful target.

Then concede the second part, which is newer. Demolition has continued since the fighting stopped. Al Jazeera dated further demolitions in Shujayea and Tuffah to between 5 November and 13 December 2025, on satellite imagery. The New York Times counted more than 2,500 buildings razed after the October 2025 ceasefire.

So “the buildings were empty” answers nothing about whether flattening them was lawful. Offer it as though it did and you will be finished in one move.

There is still one thing the figures show, and it is narrow. They show where the population was.

The UN Satellite Centre assesses damage in Gaza from satellite imagery. On the imagery of 11 October 2025, published on 31 October, it found 198,273 structures affected, roughly 81 per cent of every structure in the Gaza Strip. Of that total, 123,464 were destroyed outright, with an estimated 320,622 housing units damaged. Get the split right when you say it. The 81 per cent is the affected total. The 123,464 is the destroyed subset inside it, about half of everything standing. Say it the other way round and the first person to open the report has caught you inventing a figure.

Now set the destroyed count against the recorded dead. The Gaza health ministry’s named list carries 72,835 people to 7 May 2026. The average Gaza household ran to 6.5 people at the last census, and much of the housing stock was blocks of flats holding several households each. Occupied at anything like that density, 123,464 destroyed structures would have produced a toll in the hundreds of thousands. It did not. They were not occupied when they came down.

That is a fact about intent and about nothing else. A force whose object is to destroy a population does not move it out of the buildings first. Every figure comes from the United Nations and the Gaza health ministry, so the point asks nobody to trust Israel about anything. It belongs in the genocide entry, where the question is what the war was for. It does not belong here as a defence of the destruction, because it is not one.

State the remaining limits yourself. Damage assessments count structures, not occupancy. Displacement is not safety: those people were somewhere else, often somewhere worse, under near-daily evacuation orders into shrinking ground, and the UN has documented exactly that. And none of it speaks to any individual strike.

Say it like this: “I’m not telling you flattening empty homes is lawful. It can be a grave breach on its own. I’m telling you the UN’s own damage figures don’t fit a population that was still inside them, and that is an argument about intent, not about lawfulness.”

A standard nobody has ever applied to anybody

One more test, and it is a test of their standard rather than a defence of Israel. Run the ratio now being treated as proof of criminal intent across the Allied campaigns of 1939 to 1945. No campaign in that war passes, including the ones fought against an unambiguously genocidal enemy.

Between 1940 and 1945 Allied air forces dropped nearly 600,000 tons of bombs on France, a country they were liberating, killing perhaps 60,000 French civilians. That is comparable to everything German bombing and the V-weapons killed in Britain across the whole war. The Dresden Historians’ Commission, appointed by the German city itself, reported in March 2010 that up to 25,000 people died there over two nights in February 1945. Operation Meetinghouse burned sixteen square miles of Tokyo on the night of 9 March 1945 and killed an estimated 110,000 people in a few hours.

Two limits, and say them yourself before anybody else does. Nobody defends those campaigns now, several were arguably unlawful by the standards of their own time, and an advocate who looks like they are defending Dresden has lost the room and deserves to. And the law changed afterwards, deliberately, largely because of what happened then. Additional Protocol I dates from 1977, not 1945, so “the Allies did worse” is not a legal argument and must never be offered as one.

It is an argument about the standard instead, and notice the direction of travel in the figures. One night over Tokyo killed more people than two years in Gaza. That fall is what targeting law and precision were built to produce, and it means campaigns now called uniquely savage would have been unremarkable, or exceptionally careful, in the wars the condemning countries fought themselves. A test never applied to anybody, including to themselves, is not a test. It is a conclusion looking for a number.

Say it like this: “Allied bombing killed about 60,000 French civilians while liberating France. I’m not defending that and nobody does. I’m asking whether the standard you’re using is one anyone has ever used, on anybody.”

The honest limit, and it cuts both ways

This is where advocates for Israel overreach, and it costs them the room. You have no more access to the targeting file than your opponent has, so anybody saying every strike in Gaza was proportionate is making a claim they cannot support. It is the same category error as saying the death toll proves the opposite, and a briefed opponent will name it as such.

International bodies have not treated the question as unanswerable either. On 19 June 2024 the UN Human Rights Office reported on six specific Israeli attacks carried out between 9 October and 2 December 2023, concluding that the principles of distinction, proportionality and precautions in attack may have been systematically violated. You are entitled to think that finding is wrong. You are not entitled to call proportionality unknowable and then treat your own confidence as knowledge.

So take the narrower position, because it is the one that holds. The word as used in public is not being used in its legal sense, and the legal question cannot be settled from a casualty figure by anyone, including you. That is less than an acquittal, and worth more, because it survives the follow-up.

Ask before you answer

The strongest move here is a question rather than an assertion. Ask what they mean by the word. Most people answer with a body count, and that is the opening: they have described a ratio and invoked a legal rule in the same breath, so you can explain the rule without contradicting anything they said. Correcting somebody and informing them look nothing alike from the outside.

The lines

Nobody follows these mechanically. Take the shape and use your own words.

What do you mean by disproportionate? I want to answer the thing you’re actually putting to me.

If the side that loses fewer people is automatically the guilty one, then the better a country’s shelters work the guiltier it gets. Nobody believes that, which is why the test isn’t a body count.

I’m not telling you every strike was proportionate. I’ve no idea and neither has anyone in this room. I’m telling you a casualty total doesn’t settle it either way.

The test is expected harm against anticipated advantage, on what the commander knew before the strike. Which strike do you want to run that on?

A reasonable commander and a human rights lawyer won’t price military advantage the same way. The tribunal that investigated NATO said so. So whose judgement is the test?

Where this backfires

Three ways, all avoidable.

Delivered as a correction, this sounds like pedantry over dead children, and the reaction you get will be deserved. Acknowledge what they are reacting to first, in plain language, then say the word has a technical meaning. The entry on the death toll carries the same risk and the same fix.

The second is worse. Reply to every allegation by saying that only the commander can know, and an audience hears a defence built to be unfalsifiable. They will be right. Keep the claim to what the word means. The same discipline applies to the genocide charge, where answering the legal threshold says nothing about whether other charges hold.

The third is the 1945 comparison used carelessly. Reach for it as a defence rather than as a test of their standard and you will be heard defending the firebombing of cities, which nobody can do and survive.

Three things by heart, about forty minutes

Learn three things by heart. The wording of Article 51(5)(b), which is one sentence, the three features of the test, and the reasonable military commander standard with the line about the human rights lawyer. That is about forty minutes of work, and it is what survives a follow-up question.

Learn the damage figures only if you will also learn the concession that goes in front of them. Said alone they hand the room to the other side.

Then: if somebody tells me Israel’s response is disproportionate, then I will ask what they mean by the word before I say anything else.

And a coping plan, because the likeliest obstacle is not a counter-argument. It is being told you are hiding behind a technicality: if that happens, then I will say the word they chose is a legal term and I am answering it as one, and that I am not claiming any particular strike was lawful.