The Limited Impact of Changes to the Defense Department’s Law of War Manual

On July 31, the Department of Defense released an updated Law of War (LOW) Manual, which establishes a presumption of civilian status, as well as significant revisions and additions on (a) feasible precautions to verify that the person or object is a military objective and (b) assessing information when conducting attacks. Before July, the manual was last updated by the Defense Department’s Office of General Counsel in 2016. The manual’s most recent updates are largely the result of several critiques by prominent subject matter experts and congressional members. While much of the discussion of the updates to the manual has focused on the new presumption of civilian status, to understand the impact of this change, it is necessary to review how decision-makers may overcome the presumption and how other nonupdated sections interact with the presumption. The update also added language on civilian harm mitigation policy yet does not provide the desired clarification of the line between law and policy.

Presumption of Civilian Status

The update to the Law of War Manual includes a reversal on whether there is a presumption of civilian status under customary international law. The previous version explicitly stated that no presumption exists, while the updated version states that under the principle of distinction, decision-makers must presume persons or objects are protected from being made the object of attack unless available information indicates they are military objectives.

The 2016 version of the LOW Manual stated:

5.4.3.2 [...] Under customary international law, no legal presumption of civilian status exists for persons or objects, nor is there any rule inhibiting commanders or other military personnel from acting based on the information available to him or her in doubtful cases. [footnote omitted]

The 2023 version of the LOW Manual states:

5.4.3.2 Classifying Persons or Objects as Military Objectives When Planning and Conducting Attacks. The law of war requires that only military objectives be made the object of attack and imposes other requirements for the protection of civilians and other protected persons and objects. In planning and conducting attacks, decisions or determinations that a person or object is a military objective must be made in good faith based on the information available at the time. In addition, these decisions must be consistent with the obligation to take feasible precautions to verify that the objects of attack are military objectives and with other obligations to seek to reduce the risk of incidental harm to civilians and other persons and objects protected from being made the object of attack.

Under the principle of distinction, commanders and other decision-makers must presume that persons or objects are protected from being made the object of attack unless the information available at the time indicates that the persons or objects are military objectives. This presumption is the starting point for the commander or other decision-maker’s good faith exercise of military judgment based on information available at the time. For example, if there is no information indicating that a person is a combatant or a non-combatant member of the armed forces, then commanders or other decision-makers must presume that person is a civilian. [footnotes omitted] [italics added]

The language of the 2016 version was critiqued extensively for claiming there was no legal presumption of civilian status and that no rule exists that inhibits commanders or other military personnel from acting based on available information in doubtful cases. To the latter point, no authority or scholar currently claims decision-makers cannot act if there is any doubt whatsoever. However, a reasonable approach would suggest there are levels of doubt, even above merely hypothetical or speculative considerations, that the law would not consider to be making a decision in good faith based on the information available at the time. 

This new presumption of protected status applies to both individuals and objects. This language is similar to two of the provisions of Additional Protocol I to the Geneva Conventions, of which the United States is not a party. Article 50 states in part, “[i]n case of doubt whether a person is a civilian, that person shall be considered to be a civilian.” Article 52 (3) states in part, “[i]n case of doubt whether an object which is normally dedicated to civilian purposes, such as a place of worship, a house or other dwelling or a school, is being used to make an effective contribution to military action, it shall be presumed not to be so used.” The diplomatic conference deliberately used “considered” instead of “presumed” when referring to people, to avoid confusion with the use of a presumption in favor of prisoner of war status after capture. While some observers have expressed concern about the ability to use a presumption on the battlefield, the analysis below begins from the position that it is not only possible but also already done in some circumstances.

Before analyzing the application of this “new” presumption of civilian status, it is important to note that it is already used within the U.S. government. First, as treaty law, the United States ratified the Amended Protocol II to the Convention on Certain Conventional Weapons in 1999, which as it applies to mines, booby traps, and other devices designed or of a nature to cause superfluous injury or unnecessary suffering, already prohibits in Article 3(8) the indiscriminate use of these weapons and “[i]n case of doubt as to whether an object which is normally dedicated to civilian purposes, such as a place of worship, a house or other dwelling or a school, is being used to an effective contribution to military action, it shall be presumed not to be so used.” The protocol’s letter of transmittal to the Senate stated that “[t]his prohibition is already a feature of customary international law applicable to all weapons.” While the letter does not explicitly mention the presumption, the support for the general prohibition in customary international law, without a caveat or correction to the subsequent presumption, lends support to its acceptance by the United States. Granted, this is only in regard to objects, not individuals. 

