Modelling the intelligence requirements and priorities process: the US response to the Rwandan genocide

ABSTRACT In the US, national security outcomes tend to be categorised as either the responsibility of the intelligence or policy community. Few discuss systemic outcomes emanating from the requirements and priorities (R&P) process, a top-level collaborative effort that determines national security objectives and establishes the means to address them. Here, a holistic model is introduced to examine the R&P process alongside the binary functions of intelligence and policy, and tested against two mandates of the US response to the Rwandan genocide: evacuation of American expats, and broader intervention. Such macroscopic investigations can better identify the root causes of national security outcomes.


Introduction
This research provides an overview of the components of the US Requirements and Priorities (R&P) process, and an exploration of their function during the escalation of urgent events.While the R&P process has undergone numerous iterations, specific attention here is paid to the process as it existed during the Clinton administration, through the aims set out by Presidential Decision Directive 35.A process-tracing model is applied to evaluate the effectiveness of the process when crises emerge.This model is then tested against one of the most significant foreign policy events of the Clinton administration, the Rwandan Genocide.Two aspects of the Rwandan Genocide are tested: the action to evacuate US nationals, and the US response to quelling the genocide itself.
For the purposes of this research, the phrase 'national security failure' broadly indicates a failure to protect the interests of a nation, its citizens and/or institutions, regardless of where responsibility lies.Since the Japanese attack on Pearl Harbor during World War II, investigations into national security failures in the US have tended to be categorized in binary terms: they are seen as the responsibility of either the intelligence or policy community. 1During the 1970s, these examinations began to skew heavily toward concentrating on intelligence community actions, while limiting investigation on the policy side. 2 Few post-mortem investigations addressed the potential for systemic failure.There is little acknowledgement of the idea that failure can occur as the result of the formalized, systemic mechanism that inextricably binds the two communities.
In the US, this mechanism is known as the requirements and priorities process (R&P).The process has been given various names over the course of the past 75 years (for example, Key Intelligence Questions, 3 National Intelligence Topics, 4 National Intelligence Priorities Framework 5 ), but the purpose and function more or less remain the same.Through this process, intelligence and policy leadership determine US national security objectives and establish the means to address them.While the process is integral to the intelligence and policy interface, it is seldom discussed, little understood, and overwhelmingly neglected in investigations of national security outcomes.This may be for two reasons.First, intelligence authors often focus on single agencies or departments and display 'almost an aversion to discussing interagency mechanisms and processes'. 6Second, the R&P process, particularly in the United States, is a massive undertaking that has been mired with problems since its inception.If examinations of single agencies or departments pose a daunting task, factoring in the R&P can seem insurmountable.
As a result, there is a tendency to zoom in from the macroscopic and observe smaller components.This can cause post-mortem investigations to misconstrue strategic failure as operational failure.For example Jervis, in his assessment of NFAC and warning failures around the Iranian Revolution, noted several strategic intelligence failures, including weak analysis in NFAC cables from the Tehran office and a reliance on information provided by SAVAK or other elites in the Shah's circle. 7While Jervis does at times hold both the intelligence and policy communities responsible for certain aspects of failure, he only briefly mentions the impact of the role of low prioritization of Iran, as existed in the Key Intelligence questions, which was the R&P process du jour.Yet the low prioritisation caused the consequence of failing to recognize that the Shah was in danger.
Failures within the R&P process often manifest in a way that looks like intelligence failure.As a result, recommendations for reform will sometimes target the symptoms, rather than the root causes of the failure.To demonstrate how this can occur, this research puts forth an explanation of the components of the requirements and priorities process and introduces a holistic model which incorporates the R&P into post-mortem examinations.The model demonstrates outcomes as a series of nested successes and/or failures, which allow investigators to pinpoint areas of weakness across both strategic and operational national security processes.

