The risk management framework

To assess the contribution of the CHMR policy framework to limiting civilian harm from military operations, it is instructive to take the combatant commander’s perspective. Inherent to command decision-making throughout planning and operations is the management of risk, particularly risk associated with the use of lethal force. Much of a command’s “battle rhythm”Footnote ²⁰ – including operational planning teams, cross-functional working groups and cells, and decision boards – is geared towards identifying potential risk, estimating its severity, assessing whether the consequences of action would outweigh its potential benefits, and applying the means to reduce or offset that risk in mission planning.Footnote ²¹ Risk management is continuous and cyclical in all phases of planning and operations. It allows a commander to define her or his level of risk tolerance (including as informed by IHL obligations), make it clear to all personnel, and issue corresponding directives and orders. The process by which these considerations are weighed brings structure to an inherently subjective decision-making process.Footnote ²² Specifically, risk management is

A robust risk management process becomes even more important when command authorities are delegated and decentralized, whereby

The notion of risk and how to manage it is thus co-created by commanders holding ultimate responsibility and those receiving delegated authority. This depends on having a continuous shared understanding of risk factors and how they may evolve throughout an operation, as well as command awareness of whether the necessary information and capabilities are in place to mitigate the risk, including risk to civilians. One of the richest civilian harm studies commissioned by the DoD, RAND’s examination of the 2017 campaign in Raqqa, found that the Special Operations Joint Task Force leading Raqqa operations did not have a fully staffed operational-level headquarters and lacked functions for battlefield-shaping operations, leading to heavy reliance on dynamic targeting, which inherently carries greater risk of civilian harm. In the same operation, a US ground forces commander noted that rotations in personnel led to decreased proficiency and attention to civilian casualty risks, requiring thirty to sixty days for proficiency to improve. This review found that commander’s intent did include appropriate emphasis on minimizing civilian harm, but it was communicated only in a broader combatant command policy and tactical directive. Commander’s intent will have greater impact if it is explicitly integrated into specific contingencies, directly linked to objectives and tasking orders, and communicated regularly to operational forces at the smallest unit levels.Footnote ²⁵

This is where a structural blind spot becomes apparent, with consequences for managing civilian harm risk. The overarching Joint Risk Analysis MethodologyFootnote ²⁶ framework sets out three types of risk: strategic risk (to national interest), risk to mission (to mission execution), and risk to force (to generating, sustaining and developing forces). Assessments of all three types of risk are conducted by staff and flow upwards in a chain of command. “Risk to civilians” is not explicitly identified. The presence of civilians and the potential for non-combatant casualties is commonly mentioned in military guidance as a contextual risk factor,Footnote ²⁷ alongside weather and terrain, but it is not elaborated in risk management frameworks or treated as an overarching governing factor alongside other types of risk.

Mirroring the commander’s risk management framework, CHMR policy is based on concepts of risk and risk reduction and provides commanders with an expanded vocabulary, operational concepts, capabilities and procedures for managing civilian harm risk. Doctrinal updates incorporating CHMR policy now include among a commander’s strategic considerations “efforts to protect and restore elements within the civilian environment” alongside international law, diplomacy and other factors.Footnote ²⁸ Updates to Joint Force Headquarters doctrine also emphasize the joint force commander’s responsibility to “ensure forces are prepared to mitigate and respond to civilian harm during operations”.Footnote ²⁹ Following the first combatant command exercise integrating CHMR, command leadership and senior staff highlighted the value of identifying, assessing and reporting risk to civilians up the chain, alongside other overarching risks.Footnote ³⁰ Prominent placement of civilian harm risk considerations along these lines lays a foundation for inculcating the protection of civilians as a constant consideration, conveying its importance to subordinate personnel and counteracting the factors contributing to civilian harm.

Cognitive bias

Analysis of how harm happens has illuminated the central role of cognitive bias, including flawed assumptions and path dependencies in a long sequence of intelligence, planning, operational and tactical decisions. Flawed assumptions may be explicit and embedded in planning – for example, a common assumption is that as hostilities escalate, most or all civilians will flee, thereby diminishing the need to exercise precaution and restraint. Operational personnel may believe in the superior situational awareness of their own forces, leading them to discount external information sources about the presence of civilians and civilian casualties as irrelevant, inaccurate or even malicious disinformation. Group behaviour within a command structure may reinforce flawed assumptions and suppress alternative perspectives, due for example, to pressure to agree, cooperate and maintain speed and efficiency in operations. There may be habitual reliance on familiar and “good enough” shortcuts whereby tools designed for other problems are applied to the situation at hand.Footnote ³¹ A failure to question beliefs and revise assumptions relating to civilian populations in planning and operations can create path dependencies – whereby past choices influence the menu of options available – that lead to civilian harm. A lack of familiarity with patterns of civilian harm, and a lack of training on common fallacies and errors resulting in civilian harm, can leave forces ill-equipped to recognize and counteract these tendencies.

Other assumptions may operate at the tactical level, such as perceptions of hostile intentFootnote ³² leading to the mistaken targeting of a civilian. Tactical commanders and targeteers may have unrealistic confidence in the intelligence on which tactical decisions are based and in the predictability of the effects of precision-guided munitions in populated areas. In addition, implicit assumptions may shape planning and operations, such as a belief that the best capabilities are already being deployed, that all feasible and required steps are already being taken to limit civilian harm, and that any civilian harm which does occur is a tragic but unavoidable consequence of war.