Second, per the Commander’s Handbook on the Law of Land Warfare, the Army, as “a matter of practice due to operational and policy reasons,” already applied this presumption of civilian status to individuals and objects. However, the Law of Armed Conflict Deskbook released by the International and Operational Law Department of the Army’s Judge Advocate General’s Legal Center and School explicitly states that the United States does not apply the presumption to objects as a matter of law and instead “applies the same test to all targets, requiring commanders to act in good faith based on the information available at the time and not on hypothetical or speculative considerations,” citing to the LOW Manual at 5.5.3.2.

The final argument is that the previous version of the LOW Manual all but explicitly states the presumption. In Section 5.4.3.2, the 2016 LOW Manual stated, “The law of war requires that only military objectives be made the object of attack.” Then, at 5.4.3.2, the 2016 LOW Manual stated, “[A]ttacks … may not be directed against civilians or civilian objects based on merely hypothetical or speculative considerations regarding their possible current status as a military objective.” This means that if there is no valid information indicating an individual or an object has military value, the decision-maker cannot presuppose that an individual or object is a military objective. To make an individual or object the target of an attack, there must be some evidence that the individual or object is a military objective. Without such evidence, the individual or object cannot be targeted, and those who cannot be targeted under the law are “protected” by the law from such action. This would seem to take a decision-maker to a similar but not identical place as an explicit presumption of civilian status. Additionally, there is an argument that this change in framing as one of protection by default will have a further psychological impact on the analysis of decision-makers, especially when it comes to overcoming doubt in deciding whether the individual or object is a military objective.

Acceptable Levels of Doubt 

In the previous LOW Manual framing, the individual or object in question is a blank slate, but before they can be targeted, the decision-maker has to determine that, based on the information available at the time, the individual or object is a military objective. In the new LOW Manual framing, the individual is presumed to be a protected civilian and the object a civilian object, but the decision-maker can determine that, based on the information available at the time, the individual or object is a valid military objective instead. In both framings, the evidence needed to indicate the individual or object is a military objective and, therefore, able to be legally targeted, would seem to be the same. However, the protected civilian framing at the beginning is expected to result in better protections for civilians and civilian objects.

Theoretically, the evidence required to indicate an individual or object in a specific situation is a military objective should be the same, but this starting point of a presumption of civilian status requires a new approach to analysis. For an individual or object to be a legal target, the presumption of civilian status must be rebutted. In the case of individuals, rebutting the presumption of civilian status requires specific information and analysis that would indicate the individual is not a protected civilian, due to either status or conduct. In a conflict scenario, information and analysis are never perfect. Decision-makers must decide what to do with imperfect information and how certain they are in the assessments provided to them.

Assuming the decision-maker would not be contemplating the decision on whether to target an individual or object unless there was at least some information that indicated the individual or object was not a protected civilian or civilian object, the analysis should begin with the confidence the decision-maker has in the information and assessments provided. There will likely frequently be some doubt whenever an individual is not wearing a military uniform and appropriate markings because the information is incomplete, or the assessment, while based on a reasonable interpretation of the information currently available, is not the only interpretation possible.

The LOW Manual states in 5.4.3.2, “Good Faith. Commanders and other decision-makers must assess whether persons or objects are military objectives in good faith. They must have an honest and genuine belief that a person or object to be attacked is a military objective.” The question then becomes, what is “good faith”? The above description appears to be wholly subjective, but the language that follows in the LOW Manual states that “the law of war requires commanders and other decision-makers to exercise professional judgment in making any assessment that a person or object is a military objective, and what is reasonable in making that assessment depends on the circumstances.” This adds an objective element, although the makeup of that objective element is not entirely clear, especially because it falls under a section entitled “Good Faith.” A reasonable interpretation is that the LOW Manual requires decision-makers to make choices that are both reasonable and in good faith—the decision-maker must decide what level of doubt is reasonable in exercising good faith in overcoming the presumption of civilian status in a specific situation. Geoff Corn, however, discusses how it could be interpreted in other ways, including with a purely subjective analysis or with a range of discretion so large it arguably negates the newly adopted civilian presumption.