Understanding requirements and priorities
It is easiest to first consider the R&P process in terms of a business.In public and private sector organizations around the world, executives regularly meet with leadership from various departments to define their objectives and establish their level of priority.Once established, organization determine which people or teams will conduct the associated tasks and provide the resources and authorizations necessary to accomplish these goals.
In the US (and throughout most Western governments), the R&P process works in a similar manner.Consultations are conducted with policy officials to gather an inventory of national intelligence needs.The coordinating body in the intelligence community filters the list of needs and the remaining requests are sent to the lead IC official, who examines and adjusts the requests, and delivers recommendations to the National Security Council.The NSC will make final adjustments.Once approved, the requests are formalised into intelligence requirements, which are ranked, aligned with the most appropriate departments or agencies, and provided with a budget and resources.Intelligence requirements of the highest priority receive the greatest share of the budget.As the priority level decreases, so too does the associated budget.Finally, the agencies are provided with operational authorisations to conduct activities relevant to the requirement. 8ntelligence objectives sometimes require input from multiple agencies with specialized collection channels to draw from strategic, operational, or tactical concentrations. 9Therefore, the R&P process informs the guidance for interagency collaboration, including determining the lead and cooperating agencies for each priority. 10he R&P process also serves as an audit trail by introducing a system of controls to provide authorizations and restrictions regarding intelligence activities.Both action and expenditures are monitored to prevent operations from occurring without prior notification or approval.
In the US, the Executive branch, Congressional committees, and other organizations monitor intelligence community activities to ensure that 'reasonable and lawful means' are used to obtain reliable intelligence.Low-risk activities, such as the collection of open-source information, require a low level of authorization, whereas activities that carry a greater amount of risk, including those that could detrimentally impact foreign policy, require higher levels of authorization.For instance, the US Foreign Intelligence Surveillance Act of 2008 (FISA) requires that the intelligence community provide justification for the electronic surveillance of targets located outside of the United States.
As a result, the R&P process determines the intelligence priorities, and informs the allocation of resources, authorizations, and interagency or departmental cooperation to guide intelligence community actions.Together, these comprise the mandate for the intelligence community, and are therefore referred to here as mandate-level functions (Figure 1).
Generally, the intelligence priorities are established on a long-term basis, and tend to remain in place for a protracted timeframe.Occasionally, however, an unexpected event will arise in a low priority area.In these instances, the urgent issue may become a temporary, or ad-hoc priority.When this ad-hoc issue is escalated in priority, it necessitates an adjustment to the mandate-level functions.This requires a rapid realignment of resources, authorizations, and cooperation until the issue has been resolved.Ad-hoc priorities do not supersede other priorities but stand alongside them (Figure 2).For example, the COVID-19 pandemic caused several governments to rapidly escalate the issue and realign intelligence resources, cooperation, and collaboration.However, this does not mean that other issues of national security became less significant.
For the purposes of this research, the rapid escalation of an ad-hoc priority and subsequent adjustments to the mandate-level functions are referred to as a mandate shift.Urgent events that are assumed transient will not be added to the long-term, formalized list of intelligence priorities. 11If the  event is temporary, the incident is addressed, and the priority is de-escalated.However, if the situation poses a long-term threat or requires long-standing response (such as demonstrations in Syria evolving into civil war) the priority could be formally escalated in subsequent reviews of national requirements and priorities.However, even when formalized, this does not always mean it becomes a top priority.