Cognitive bias in warfighting has been described as follows: “When relevant information is not instantly accessible, the likelihood of events cannot be reliably estimated; circumstances are marked by stress and time constraints; and decision-making processes become prone to systematic errors.”Footnote ³³ Scholarship on cognitive bias has established several distinct patterns, including anchoring bias, confirmation bias, optimism bias and false dichotomies, which compromise decision-making and mission effectiveness.Footnote ³⁴ The 29 August 2021 US strike in Kabul is a now well-documentedFootnote ³⁵ example of anchoring bias in a high-pressure situation whereby initial intelligence mistakenly identified one civilian as an Islamic State of Iraq and Syria (ISIS) operative preparing to carry out an attack, leading to biased interpretations of subsequent information and ultimately resulting in the death of ten civilians.

According to US Air Force Lieutenant General Sami D. Said, who was tasked with investigating the strike, “what I found is the disconnects were an aggregate process breakdown in which many people are involved, and it wasn’t any particular individual that was causal to that”.Footnote ³⁶ Comparable aggregates of faulty assumptions and cognitive bias – and failure to interrupt path dependency – have been cited in other past cases of civilian harm, such as the protracted series of strikes on a Médecins Sans Frontières-supported hospital in Kunduz, Afghanistan.Footnote ³⁷ An aggregation of errors dispersed across several operational functions and decision points speaks to the importance of addressing the critical gaps not in a piecemeal manner, but holistically and systemically.

The US military’s approach to overcoming the negative effects of cognitive bias is known as “red teaming”. It is described as

Trained red teaming personnel serve as designated contrarians tasked with questioning problem statements, checking assumptions, interrogating bias, identifying influential variables which are not yet being considered and generating alternatives. This capability has been primarily used to simulate the adversary’s perspective as a means of stress-testing and strengthening planning to fulfil mission objectives and achieve desired end states. The CHMR-AP repurposed this capability with explicit tasking to use a civilian harm risk lens in the joint target cycle.Footnote ³⁹ Numerous staff highlighted the broader value of red teaming for improving civilian harm risk reduction and called for red teaming training holistically tailored to CHMR to be made available to staff with key CHMR-related functions across intelligence, planning and operations.

Information flow and the civilian environment

Civilian harm risk in armed conflict cannot be reduced to zero, and there will never be a perfect information environment leading up to and during an armed conflict. However, the premise of CHMR is that patterns of harm are often knowable and foreseeable, as are the factors contributing to harm, and therefore some civilian harm can be avoided or at least minimized. Critically, the process of risk management relies not only on communication of command intent but also on effective horizontal and vertical information flow within a command structure. CHMR policy directly addresses the common absence of explicit end states and mission objectives relating to the civilian population and directs that “strategic guidance, goals, objectives, and mission orders clearly incorporate end-state objectives with respect to the protection of civilians and civilian objects as part of overall mission objectives”.Footnote ⁴⁰ A commander whose end states and mission objectives include the greatest possible preservation of the civilian environment infuses their command with an anticipatory posture which drives information flow and the planning process.

CHMR policy envisions information flow as being blended into existing processes and not as a parallel add-on, an approach whereby “mitigating civilian harm and operational effectiveness are two sides of the same coin”.Footnote ⁴¹ Tailoring information on civilian harm risk within a command has become a central theme for operationalizing CHMR – specifically, what and how information is generated, curated, analyzed, circulated and applied in decision-making. In this regard, US doctrine emphasizes that commanders must be able to visualize the battlespace from the current state to the end state, and the sequence of actions in between. More than a physical observation, battlespace visualization “is a cognitive reconstruction of the battlefield framework that allows the commander to see the enemy and friendly forces, terrain and weather, in terms of time, space and purpose”.Footnote ⁴² By introducing the notion of “the civilian environment”, comprised of “the civilian population and the personnel, organizations, resources, infrastructure, essential services, and systems on which civilian life depends”,Footnote ⁴³ the DoD’s CHMR policy sought to ensure that commanders would be able to visualize these protected elements in relation to that battlespace, including how they overlap and co-mingle with the adversary.Footnote ⁴⁴

The CHMR policy description of the civilian environment has been criticized for being more of a list than a definition.Footnote ⁴⁵ However, it had the practical effect of catalyzing a departmental effort, for example through cross-departmental working groups, to grapple with its implications across the land, air, maritime, cyber and space warfighting domains as well as to further develop and operationalize core concepts. External resources, such as analysis and tools addressing indirect harm in urban warfare,Footnote ⁴⁶ were essential to understanding the interconnectedness of systems and services, envisioning the cascading and reverberating effects which could arise from disruption of critical infrastructure nodes, questioning assumptions about the presence of civilians, and forecasting impacts on, for example, food production and access to health care.

Intelligence analysts and planners warned about the seemingly endless information needs that could theoretically flow from the broad notion of “the civilian environment”.Footnote ⁴⁷ Information overload can be paralyzing, and there is a risk that increased information will create a false perception of enhanced effectiveness. Critically, civilian environment analysis requirements should not be viewed in isolation but should be purposefully tailored to information requirements, enabling preparations for civilian harm mitigation in a specific operation. This is not to suggest that frameworks for understanding the civilian elements of an operating environment did not exist previously, but the pre-existing frameworksFootnote ⁴⁸ primarily supported broad contextual understanding of an area of operations while lacking explicit intent to mitigate civilian harm in planned operations. Civilian environment analysis is now absorbed into joint intelligence preparation of the operational environment (JIPOE)Footnote ⁴⁹ doctrine, which anchors intelligence production to the military planning process. The CHMR-AP created dedicated Civilian Environment Teams in recognition that such analysis could only be undertaken systematically and consistently if explicitly tasked and resourced.Footnote ⁵⁰

Courses of action and harm mitigation

The Joint Planning Process entails a continuous and iterative relationship between intelligence, planning, joint firesFootnote ⁵¹ and the targeting process. Intelligence informs commander’s intent, desired end states, and mission objectives. These serve as a basis for courses of action (COAs) laying out a sequence of steps to achieve mission objectives. Joint fires translate these into kinetic and non-kinetic actions to shape the battlespace and targeting in order to degrade and defeat adversary forces. These planning and operational processes also define specific information requirements shaping intelligence priorities and commander risk management.