The previous version of the LOW Manual already included good faith in decision-making regarding targeting, albeit in a slightly different form. In 5.3.2, “Decisions Must Be Made in Good Faith and Based on Information Available at the Time,” the 2016 LOW Manual stated:

Decisions by military commanders or other persons responsible for planning, authorizing, or executing military action must be made in good faith and based on their assessment of the information available to them at the time. … The requirement that military commanders and other decision-makers make decisions in good faith based on the information available recognizes that decisions may be made when information is imperfect or lacking, which will often be the case during armed conflict. [footnote omitted]

In 5.4.3, “Assessing Information in Conducting Attacks,” the 2016 LOW Manual stated:

Persons who plan, authorize, or make other decisions in conducting attacks must make the judgments required by the law of war in good faith and on the basis of information available to them at the time. For example, a commander must, on the basis of available information, determine in good faith that a target is a military objective before authorizing an attack against that target. [footnote omitted]

Additionally, 5.4.3.2, which disavowed the legal presumption of civilian status for persons or objects, also stated:

Attacks, however, may not be directed against civilians or civilian objects based on merely hypothetical or speculative considerations regarding their possible current status as a military objective. In assessing whether a person or object that normally does not have any military purpose or use is a military objective, commanders or other decision-makers must make the decision in good faith based on the information available to them in light of the circumstances ruling at the time.

The 2016 LOW Manual did not directly require a reasonableness standard but did, in multiple provisions, reference that the targeting decision must be based on the information available. The quoted language from 5.3.2 and 5.4.3 remain in the 2023 version, and the first sentence of the 5.4.3.2 quote remains as well. Therefore, the latest update of the LOW Manual more directly, but without explicitly using the term, requires a reasonable assessment in addition to one in good faith due to the additional language added to the “Good Faith” section of 5.4.3.2

The decision-maker must determine what level of doubt is reasonable in exercising good faith in identifying an individual or object as a military objective. This “level” is not static. While the concept of a continuum on a line has been critiqued as overly simplistic, the description is useful to emphasize the range of confidence levels that exist and that what is determined to be a reasonable confidence level may vary depending on the circumstances. While there is a level of doubt that would likely always be “too high” for a good-faith determination that an individual is a military objective (“mere speculation” in the language of the LOW Manual), it does not appear that specific levels could be set in advance of operations because what is reasonable will be heavily dependent on the narrow circumstances of the situation at hand. Because the reasonable confidence level is circumstance based, it should result from several circumstance-based factors.

In 5.4.3.2, the LOW Manual indicates several circumstances that may affect a decision-maker’s good-faith analysis by aiding in determining the level of certainty needed in a specific situation. These circumstances include the time and resources reasonably available, the risks to civilians from an erroneous decision, the risks to friendly forces, and the military advantage from the attack. Supplemental clarity on “the time and resources reasonably available” should be provided. From a practical standpoint, time should be considered both as “time required” to gather additional information and “time available” before the military advantage changes or the risk to friendly forces becomes immediate. Overall, the language reads as a range of factors that may be considered in determining the reasonable amount of doubt that would lead to a good-faith determination that an individual is a military objective but does not provide explicit guidance on how to evaluate or weigh these factors, or even in which situations some of them may not apply, leaving significant ambiguity in application. At a minimum, uncertainty, or doubt, in the assessment of both advantages and risks should affect the value attributed to them. Although this seems like a highly quantitative assessment, in practice, numbers cannot be associated with the values, and the process is more intuitive and experience based. 

There are two conflicting concerns about the presumption of civilian status and the level of doubt allowed in overcoming the presumption, which indicate how one sees the law. Either the ambiguity around the level of certainty required will hamstring decision-makers who will be too afraid to make decisions, or decision-makers will exploit the vagueness, knowing there is no real justiciability of the “good-faith” standard. Of course, there is a third option, that decision-makers will, by and large, act reasonably as dictated by the context of the situations they find themselves in. This is generally the status quo, and for the militaries in which it is not, this presumption change would change nothing. However, the risk is that with this more instinctual process, cognitive biases, as discussed later, may cause a decision-maker to incorrectly weigh doubt, especially as it relates to certainty of identification.

Additionally, when doubt does exist, the LOW Manual update requires decision-makers to take feasible precautions to verify that the person or object is a military objective, as outlined in a new section, 5.5.3, “Feasible Precautions to Verify Whether the Objects of Attack Are Military Objectives.” So, while there are circumstances to evaluate based on the specific situation, there is also a requirement to reduce doubt through feasible precautions, which may include actions such as some types of warnings before attacks that may affect the civilian population and reviewing previously approved targets at reasonable intervals as well as when warranted in light of fresh information and changing circumstances. If the decision-maker has time and resources available to take steps to verify whether the object of an attack is a military objective, they must take feasible precautions. “Feasible” is used as a term of art here, in which it implies not just that time and resources exist but taking into consideration military and humanitarian concerns and appropriately prioritizing all the other demands on time and resources.