A complex, fragile process
In theory, the R&P process is a streamlined, collaborative effort between intelligence and policy leadership.However, even when stakeholders are diligent, there is still the potential for something to go wrong.Globally, governments struggle with their versions of the requirements and priorities process in varying degrees. 12There are a series of pathologies that can impact the prioritisation process.
At the outset, the process requires regular consultation to gather intelligence requests from policymakers, which can result in a feast or famine of requests.If there is too little, or unclear policymaker input, middle managers in the intelligence community may be left to guess which items may be relevant.For example, Lowenthal notes that a former secretary of defence, when asked about giving more precise definitions of his needs said, 'No.I assumed they knew what I was working on'. 13n contrast, if the mechanism is filled with requests that are too specific or otherwise unrealistic, it can restrict the flexibility of intelligence collection and analysis and weaken the output.Further, too many intelligence requirements can cause errors in prioritization or result in a strain on the intelligence community's finite resources.Finally, if not maintained, the list of priorities could become flawed, overloaded, or outdated.
In terms of budget, the outcome of the priorities process also determines how several billions of dollars in programming will be spent on intelligence collection and analysis.Once assets are allocated to one task, they cannot be allocated to another. 14If the list of intelligence requirements expands, it is rarely balanced by an increase in resources or cuts to other requirements.Rather, existing resources are instead stretched to cover everything.This can cause high priority issues to receive fewer resources, 15 or leaves areas of lower priority with less funding, increasing the risk of collection gaps or other issues.
The allocation of resources can also become a source for competition within the intelligence community.Agencies may fight for the ability to work on high-priority, highly funded intelligence requirements -even if they are not best suited for the project. 16Competition can lead to the stovepiping of information, a redundancy of efforts, and a dearth of effective collaboration.
Finally, the R&P process also serves as an operational and financial audit trail, and provides authorizations or restrictions for activities conducted in the pursuit of information.Actions and expenditures are monitored to prevent operations from occurring without prior notification or approval by the lead policy officials.Low-risk activities, such as the collection of open-source information, require a low level of authorization.In contrast, activities with a greater amount of risk, including those that could detrimentally impact foreign relations, require higher levels of authorization.For instance, the US Foreign Intelligence Surveillance Act of 2008 (FISA) requires that the intelligence community provide justification to the Foreign Intelligence Surveillance Court for the electronic surveillance of targets located outside of the United States. 17However, the stifling of authorisation can hamper intelligence collection and analysis, and the authorization process has been targeted as an overly bureaucratic anchor that slows the progress of meeting national security requirements.
Finally, it should be noted that policymakers are often driven by factors that exist beyond the R&P process, and beyond their relationship with the intelligence community.They may have different perceptions or goals, and may disagree with what the intelligence community (or even other policymakers) have agreed is worthy of prioritisation.For example, there has been a growing divide amongst Republicans regarding continued support to Ukraine in their war against Russia (NBC News, 29 September 2023).In these instances, even when early warning is provided, it might be disregarded in favour of the policymaker's existing priorities or views.
For the reasons noted, the R&P process has been particularly troublesome in the US.Over the course of decades, successive reports have lambasted the inefficiencies of the process.Senior leadership in the intelligence and policy communities have attempted numerous reforms to create a sustainable, enforceable, and effective mechanism.But these well-intentioned endeavours have either created problems in other areas or added layers of bureaucracy to an already encumbered system. 18As a result, the R&P process historically been clumsy and ineffective, causing both politicians and the intelligence community to treat it with indifference.This is an open secret, yet it is rarely, if ever, factored into examinations of national security failure in the US.
Weaknesses within the R&P process can result in delays to (or denial of) the means necessary to conduct intelligence activities.As mentioned, these constraints manifest as weaknesses in intelligence collection or analysis, and can generate a 'false positive', when in fact, the root cause of failure may have occurred prior to the provision of intelligence direction.