Specific legal obligations to limit civilian harm, including precautions to spare civilians and civilian objects, are linked to individual attack decisions. Tactical tools and guidance are essential for this purpose; however, operational level design and warfighting processes have a considerable influence on the overall impact of operations on civilians.Footnote ⁵² When decision-makers are equipped with civilian environment analysis, civilian harm considerations can be woven into planning earlier and more robustly than they typically are. The development of COAs is where civilian harm mitigation starts to become tangible:

Civilian harm risk mitigation through COA development, evaluation and approval is now embedded in doctrine,Footnote ⁵⁴ whereby COAs are expected to specify potential impact on the civilian environment and establish relevant civilian harm risk criteria for comparing and evaluating COAs. COA analysis serves as an opportunity to develop options to mitigate harm, determine the required resources, identify potential decision points for mitigating risk, and refine information requirements in support of risk mitigation prior to commander approval. While some guidance and toolkits are available to help devise civilian harm mitigation measures,Footnote ⁵⁵ effective mitigation is highly contextual and such guidance is only likely to be utilized if the planning process requires it. As one military official has noted, “If you’re serious about reducing CIVCAS [civilian casualties], you can design the operation around where your humans are.”Footnote ⁵⁶ This is arguably exactly what the law requires,Footnote ⁵⁷ and what CHMR capabilities and procedures enable.

Concerted effort to develop precautionary measures specific to the planned action before they are needed, rather than leaving this primarily to the target cycle prior to strike decisions, allows time for an expanded range of options to be developed and considered. Command staff and leadership habituated to civilian harm risk analysis and mitigation during planning will not only have a menu of “pre-baked” options available but will also be better able to take civilian harm risk into account during compressed decision processes in high-tempo operational command environments. The COA development, analysis and approval process also creates a record of how civilian environment information was applied and how it shaped the COA that was ultimately executed. This strengthens the basis for after-action review and evaluation of civilian impacts, offering insight for operational and institutional learning.

While information needs and decision points addressing civilian harm should be embedded into, rather than run parallel to, existing command processes, there is still value in mapping out the specifics of civilian harm risk information flow supporting COA development, evaluation and approval. A cross-functional and multi-service tabletop exercise at US Indo-Pacific Command sought to do this, using fictional but realistic scenarios whereby civilian harm incidents were used to identify what factors needed to be considered, what specific information would be needed, who would need it, who would generate it, how they would generate it and by when, to better avoid or limit the harm.Footnote ⁵⁸ This allowed for a deeper understanding of the unique roles being performed by individual staff and where the constraints on information flow needed to be addressed.

The target cycle, red teaming and dynamic strikes

With respect to the target cycle, CHMR policy reiterates IHL obligations – for example, regarding the implementation of precautions – and underscores existing good practice, such as the maintenance of no-strike lists.Footnote ⁵⁹ It also emphasizes information-sharing on civilian harm risk “between all elements, nodes, and cells responsible for the joint targeting process”.Footnote ⁶⁰ Notably, while collateral damage estimation remains relevant as an existing rigorous and robust methodology, it is narrowly concerned with the direct incidental effects of an attack within a confined radius around an intended military target. CDE does not incorporate indirect and reverberating effects,Footnote ⁶¹ for example, which may arise from damage to civilian objects or the contamination of natural resources or urban habitats. Rather than applying changes to the CDE methodology itself, CHMR policy enriches the information environment shaping and informing joint fires and the processes leading up to and following strike decisions, thereby complementing CDE as a specific tool.

While planned or deliberate strikes involving static military targets are typically vetted, scheduled and approved in advance, dynamic strikes are unscheduled, unanticipated, and involve targets of opportunity addressing an emerging threat in time-pressured circumstances.Footnote ⁶² These include strikes undertaken in defence of one’s own forces or those of a partner force, where the urgency of the threat means less time to marshal intelligence, verify targets, apply precautions, undertake tactical manoeuvres and select suitable munitions – steps which would help to minimize risk if time allowed. Unsurprisingly, the preponderance of civilian harm results from dynamic engagements, particularly in urban settings, rather than planned strikes.Footnote ⁶³ This illustrates

This tension speaks to the value of robust COA development for anticipating tactical conditions and constraints during execution. For example, COA development can help to anticipate the need for a diversity of air, ground and intelligence assets to support a tactical engagement. Forecasted scenarios may help anticipate measures to mitigate heightened risk in situations where targeting decisions are delegated to lower echelons. Anticipation of risk along these lines may help to avoid situations whereby troops-in-contact are left with no good options but to take actions resulting in civilian harm. Such situations may not be avoided entirely, but preparing for them – including through pre-deployment training based on COA risk analysis – may help to limit risk.