Section 5.5.3 explains that if time is not available, decision-makers may make decisions and take actions “at the speed of relevance … based on their good faith assessment of the information that is available to them at the time.” Feasible, as expounded upon in 5.2.3.2, “What Precautions Are Feasible,” sets the standard as one of due regard or diligence and not an absolute requirement to do everything possible. There is ambiguity around what may actually be considered feasible or not under this standard—this ambiguity is not new. The update to the LOW Manual does provide examples of what may be included in feasible precautions to verify the person or object is a military objective, which, although helpful, does not address the ambiguity around feasibility. The examples, as provided in 5.5.3 include, among others, “gathering more information, such as visual identification of the target through intelligence, surveillance and reconnaissance platforms” and “reviewing previously approved targets at reasonable intervals as well as when warranted in light of fresh information and changing circumstances.” Understanding feasibility will continue to be a challenge, especially in the context of future large-scale combat operations in which competing priorities will vie for resources, including intelligence, surveillance, and reconnaissance assets.

Finally, in understanding the effect this change in the presumption of civilian status will have in practice, it is vital to look at the effects of other sections of the LOW Manual that did not change. Even if the presumption of civilian status results in a more protective analysis, which from the above analysis, it is not clear it will, there are still at least two critical analyses in the LOW Manual that may reduce this protection, one involving losing the protected status and the other involving providing less weight to civilians in the proportionality analysis.

First, as the letter from Sen. Dick Durbin (D-Ill.) and Rep. Sara Jacobs (D-Calif.) to Defense Department General Counsel Caroline Krass highlighted, the LOW Manual suggests that civilians working on “war-sustaining activities,” such as an oil refinery that generates revenue, may be directly participating in hostilities, which would make them lose their protected civilian status and allow them to be targeted under the law. In relevant part, the LOW Manual states in 5.8.3 that “[t]aking a direct part in hostilities extends beyond merely engaging in combat and also includes certain acts … that effectively and substantially contribute to an adversary’s ability to conduct or sustain combat operations.” Section 5.6.8.5 provides examples of economic objects in war-supporting or war-sustaining industries that are considered military objectives, such as electric power stations and oil refining and distribution facilities. Turning to the “conditions that may be relevant” in 5.8.3 to determine whether an act by a civilian is directly participating in hostilities, enough ambiguity exists to allow for an interpretation to include a large amount of war-sustaining activities. This may significantly expand the scope of activity that is included in “directly participating in hostilities” because civilians working at “economic objects with … war-sustaining industries” may lose their protected civilian status and be able to be directly targeted. This is not a new critique but a reminder of the pervasive possible impacts of such language. The Office of General Counsel chose not to modify or add clarifying language to this issue in the LOW Manual update.

Second, also detailed in the Durbin-Jacobs letter, the LOW Manual has ambiguous language on whether civilians, because they are employed in or on military objectives, could be discounted in a proportionality analysis. Specifically, the LOW Manual states that “[w]hen the attacking force causes harms that are the responsibility of the defending force due … to the employment of civilian personnel in or on military objectives, the responsibility of the defending force is a factor that may be considered in determining whether such harm is excessive.” The later section 5.12.3.3, “Civilian Workers Who Support Military Operations In or On Military Objectives,” states: 

In general, reasonable steps must be taken to separate the civilian population from military objectives, such as by removing members of the civilian population from the vicinity of military objectives and combat operations. However, sometimes civilian personnel work in or on military objectives in order to support military operations. …  Such persons assume a certain risk of injury.⁴⁶² Provided such workers are not taking a direct part in hostilities, those determining whether a planned attack would be excessive must consider such workers, and feasible precautions must be taken to reduce the risk of harm to them. Those making such determinations may consider all relevant facts and circumstances.

