When the world's at stake,
go beyond the headlines.

Urban warfare is often invoked as an alibi: dense terrain, an embedded enemy, human shields, imperfect intelligence. These conditions are real but they do not, on their own, explain outcomes. Every modern military that has fought in cities has faced uncertainty, civilian presence, and adversaries who exploit it. What distinguishes campaigns is not whether these problems exist, but how commanders choose to respond to them, whose risk is prioritized, and what forms of harm are treated as acceptable.

The war in Gaza has produced civilian death on a scale that cannot be explained by inevitability alone. More than 73,000 people — over 71,000 Palestinians and nearly 2,000 Israelis — have been killed in a battlespace smaller than Philadelphia.

This outcome was not the unavoidable result of fighting Hamas in dense urban terrain. It was the product of institutional choices about how force was applied under uncertainty. Israeli targeting practices consistently defined military value broadly, treated intelligence gaps permissively, and shifted risk away from Israeli forces and onto civilians. Those choices shaped humanitarian outcomes and now threaten Israel’s strategic position.

Having commanded U.S. forces in urban combat, I have seen how narrow the margin can be between military necessity and civilian catastrophe. Uncertainty does not disappear in war, but it should tighten restraint rather than license destruction. In Gaza, uncertainty repeatedly functioned in the opposite direction. The result was not just tragic harm, but a pattern of civilian death that reflects how decisions were made, what standards were enforced, and what consequences commanders expected to face.

Numbers alone cannot settle questions of responsibility. To understand whether civilian harm in Gaza was lawful, avoidable, or excessive, we must examine process rather than intent, and institutional practice rather than rhetoric. That requires asking not whether the Israeli military possessed modern targeting tools, but how it used these under pressure, how it handled intelligence uncertainty, and how it allocated risk when civilians could not be separated from the battlefield.

What the Numbers Show

Despite methodological differences, independent U.N. reporting, human rights investigations, and peer-reviewed demographic studies all point to the same conclusion: The majority of those killed in Gaza have been civilians. By even the most conservative estimates, assuming that all adult males were combatants, civilians still account for roughly two-thirds of reported deaths.

In early 2025, The Lancet estimated that recorded civilian trauma deaths had been undercounted by roughly 40 percent and that most indirect deaths were absent from official tallies, placing total Palestinian deaths plausibly above 100,000.

Other independent analyses, including Israeli intelligence reporting cited by international media and the Washington Institute indicate that civilians likely comprise more than 80 percent of this higher figure.

But numbers alone cannot resolve whether harm was unavoidable. That question requires a shift from outcomes to process — how force was applied, and under what constraints.

When Civilian Harm Is Lawful

The central issue, therefore, is whether the Israeli military took all measures that could reasonably be expected to reduce civilian harm while pursuing military objectives. International humanitarian law provides a benchmark. Its core principles of distinction and proportionality require attacks to be directed at legitimate military objectives and prohibit civilian harm that would be excessive in relation to anticipated military advantage.

“Excessive,” however, is not a mathematical threshold. It is a contextual judgment shaped by the value of the target, the quality of intelligence, the means employed, and the feasible alternatives available to commanders. Uncertainty is inherent to war. It does not bar action, but it should shape how and when force is applied — tightening constraints as doubt increases, rather than serving as a rationale for shifting risk onto civilians.

Nowhere to Go

Whether legal standards can function in practice depends first on whether a military makes all reasonable efforts to remove civilians from harm’s way.

Gaza’s pre-war density — more than 8,000 people per square kilometer — was compounded by evacuation orders that pushed much of the population into a shrinking fraction of the territory. In al-Mawasi, aid agencies reported densities exceeding 34,000 people per square kilometer, with minimal infrastructure.

At the same time, Gaza was effectively sealed. Civilians could be displaced internally but were rarely permitted to leave. U.N. officials repeatedly warned that there was “no safe place in Gaza,” as bombardment followed people from one designated zone to another.

Other militaries have faced similar dilemmas. In Fallujah and Mosul, U.S. and coalition forces made sustained efforts to enable civilian evacuation.

Israel had the technical and operational capacity to establish internationally monitored safe zones early in the campaign. Evacuation corridors could also have been stabilized through clearly announced and monitored movement windows. None of this would have ended Gaza’s suffering, but it could have reduced the number of civilians trapped between evacuation orders and violence.