The tensions created by dynamic tactical engagements also point to the benefits of dedicated civilian harm-focused red teaming capabilities, augmenting legal advice from staff judge advocates, in a Joint Operations Cell or co-located with a designated target engagement authority. Red team personnel can aid in recognizing a path dependency built on unverified intelligence, raise civilian harm implications associated with certain munitions, help avoid worst-case scenarios, and inject alternatives in a manner that keeps pace with operational tempo and command decision-making. For example, if a commander is weighing a potential strike on a civilian structure or infrastructure being utilized by adversary forces, a red team perspective could interrogate assumptions about the reliability of surveillance and the effective contribution that the object makes towards the adversary’s military actions or propose measures to limit the civilian impact. In a high-stakes, high-intensity combat operation, and a military culture with a bias for decisive use of force to achieve its targeted results, red teaming offers commanders an added means of managing operational tempo, and exercising strategic patience and tactical restraint, in order to ensure that capabilities are in place and conditions are appropriate for effective civilian harm mitigation before taking action, thereby strengthening the basis on which attack decisions are made in compliance with IHL.

Civilian harm assessments

The introduction of dedicated capabilities for civilian harm assessments is one of the most significant elements of CHMR and is worth examining closely. While they are undertaken after an attack, civilian harm assessments stand to inform the entirety of a mission as well as future operations, policy and capabilities development. Standard battle damage assessments,Footnote ⁶⁵ even when done consistently and well, do not offer sufficient information on the impact of strikes on civilians and civilian objects, being primarily concerned with whether a weapons system performed as expected and a strike has had the intended effect on the military target. The higher the operational tempo, the greater the pressure to move quickly, thereby degrading the quality and consistency of assessments. Information on civilian impacts is gathered only incidentally: it is recorded when time allows and when civilian casualties or damage to civilian objects is apparent while gathering information, primarily visual, on the impact of a strike on a military target.

Without an understanding of the impacts of operations on civilians, commanders are left with a considerable blind spot in their understanding of the operating environment and are unable to detect when repeated errors are leading to civilian harm. The public narrative about military operations is very much part of a commander’s theory of victory, whereby countering mis- and disinformation is central to shaping the information environment.Footnote ⁶⁶ The commander cannot do so effectively, however, if operational data is insufficient to distinguish between real and false reports of civilian harm. This can lead to a mounting mismatch between a command’s internal and public-facing narrative and the observable reality as experienced and documented by civilians. A military command that is unaware of its impact on civilians also leaves victims unacknowledged and with no place to turn to understand who was responsible or to learn what, if any, assistance they may be able to access in order to alleviate the harm suffered.Footnote ⁶⁷

The CHMR policy framework drew from good practice emerging in Afghanistan and Iraq operations to standardize assessment procedures and create a new operational capability in the form of Civilian Harm Assessment Cells (CHACs).Footnote ⁶⁸ Recognizing that it may not be feasible to conduct assessments in response to all civilian harm reports, CHMR policy requires combatant commanders to establish guidance and criteria for conducting assessments “at the most detailed scale practicable given mission requirements, the availability of resources, and other operational factors”.Footnote ⁶⁹ This guidance must include criteria for assessing damage to civilian objects based, for example, on their criticality. As with all military capabilities, CHACs must be established, with procedures and manning in place, well prior to the onset of hostilities. Doing so allows time to rehearse their functions in exercises so that command leadership can experience how assessments inform decision-making and staff can refine procedures.

Civilian harm assessments are primarily a fact-finding procedure. They seek not to characterize the action or the harm in terms of their lawfulness but, rather, to understand what happened and how, and whether one’s own operations were responsible for civilian harm. Recognizing the challenge of imperfect information in an ongoing operation, particularly in non-permissive environments, CHAC procedure entails applying a “more likely than not” standard,Footnote ⁷⁰ reflecting the command’s best understanding based on a preponderance of evidence, to arrive at informed conclusions about whether to attribute specific incidents of civilian harm to US action. CHAC procedure also allows for confirmed civilian casualties to be presented as a banded estimate if precise figures cannot yet be determined. Civilian harm assessments remain open until a determination of responsibility can be made, thereby allowing new information to be integrated when it becomes available. CHACs are also responsible for identifying factors contributing to harm, using aggregated data to analyze broader patterns, and making recommendations to address those patterns.

Civilian harm assessments are not a substitute for formal investigations to determine the lawfulness of certain strikes or the conduct of forces under their command, but constitute an added capability. Although not intended to characterize lawfulness, assessments help ensure that critical records are preserved and may help commanders to identify early whether there is a risk of violating IHL, thereby addressing that risk sooner than they would if such insights were pending a formal investigation. Formal investigations to determine whether wrongful acts have been committed are often seen as cumbersome, time-consuming and a distraction from mission objectives. The pressure to complete them rapidly can lead to incomplete review, hasty conclusions and compromised credibility of their findings,Footnote ⁷¹ ultimately leaving commanders, operators and victims without answers. The added capability of CHACs thus meets operational information needs while allowing investigations to draw on assessment information in order to answer different questions about IHL compliance.

The combined effect of this approach greatly enriches the command information ecosystem and enables other critical functions. In addition to enabling operational commands to detect civilian harm when it occurs and to avoid recurring patterns, aggregated data across operations also supports institutional learning. Civilian harm assessments encompassing damage to or destruction of civilian objects stand to offer considerable insight into proportionality assessments and good practice with regard to exercising precaution and limiting impacts on civilian objects. Assessments provide a basis for acknowledging, and expressing condolences for, harm when it happens – whether communicated publicly or privately to affected people – consistent with an ethos of treating civilians in armed conflict with dignity and respect. Finally, the CHMR policy framework requires that civilian harm data be publicly reported on the relevant command’s website and in an annual report to Congress, thus enhancing transparency and informing the public discourse on the humanitarian impacts of armed conflict. A consistent and robust civilian harm assessment capability enables commanders to communicate forthrightly and transparently – including with affected people and communities – about civilian harm and the steps being taken to minimize it.