⁴⁶² U.S. Comments on the International Committee of the Red Cross’s Memorandum on the Applicability of International Humanitarian Law in the Gulf Region, Jan. 11, 1991, Digest of United States Practice in International Law 1991-1999 2057, 2063 (“Likewise, civilians working within or in the immediate vicinity of a legitimate military objective assume a certain risk of injury.”). See also Bothe, Partsch, & Solf, New Rules 295 (AP I art. 50, ¶2.2.2) (“By being within, or in the vicinity of a military objective, these civilians assume the risk of collateral injury from the effects of attack. It is also doubtful that incidental injury to persons serving the armed forces within a military objective will weigh as heavily in the application of the rule of proportionality a that part of the civilian population which is not so closely linked to military operations.”); The Sixth Annual American Red Cross-Washington College of Law Conference on International Humanitarian Law: A Workshop on Customary International Law and the 1977 Protocols Additional to the 1949 Geneva Conventions: Remarks of Lieutenant Colonel Burrus M. Carnahan, 2 American University Journal of International Law and Policy 505, 510 (“State practice, on the other hand, suggests the existence of at least one intermediate category: persons who, while not taking a direct part in hostilities, are so intimately connected with a military objective that they have forfeited the right to be free from risk of collateral damage.”); Waldemar A. Solf, Protection of Civilians Against the Effects of Hostilities under Customary International law and Under Protocol I, 1 American University International Law Review 117, 131 (1986) (“Thus, while a civilian may not lose his protection against individualized attack while working in a munitions plant, he assumes the risk of collateral injury when he is in the vicinity of the munitions plant, although he continues to retain full protection while at home.”).

The footnote raises additional questions. Footnotes are not used within the LOW Manual as footnotes typically are within legal text; according to a briefing Marty Lederman received from department officials, “readers should not assume the Department approves of, or agrees with, propositions of law contained in the footnotes.” However, considering the LOW Manual itself states at 1.2.1 that “[t]his manual uses footnotes to provide sources or cross-references to other sections of the manual in order to clarify, elaborate on, or support the main text,” the average reader will likely read this footnote to imply that there is a sliding scale of value that can be applied to civilians in a proportionality analysis. This remaining language of the LOW Manual will somewhat temper the protective effects of the change to a presumption of civilian status, to the delight of some and the chagrin of others.

Addressing Cognitive Bias in Identification

The manual also includes an update to 5.4.3.3 (previously 5.4.3.1), “Heightened Identification Requirements in Conducting Attacks,” integrating recent civilian harm mitigation policy measures from the department’s Civilian Harm Mitigation and Response Action Plan. The update includes reference to practices—in the form of policies—to mitigate cognitive biases that result in misidentification of targets in combat. However, the language fails to clarify the line between policy and law as intended. The LOW Manual recognizes in footnote 111 of Chapter 5 that several additional civilian harm mitigation policy measures have been implemented, including various heightened identification requirements. For example, the U.S. policy standard for use of force in counterrorism operations outside the United States and areas of active hostilities was a “near certainty that the terrorist target is present.” However, many other civilian harm mitigation policy measures, such as integrating the civilian environment into the operational environment part of operational plans or technical training for imagery analysts on accurate identification of civilians and collateral objects, are unrelated to heightened identification requirements.

The specific example provided by the LOW Manual in 5.4.3.3 is “practices to mitigate cognitive biases that result in misidentifying targets in conflict.” In this example, a decision-maker would likely know that cognitive biases are resulting in the misidentification of targets in combat, at least some of the time. Can a decision-maker be said to be truly acting in good faith and have taken feasible precautions to verify that the objectives to be attacked are military objectives if they know cognitive biases may be affecting the levels of certainty of information and assessments provided, yet they did nothing to overcome this deficiency? If not, the legal standard is not met. While others have addressed feasible precautions as it relates to cognitive bias in depth, the good-faith analysis should not be overlooked, particularly because of the increased focus due to the new presumption of civilian status, and the underlying concept of honest, genuine, and reasonable belief that directly ties into the American military’s ethos of moral leadership. 

The good-faith analysis required here does not necessarily mean the decision-maker must take additional identification steps, but that it should factor into their analysis in determining whether the circumstances would allow them to make a good-faith decision that the individual or object is a military objective. The confusion here likely arises from placing this reference to cognitive biases in 5.4.3.3, “Heightened Identification Requirements in Conducting Attacks.” While one approach to overcoming cognitive bias that results in misidentification could be to increase the identification requirement, it is not the only option. Instead, a good-faith analysis could involve reviewing and, as appropriate, reducing the level of certainty in the accuracy of specific analyses or conclusions in circumstances where the underlying information was known to suffer from confirmation bias frequently. This is supported by a new example provided in 5.5.3, “Feasible Precautions to Verify Whether the Objects of Attack Are Military Objectives,” which specifically provides that a feasible precaution may include “[r]eviewing the accuracy and reliability of the information supporting the assessment that a potential target is a military objective.” For example, a decision-maker may decide that a high confidence assessment identifying an individual as a military objective is of only moderate confidence. Members of the Department of Defense would benefit from more clarity on requirements of law and requirements of policy.