Instead, evacuation orders compressed civilians into areas that were subjected to repeated strikes. Using the resulting population density to explain civilian deaths, rather than to require greater restraint, misstates the operational problem.

How Militaries Exercise Restraint

When civilians cannot be fully separated from the battlefield, protection ultimately turns on how force is applied.

The U.S. military follows a formal collateral damage estimate process. A strike may proceed only if projected civilian harm remains below an authorized threshold. If not, commanders are to modify or abandon the attack.

I saw this process firsthand. In Iraq, where I commanded an infantry battalion, strikes were frequently canceled when civilians were detected, even when insurgents were present. Later, while commanding a special operations task force during the Mosul campaign, I approved hundreds of strikes. If there was any indication civilians might be present — such as ambiguous figures on a drone feed or voices on intercepted communications — the strike was delayed.

That restraint was not absolute. It was balanced against the mission and our obligation to protect Iraqi security forces as they battled to retake territory held by the Islamic State. Still, when civilian risk outweighed the tactical value of a strike, protection of the population took precedence over killing the enemy.

This expectation carried real consequences: Strikes that caused avoidable civilian harm or revealed breakdowns in targeting discipline routinely triggered investigations, loss of strike authority, and career-ending reprimands.

Looks Like Restraint

This comparison lies at the core of one of the most common defenses of Israel’s conduct of the war: the claim that the Israeli military employs a U.S.-style targeting process and should therefore be judged by comparable standards. John Spencer, a U.S. Army veteran and scholar of urban warfare, has perhaps been the most prominent voice forcefully asserting that Israel has taken unprecedented measures to avoid civilian harm. He points to formal targeting cells, the use of precision-guided munitions, and extensive warning and evacuation measures as evidence.

Spencer’s argument reflects a broad tendency to treat the existence of process as proof of restraint while discouraging scrutiny of how those tools are actually used. The decisive question is not whether such systems exist, but how they function under pressure. And it is here that a different picture emerges.

For years, Breaking the Silence —  a nonprofit composed of current and former Israeli soldiers — has collected testimony indicating that target development often rests on fragmentary or outdated intelligence, broad and permissive definitions of “military infrastructure,” and limited real-time updating of civilian patterns of life. Soldiers describe entire residential apartment blocks being designated as legitimate targets on thin leads rather than the confirmed presence of a military objective.

These accounts are not the sole basis for this assessment. Since the Oct. 7 attacks, I have interviewed over 50 Israeli officers who served in Gaza, including combat commanders, intelligence officers, and personnel directly involved in strike nomination and execution. While these interviews were conducted independently and are not publicly attributable, they consistently echoed the same patterns described by Breaking the Silence: permissive target definitions, heavy reliance on inference rather than confirmation, and little effort to update civilian presence once a target had been approved. Israeli intelligence officers have acknowledged that a building may be designated as an ammunition depot without knowing whether it actually contains weapons — or nothing at all. When the military value of a target is speculative, proportionality calculations become fragile by definition.

Taken together, these practices reflect more than permissive risk tolerance. They amount to a breakdown in the application of distinction and proportionality as those principles are normally understood and enforced in modern Western militaries.

Spencer argues that outside observers “cannot say” whether a smaller munition could have achieved the same effect and therefore are not qualified to make judgments about proportionality. But this misses the point. Commanders inevitably operate with assumptions. The issue is whether those assumptions are grounded in current, corroborated intelligence or outdated target assessments that fail to account for change.

That distinction matters. In my own experience approving strikes in urban combat, uncertainty about civilian harm did not license action. It slowed it. Questionable intelligence, an outdated pattern of life, and an unclear feed were not minor concerns. They were reasons to delay or abandon the strike altogether. Uncertainty is unavoidable. Treating uncertainty as a reason to accept civilian harm is a choice.

Spencer also points to Israel’s use of precision-guided munitions as evidence of exceptional care. But precision in delivery does not equate to precision in targeting. A weapon can strike exactly where intended and still produce excessive civilian harm if the target was defined too broadly, the intelligence was stale, or civilian presence was inadequately assessed.