CHACs serve an indispensable strategic purpose in a command which proactively seeks to comply with legal obligations and take constant care to limit civilian harm as part of its strategy to achieve mission objectives. Many commanders are acutely aware of the impact that mass casualty events and recurring patterns of harm can have on the perceived legitimacy of military operations,Footnote ⁷² the confidence of coalition members in the integrity of their shared purpose and, ultimately, mission success.

A bias towards proactive risk management puts commanders and command staff into an information-seeking posture. In this regard, one of the more significant influences of CHMR policy is the value it places on external information sources. Militaries are accustomed to believing in their superior situational awareness, but this does not extend to understanding the impact of conflict on civilians. From information on the interconnected character of urban services and systems to dynamic population movements and how civilians are injured and killed, external non-military sources of information are indispensable.Footnote ⁷³ A 2018 study directed by the Chairman of the Joint Chiefs of Staff compared DoD data with that of Airwars, a civil society organization which tracks and analyzes civilian harm, to understand the reasons for disparity between datasets. Among its findings, the study noted that of 191 credibly documented incidents related to Operation Inherent Resolve between 2015 and 2017, 58% of all reports of civilians confirmed killed or wounded originated from external sources.Footnote ⁷⁴ Without external information sources, operational commands may find that they are truly operating in the blind.

Militaries that work proactively to limit risk to civilians will utilize a diversity of internal and external information sources, ranging from witness interviews, self-reports on social media, and local and national media coverage to specialized organizations analyzing open-source information. This may entail an important shift in mindset and practices, including a deliberate effort to draw lessons from civil–military coordination with a view to sustaining a dialogue with specific external actors.Footnote ⁷⁵

Connecting expected performance to IHL

The preceding sections describe what CHMR policy set out to do and how its elements were intended to work together – but it is equally important to be candid about what the framework did not do. Most critically, CHMR policy did not establish a framework against which to assess effective performance. The lack of such a framework had several related implications.

The CHMR-AP set out objectives and tasks, and DoD-I 3000.17 assigned roles and responsibilities across DoD headquarters, services and combatant commands. However, neither described how a command with mature CHMR capabilities should perform, and no such framework was developed during implementation. In the absence of such a framework, implementation tended to be monitored as a task-based checklist recording, for example, whether a cell had been established, a position had been filled or a procedure had been written. CHMR was too often viewed as a series of administrative tasks rather than in terms of institutional and operational capabilities.

Without a framework of performance measures, staff and leadership lacked a shared understanding of what was expected of them, of how the tasks and responsibilities assigned were expected to lead to changes in performance, and of how progress towards mature CHMR capabilities would be demonstrated and assessed. Combatant commands, services, and the DoD as a whole also had little basis from which to sequence capabilities development activities, prioritize among competing tasks, tailor certain capabilities to regional theatres, align resources beyond the initial staffing established by the CHMR-AP or evaluate over time which elements were producing the greatest value for command decision-making.

Further undermining the development of concepts for mature CHMR performance, CHMR policy does not spell out how it contributes to meeting IHL obligations. CHMR policy is appropriately framed in the language of risk reduction, but while the policy affirms its consistency with IHL, the relevant obligations themselves feature largely as background. Neither the CHMR-AP nor DoD-I 3000.17 explains how the capabilities they establish serve specific IHL obligations pertaining to the conduct of hostilities. Instead, the relevance of CHMR to IHL obligations must be interpreted and pieced together, as is done at several points in this article.

The opportunities to draw a direct link are plentiful. For example, civilian environment analysis establishes baseline information which helps anticipate foreseeable direct and indirect harm and supports doing everything feasible to verify the nature of a target prior to attack. Integrating civilian harm risk projections into planning allows COA development to serve as the starting point for tailored precautions aimed at mitigating foreseeable harm in the choice of means and methods. Civilian harm assessments enable decision-makers to understand whether everything feasible is being done to minimize harm and to exercise constant care. Patterns identified across assessments, in turn, stand to inform what a command can reasonably be expected to foresee in subsequent strikes and ongoing operations and thereby enhance precautionary measures aimed at sparing civilian lives.

For example, assessments could reveal that civilians routinely shelter in lower floors and basements of buildings undetectable by overhead intelligence, surveillance and reconnaissance (ISR) or that strikes in the vicinity of power and water nodes are resulting in damage leading to a breakdown in essential services such as medical care. During US-led coalition operations against ISIS between September 2014 and January 2018, secondary explosions, often reaching far beyond the projected blast radius, accounted for nearly a third of the civilian casualties acknowledged in the US military’s own assessments.Footnote ⁷⁶ In this context, had a consistent practice of post-strike civilian harm assessments been in place, the repeating pattern could have been detected and a range of precautionary measures could have been applied: tactical guidance could have been adjusted, the risk could have been flagged in tactical planning and target selection, a larger effect radius could be applied when estimating potential harm from a strike, additional intelligence could be required prior to strike authorization, weapons selection and deployment could be adjusted or an alternative means of achieving the desired objective could be pursued.

Furthermore, the way CHMR policy defines civilian harm leaves important aspects of the protection IHL affords civilians at the margins. DoD-I 3000.17 describes civilian harm for the purpose of the policy as

This description has two consequences. First, it places the effects of hostilities on the services and systems on which civilian life depends in a secondary tier to be considered as a matter of policy. Yet reasonably foreseeable indirect harm, such as the loss of water, electricity or health services that can follow damage to critical infrastructure, must be taken into account in proportionality assessments and addressed through feasible precautions in attack, as the United States itself recognizes.Footnote ⁷⁸ In addition, an object that has become a military objective may continue to serve civilian functions, and the foreseeable harm arising from the loss of those functions must be taken into account.Footnote ⁷⁹ By presenting “other adverse effects” as a discretionary policy consideration, the policy may signal to personnel that anticipating and mitigating indirect harm is optional rather than required by law.

Second, casualties and physical damage to objects do not capture all the ways in which IHL protects civilians. IHL prohibits attacking, destroying, removing or rendering useless objects indispensable to the survival of the civilian population, such as foodstuffs, crops, livestock and drinking water installations.Footnote ⁸⁰ It prohibits acts or threats of violence whose primary purpose is to spread terror among the civilian populationFootnote ⁸¹ – a rule relevant, for example, to information operations. And it requires that medical personnel, units and transports be respected and protected.Footnote ⁸² Harm resulting from a failure to observe these rules may fall largely within the “other adverse effects” that CHMR policy treats as discretionary rather than legal obligations.

The relationship between policy and law further matters for how a military force institutionalizes its interpretation of its legal obligations and equips its personnel to understand and operationalize those obligations in the conduct of hostilities. Because CHMR is framed almost entirely in policy terms, with little explicit link made to legal obligations, personnel may believe that it goes “beyond what IHL requires” and that all of its components are discretionary – even those that in fact help give effect to legal obligations. However, there is a strong argument that the reverse is true. The implementation of feasible precautions depends on the information and capabilities available at the time to those who plan and decide on an attack. Once the factors contributing to civilian harm are known, and capabilities exist to better anticipate and avoid – or to at least minimize – that harm, these become part of the legal obligation to undertake feasible precautions. The information and capabilities must be used, and it is difficult to see how their employment in planning and conducting attacks could be viewed as merely optional.Footnote ⁸³

In sum, CHMR policy builds considerable capabilities to make harm more foreseeable and generates options to mitigate it – thereby strengthening the ability of military forces to meet their obligations – while neglecting the relevant legal standards in a policy explicitly designed to minimize harm.Footnote ⁸⁴ By connecting CHMR policy objectives and measures of performance to IHL obligations in the conduct of hostilities, these basic flaws could have been avoided.

A CHMR capability maturity framework

Measures of performance would also enable and inform the development, as specified in DoD-I 3000.17, of “professional tracks, including skill identifiers and certification requirements for key CHMR personnel and functions”.Footnote ⁸⁵ A performance measurement framework would have allowed the creation of specialized career tracks early in implementation. Without a career path with professional skill indicators matched to a performance framework, CHMR functions will inevitably compete with other career tracks offering clearer paths to advancement, leading to frequent staff rotations. In effect, this replicates one of the challenges underpinning a failure to systematize good practice which CHMR policy sought to address – namely, losing proficiency and institutional knowledge with rotations in personnel. The development of career tracks would also have signalled that CHMR capabilities entail competencies that the institution values and is committed to developing in a sustained and purposeful manner.

Measures of performance would additionally aid in defining staffing requirements, including for dedicated CHMR functions at combatant commands and operational commands. Dedicated and specialized CHMR staffing arrangements should ensure that specific planning and operational procedures, such as civilian environment analysis in intelligence processes and specific procedures for post-strike civilian harm assessments, can be performed consistently. However, effective civilian harm mitigation depends as much on existing capabilities and functions as it does on the new staff and augmented procedures. A dedicated CHMR workforce must be sizable enough to sustain specific functions but not so large that it would encourage other staff to offload their related responsibilities to those with CHMR job titles – this would amount to CHMR operating as a parallel process rather than being integrated into regular staff functions and battle rhythm. Commands assessing staffing requirements with this in mind would benefit from indicators and benchmarks of expected performance and of a joint force with mature CHMR capabilities.

These critiques do not diminish the value of the concepts and operational insights codified by CHMR policy, but they may offer lessons for future development of CHMR concepts and capabilities. A performance framework need not be elaborate; it might be described as a “CHMR capability maturity framework” and, in relation to combatant command and operational command functions, could establish a shared account of what a command should be able to do, and how well, at successive stages of development. The language of capability maturity is common to many militaries and therefore offers a basis for exchange between States building comparable capabilities. A description of the capabilities, information flows and decision points that a command should have in place prior to hostilities, together with a small number of indicators tested against the command’s battle rhythm, would give leadership and staff a clearer understanding of what they were expected to work towards. The indicators would need to be anchored in IHL, describing the conditions under which, for example, a command can be said to be doing everything feasible to verify the nature of a target, to anticipate and mitigate foreseeable direct and indirect harm, and to exercise constant care.

The US joint force already has a process of this kind. Before a headquarters assumes Joint Task Force responsibilities,Footnote ⁸⁶ it is trained and assessed against mission-essential tasks drawn from the Universal Joint Task List,Footnote ⁸⁷ each expressed as a task with stated conditions and a standard of performance. To be certified, a command defines what adequate performance means, trains its personnel to that standard and adjusts its procedures accordingly prior to hostilities. Building CHMR tasks, conditions and standards into mission-essential task lists and into certification exercises would ensure that performance assessments relate directly to operational conditions. Embedding CHMR in this process would also help direct resources and initiatives within larger processes, building the capacity to continuously learn and adapt and to develop new guidance and capabilities supporting planning and operations.