The same is true of warning measures such as roof knocking, which Spencer and others portray as humanitarian innovations. But the evidence we have from Israeli sources, not just outside advocates, challenges that characterization. Soldiers describe the practice as ineffective and often dangerous. In practice, roof knocking relies on the use of explosives as a warning mechanism. From personal experience under artillery and air attack, I can say plainly that distinguishing between a munition intended to warn and one intended to kill is neither intuitive nor reliable.

Warnings that do not reliably remove civilians from danger do not meaningfully mitigate harm. In a battlespace where electricity, cellular networks, and internet access have been deliberately degraded, claims about phone calls, text messages, and warning strikes ring hollow. In that context, warnings function less as protection than as justification for proceeding with a strike.

Custody Lost

Civilian movement in urban warfare is fluid. A building that appears empty on a morning drone feed may shelter families by noon. In Gaza, repeated evacuation orders intensified this volatility, while Israeli targeting often relied on intelligence that was days or even weeks old. The resulting collateral-damage estimates could be technically precise while remaining disconnected from realities on the ground.

Here, too, there is divergence between Israeli and U.S. practice. Over the course of the wars in Iraq and Afghanistan, the U.S. military came to emphasize the “unblinking eye”: the requirement to maintain continuous observation of a target from nomination through the moment of strike.

In Gaza’s sealed and densely populated battlespace, loss of custody was a predictable source of error. The April 2024 World Central Kitchen strike illustrates this gap. Coordination existed, markings were clear, yet custody was lost — and seven aid workers were killed.

I have seen both sides of this equation. As a commander, I approved strikes under the pressure of imperfect intelligence, aware that a single decision could save friendly forces or kill civilians. Later, as the head of a humanitarian nonprofit delivering aid on the front lines in Ukraine, I encountered the same problem from the opposite side of the coordination boundary. Humanitarian movement depends on disciplined procedures and continuous tracking.

Taken together, these points of contrast with U.S. practice undercut the claim that Israeli targeting should be evaluated as functionally equivalent. While the Israeli military possesses many of the same tools, it has not adopted the same institutional culture of restraint that emerged from U.S. experience. Israeli practice accepts a higher risk to civilians as normal.

This helps explain why the Israeli military routinely conducts strikes that American commanders would almost certainly have rejected. The destruction of entire residential towers to eliminate a single militant apartment, the bombing of multi-family homes where civilians were likely present, the routine use of 1,000- and 2,000-pound munitions in dense urban blocks — these are not aberrations. They reflect a calculus that accepts predictable civilian death on a scale that U.S. doctrine would not permit.

In comparable U.S. operations, strikes conducted with stale intelligence, loss of custody, or predictable civilian presence would have triggered legal review, command intervention, and curtailment of strike authority. They would not have been treated as unfortunate outcomes of a difficult fight, but as indicators of command failure requiring correction.

Limits to the “Human Shields” Argument

Defenders of Israel’s conduct frequently point to Hamas’ use of civilians as shields, arguing that civilian casualties in Gaza are an unavoidable consequence of enemy tactics rather than Israeli operational choices. Spencer advances this claim in categorical terms, asserting that the sole reason for civilian deaths in Gaza is Hamas and that Israeli precautions fully discharge legal and moral obligations.

Armed groups in Gaza do operate from within civilian areas, exploit protected infrastructure, and seek to constrain Israeli freedom of action through civilian presence. But international humanitarian law distinguishes between fighters operating among civilians — a common feature of urban insurgency — and the deliberate use of civilians as shields.

The available evidence provides little support. A detailed analysis by the Middle East Institute documents how Israeli officials frequently invoke “human shields” as a generalized justification for civilian casualties while rarely providing case-specific evidence demonstrating that civilians were intentionally coerced, positioned, or prevented from leaving for the purpose of shielding military objectives at the time of attack. By conflating proximity with coercion, Spencer’s argument lowers the threshold for acceptable civilian harm. By his calculus, individual strikes need not be examined — they are absolved in advance.

In Mosul, the Islamic State employed tactics closely resembling (and in some cases exceeding) those attributed to Hamas, including embedding within civilian areas, use of protected sites, and coercive control over civilian movement. Yet civilian-to-combatant ratios in those battles were significantly lower than those observed in Gaza. If estimates of civilians killed by the Islamic State and not coalition forces are used instead, the contrast is even sharper.