Tailoring and scaling CHMR

As with any military capability, applying CHMR to best effect depends on the type of operation being planned and executed. In particular, the tempo, scale, and domains of operation have a significant bearing on the risk that operations may pose to civilians and how that risk may be managed. Purposeful tailoring of approaches to civilian environment analysis, planning, red teaming and civilian harm assessments is essential. In a low-intensity conflict with limited mission objectives, the overall pace of operations may allow time to thoroughly understand the civilian environment, develop and adjust COAs and targeting, train on harm mitigation measures, refine operational and tactical integration across the relevant domains, and forge good practice with partnered forces. In a potential large-scale combat operation (LSCO) against a peer adversary, where the fight encompasses a broad geographical area and the stakes are perceived as existential, it is logical to expect that systemic blind spots contributing to civilian harm will be amplified by the scale and complexity of the operations. The expected volume and tempo of kinetic and non-kinetic effects greatly increase the inherent civilian harm risk, particularly in densely populated urban centres.Footnote ⁸⁸

Staff exercises can facilitate grappling with the complexities of the civilian environment in preparation for a high-intensity LSCO. For example, how can a dynamic civilian environment common operating picture be maintained once hostilities have commenced? Is there a hierarchy of civilian objects which must be protected, and on what basis are objects prioritized? What information is needed to understand continuous civilian functions of a military objective, and how can the effects of potential damage to an object on which civilian life depends be forecasted?

Artificial intelligence (AI) may help civilian environment analysis keep pace with operational tempo, for example by identifying changes in civilian structures or infrastructure, detecting unanticipated civilian presence or forecasting effects across a range of structures and essential services.Footnote ⁸⁹ Importantly, this also requires an understanding of the risks to civilians associated with AI replacing the role of human critical thinking and good-faith judgement in efforts to comply with IHL and minimize civilian harm in the use of lethal force.Footnote ⁹⁰ Two risks warrant particular attention. The first is that weaknesses and blind spots already present in a military’s own practice as discussed above – such as flawed assumptions about the presence of civilians, overconfidence in the reliability of intelligence, and reliance on familiar shortcuts – are replicated in systems designed on the basis of legacy practices. Compounding this risk, automation bias leads to a tendency of operators to defer to machine outputs without adequate scrutiny. The second risk is that the speed at which such systems operate may overwhelm the procedures through which judgement is exercised, so that authorization becomes procedural rather than the product of informed assessment.Footnote ⁹¹ Purposeful design of AI applications with the objective of anticipating and mitigating civilian harm as part of decision support functions – consistent with IHL and informed by known factors contributing to civilian harm – may help to counteract the inherent risks and even heighten the feasibility of taking precautions in military operations.Footnote ⁹² A CHMR capability maturity framework would aid in establishing the necessary conditions, observable indicators, and an auditable record for the use of AI-enabled decision support.Footnote ⁹³

In addition, potential future great-power LSCOs are expected to be all-domain, encompassing sophisticated and technologically advanced kinetic and non-kinetic capabilities in land, air, maritime, cyber and space operations. Increasingly, these capabilities may include large-scale use of uncrewed and autonomous weapons systems which greatly expand the reach of military operations while accelerating their tempo and compressing decision time. Operational commands may be decentralized, operate with substantial autonomy from headquarters and be expected to rapidly respond to adversary action. Conflict parties must also anticipate intermittent or sustained degradation or loss of ISR and communications capabilities throughout operations.Footnote ⁹⁴ The consequences of this for civilian harm mitigation warrant attention. Where forces do not sustain air supremacy and cannot rely on a permissive ground environment, the information on which civilian harm assessment depends may be obtainable only through air, space and cyber capabilities, and even that access may be intermittently or persistently disrupted. Ground forces which would under other circumstances develop relationships with local information sources and undertake ground-level information-gathering may not be present. Without diversified information sources and channels, forces will lack the necessary information on the civilian environment to manage civilian harm risk.Footnote ⁹⁵ Taking these factors together, an armed conflict of this character not only poses tremendous risk to civilians but also greatly complicates risk management.

A commander’s risk management process should thoroughly establish the scope of projected risk to civilians and determine how best to focus CHMR procedures and capabilities within planning and operations. For example, prior to hostilities, front-loading civilian environment analysis and a diversity of COAs suitable for a range of potential operational and tactical contingencies, across all domains, may help to ensure that operational commands have a range of options available to them to limit civilian harm in situations that arise. While it may not be possible to conduct civilian harm assessments on all strikes or non-kinetic attacks, commanders can direct assessments covering a series of strikes, or within a certain area or time frame, to ensure that relevant operational insights are steadily generated. Civilian environment analysis, mitigation measures and civilian harm assessments will require tailoring to the maritime, cyber and space domains.Footnote ⁹⁶ Building specific mechanisms and workflows to ensure that civilian environment information is rapidly updated in tandem with other battlespace information will be essential to sustaining a dynamic common operating picture. These mechanisms will need to be designed to enable appropriate all-domain information flow, including accounting for the compartmentalized character of cyber and space operations.

Operational commanders and target engagement authorities at all levels across domains will need to have a clear understanding of risk management expectations as well as what capabilities are available to them to manage risk. A shared understanding of civilian harm risk and the co-creation of risk management measures are especially important for periods of time when command communications may be degraded. All forces will need to understand the commander’s risk tolerance and expectations for civilian harm mitigation measures when they lose ISR and lack the battlespace visibility to which they are accustomed. In conjunction with the decentralization of command authorities, combatant commanders should consider which specific CHMR capabilities should be replicated at lower-level commands. This may, for example, include assigning red teams to the Joint Task Force level and below to support mitigation in COAs throughout operations. Civilian harm assessments may need to be decentralized, whereby dedicated cells at lower echelons proactively conduct rapid fact-finding and channel it upwards for a combatant command CHAC to complete the assessment process, conduct analysis and channel recommendations.

Adding to this complexity are coalition and partnered operations where allies and partners may bring to the table significant variation in their interpretation of applicable law; doctrine, guidance and procedure; restraint in the use of force; and risk tolerance, including with regard to civilian harm.Footnote ⁹⁷ Cognitive bias may operate differently and take different forms across decision-making cultures. Coalition members may include military forces that do not have recent combat experience and therefore lack recent operational lessons informing civilian harm mitigation capabilities and procedures. National differences can be challenging to manage even in a standing alliance, such as the North Atlantic Treaty Organization, and will be even more so in a coalition operating together for the first time.Footnote ⁹⁸ Civilians will be at greatest risk where responsibility is diffused across a combined command and where national interests and objectives diverge.Footnote ⁹⁹

However, coalitions operating with a shared interest in meeting their legal obligations and limiting civilian harm have ample opportunities to use combined planning, and investments in coalition integration, to build a shared understanding of civilian harm risk and develop complementary capabilities, procedures and practices.Footnote ¹⁰⁰ Just as it is valuable to establish an overarching risk management framework and corresponding information flow to mitigate civilian harm within a single command structure, it is critical to design appropriate decision-making frameworks and information flow in a combined coalition operational command structure. These preparations must also consider the needs and perspectives of coalition members defending their own territory against an external threat and fighting to protect their own population. National civil defence planning stands to serve as an important resource for all coalition members to help them understand life-sustaining elements of the civilian environment and ensure that these are factored into combined planning.Footnote ¹⁰¹

Initiating an effort to strengthen policy and practice

The systemic challenges contributing to civilian harm are not unique to the United States. Every military force stands to benefit from understanding civilian harm risk and developing capabilities to adapt to new operational contexts, types of operations and emerging technology. Perhaps most essential is to adopt a proactive posture towards upstream and systematic investment in institutional capacity to anticipate, mitigate and respond to civilian harm, rather than being forced to contend with a deteriorating situation once operations have commenced.

Individual militaries will approach this differently based on their existing processes for distilling and applying operational lessons, building new procedures and capabilities, and planning for future operations and contingencies. Militaries that have begun this work have tended to use different entry points: the Netherlands Ministry of Defence, for example, has developed new and strengthened procedures for processing external reports of civilian harm, conducting civilian harm assessments and investigations, and reporting transparently on these processes, while the UK Ministry of Defence has used US CHMR policy as a point of reference to conduct a baseline review of its own policies and practices with a view to identifying opportunities to strengthen them.Footnote ¹⁰²

The entry point matters less than a sustained commitment to building on it. Not every military will have the time or resources to develop a comprehensive institutional policy all at once, and indeed, the approach to the development of new capacities does not need to be linear and begin with overarching policy – for example, a review of past after-action reports and operational lessons learned reports may yield valuable insights. Such a review could gather the experiences of a range of personnel from tactical, operational and strategic perspectives in order to reflect on risk factors and identify valuable good practice. Mapping horizontal and vertical information flow – specifically, how relevant information is generated, curated, analyzed, disseminated, and applied in decision-making – could serve as a useful starting point for identifying blind spots and opportunities to strengthen procedures within existing functions. This would also allow command leadership to reflect on and strengthen their risk management frameworks and practices. A series of steps along these lines could serve as a starting point for then developing and testing a CHMR capability maturity framework along the lines described above.

Large exercises involve complicated design where it can be challenging to blend in CHMR considerations and capabilities. Large exercises are also not likely to yield much insight before there is a shared understanding of civilian harm risk and how to mitigate it. Conversely, tabletop exercises (TTXs) offer a low-stakes setting for fostering analysis and dialogue based on realistic warfighting scenarios. TTXs are ideal for problem-solving and can be sequenced to build progressively toward, for example, standard procedure and capabilities requirements and to determine the distribution of critical functions across a command structure and military domains. This may enable the exploration of options and the definition of mature capabilities before major changes or additions are decided. TTXs could also be designed around development, analysis and evaluation of COAs within a larger operational planning process. This approach may be especially constructive and valuable when initiating dialogue with allies and partners on civilian harm mitigation and exploring how to ensure comparable, or even interoperable, practices.

While policy and practice must necessarily be developed at a national level, States will benefit from using their regular bilateral and multilateral security cooperation activities and fora to highlight and address challenges and opportunities relating to civilian harm mitigation. Promising initiatives include the CHMR International Contact Group, comprised of a small number of allied States co-chaired by the Netherlands and the United Kingdom, which convenes semi-annually to exchange lessons and encourage the development of good practice.

Finally, civil society organizations have honed information-gathering and analytical methods for understanding and addressing civilian harm for many years and have developed detailed insights on the critical factors contributing to civilian harm. Military lessons learned processes, exercises, and the formulation of good practice stand to be greatly enriched by soliciting the perspectives of a diversity of non-military actors, including academics and non-governmental subject matter experts.