This does not absolve Hamas of responsibility for endangering civilians. But international law does not permit attacking forces to treat enforced civilian presence as an exculpatory condition rather than a constraint. When civilian harm occurs at scale and with demographic consistency, the decisive variable is not simply whether fighters embedded among civilians, but how the attacking force chose to respond. And as comedian Bill Burr has observed in criticizing the “human shield” defense, if you want to beat up your neighbor but he is holding a baby, you do not try to punch him through the child.

Spencer’s argument and others of this kind have been widely cited in Israeli media and official commentary to assert that responsibility for Gaza’s civilian death toll lies solely with Hamas. But these claims do not withstand scrutiny. Informed judgments about legality and restraint turn on how targeting decisions were made, the quality of intelligence available, the choice of weapon, and the degree of error commanders were willing to accept. In Gaza, these decisions consistently shifted risk onto the civilian population.

Institutional Culture and the Allocation of Risk

So, why did the Israeli military repeatedly accept levels of uncertainty and civilian risk that American commanders would not? One plausible explanation lies in how Palestinian civilian life was assigned value.

In my own experience approving strikes in urban combat, ethical decision-making was shaped less by written rules than by the signals sent from senior leaders. I was repeatedly reminded that we fight with the values we represent. We do not adopt those of our enemy.

In Gaza, senior-level rhetoric and policy decisions signaled a markedly different environment. In leaked recordings, Maj. Gen. Aharon Haliva — then head of Israeli military intelligence — stated that for every person killed on Oct. 7, “50 Palestinians must die,” adding that “it doesn’t matter now if they are children.” He described mass Palestinian deaths as “necessary” to send a deterrent message.

Israeli Defense Minister Yoav Gallant’s declaration of a “complete siege” on Gaza — cutting food, electricity, fuel, and water — was accompanied by explicitly dehumanizing language. Announcing the policy on Oct. 9, Gallant stated: “We are fighting human animals, and we are acting accordingly.” Israeli President Isaac Herzog’s assertion that “an entire nation out there is responsible” further blurs the institutional line between civilian and combatant.

Such statements do not determine individual targeting decisions, but they shape the environment in which those decisions are made: how civilian life is valued, how much civilian harm is expected to be scrutinized, and how much is implicitly excused.

Investigative reporting on Israeli targeting practices indicates a permissive posture toward civilian risk, including reliance on automated target generation and acceptance of high civilian-casualty thresholds. The Israeli military disputes aspects of this reporting, and it should be treated with appropriate caution. But the credible possibility that such thresholds were considered acceptable is a better explanation of civilian casualty figures than Spencer’s argument of unprecedented restraint.

Taken together, these factors help explain why risk normally borne by commanders was instead shifted onto civilians.

Conclusion

Civilians are not killed at scale because urban warfare makes restraint impossible. They are killed when the protection of civilian life is not treated as a governing requirement of how force is applied.

In every war, commanders must accept risk. The decisive question is who bears it. In Gaza, risk was routinely transferred from Israeli forces onto civilians. The resulting toll is therefore not an unavoidable consequence of dense urban combat, but the product of deliberate decisions about how force was used and whose lives were endangered.

That distinction matters because civilian harm at this scale is not only a humanitarian or legal concern. It is a strategic liability, straining alliances and fueling the dynamics of recruitment, radicalization, and insurgent regeneration observed repeatedly in Iraq and Afghanistan.

Urban warfare cannot be made clean, and civilian harm cannot be eliminated. But civilian protection ought to function as a real operational constraint. Other militaries learned — often painfully — that when uncertainty defaults to force, the long-term costs are severe. Israel now faces a choice: to treat Gaza as an exceptional case, or as a warning about how civilian harm undermines both military effectiveness and long-term security.

Andy Milburn is a retired U.S. Marine Corps colonel and former infantry and special operations officer. He fought in Fallujah, led an infantry battalion in Iraq’s Anbar Province, and commanded a special operations task force during the campaign to retake Mosul from the Islamic State. He later led a humanitarian organization operating on the front lines in Ukraine. He is author of When the Tempest Gathers: A Marine Special Operations Commander at War.

Image: Tasnim News Agency via Wikimedia Commons