The problem with 'intelligence failure'
Michael Morell, former deputy director of the Central Intelligence Agency, famously said in an interview, 'There is a saying in the intelligence community: "There are no such things as policy failures, only intelligence failures".Every president that I served with did this, and I think every future president is going to do this' (CBS News, 1 October 2014).Morell's aphorism speaks to an irksome point: the cause for national security failure is often in the eye of the beholder.Generally, intelligence is considered successful if it provides timely, relevant, reliable, and actionable estimates to decision makers.If it does not meet these conditions, it has failed.
Robert Jervis further argues that intelligence fails if estimates fall short of what is expected of good intelligence. 19But this is highly subjective.From this perspective, it is effectively left to the intelligence customer to determine whether the estimates provided meet the requirement of good intelligence -and as they say, the customer is never wrong.Thus, a decision maker may consider a warning intelligence report unreliable because it does not meet an unspecified degree of certainty.Another may feel that an estimative intelligence piece has too much background information and is therefore irrelevant.In short, intelligence is subject to being measured against arbitrary constructs, with little consideration of political will or cooperation between the decision makers and the intelligence community.In this instance, failure will almost always be the fault of the intelligence community.And in the enduring drive to optimize national security intelligence, investigations that focus almost entirely on the intelligence community will almost exclusively find intelligence failures.This is not to say that intelligence failure never occurs -there are, of course, instances where the intelligence community gets it wrong.Kerr observes that failure can occur due to challenges related to warning, politicization, gaps of information, cognitive bias, and building expertise. 20These points are visible in Brand's exploration of the impact of intelligence on policymaking during the Prague Spring.Brand identified that early estimates 'failed to report certain pieces important intelligence and misunderstood others' 21 regarding Russian intentions for military engagement in Czechoslovakia.In this instance, information gaps and cognitive biases had an impact on the initial estimates.
However, it should be noted that the issue was a low priority at the time. 22As has been common in other cases, once priority was escalated, subsequent reporting from the intelligence community became increasingly accurate.Despite this, the Prague Spring can still be considered an intelligence failure because of the community's inability to convey the urgency of events.
Despite intelligence growing in accuracy, officials were still surprised by the Russian invasion of Czechoslovakia.Brand observed that policymakers needed 'a warning that was clear and definitive as to the threat and imminence of military action'. 23Simply, the inability to effectively convey warning to decision makers, in a manner that engenders a timely response, is an intelligence failure.

A holistic approach to investigation -the RPC model
Currently, there is no macroscopic (or even uniform) approach to examining the outcomes of national security events.Without appraisal of systemic functions, governments are likely to overlook the root causes of national security failure and ultimately provide (and spend a lot of money on) ineffective recommendations for reform.To this end, presented here is a macroscopic model for examining national security outcomes.The Requirements and Priorities centric (RPC) model illuminates and centralizes the components of the requirements and priorities process that guides national security strategy, while also stringently examining the core functions of intelligence and policy making.
This RPC model addresses a need to consider and improve accountability at the strategic, joint levels of cooperation.Over multiple, consistently applied investigations, the model can serve as an audit trail, highlighting repeated areas of weakness and identifying nested causes of failure.In turn, this allows for more effective recommendations for reform.
To test its efficacy, the RPC model is applied against two aspects of the US response to the 1994 Rwandan genocide: the mandate to evacuate American expats, and intervention to slow the genocide itself.The former has largely been considered a national security success, the latter a national security failure.By incorporating the RPC model, we can observe the US response from a macroscopic view, and better identify the functions that led to success or failure in both events.

Methodology
While most post-mortem investigations focus heavily on the operational inputs and outcomes, the Requirements and Priorities centric model incorporates actions taken at the strategic level.As visualized in Table 1, the RPC model examines the four mandate-level functions, and incorporates the actions taken solely by the intelligence and policy communities, referred to here as the 'binary model'.
In this regard, failure or success is not viewed in the binary context of intelligence or policy community success and failure, but as the result of a series of nested outcomes that impact the overall process.
Because most organizations operate with similar 'mandate-level functions' (albeit with various names), the model has transitive properties; with slight modifications to accommodate relevant functions, it can be used across a large swath of government operations.However, it must be noted that viewing outcomes in a holistic manner does not do away with the notion of intelligence or policy failure; it cannot.All failures are not joint failures; sometimes the responsibility for failure does rest squarely in one community.
The RPC model relies on theory testing process tracing. 24Process-tracing provides a method to study causal mechanisms and generate a hypothesis that links these causes with outcomes. 25In this case, the causal mechanisms are the components of the R&P process.Process tracing provides more systematic approach to qualitative evidence than those offered by historical methodologies often used in political science.This method can be used to organize evidence in a manner that captures historical narrative while making observations regarding a hypothesis.Finally, process tracing can be applied without a control group, allowing the method to be applied to historical cases where no control is possible.Historical evidence can be used to understand whether a causal mechanism can lead to a change in the outcome. 26rocess tracing has been used in political and social sciences for some time, 27 but it has not been significantly employed in the field of intelligence studies.The work that most closely approaches this is Robert Jervis' Why Intelligence Fails: Lessons from the Iranian Revolution and the Iraq War. 28However, Jervis could be seen to have employed process tracing in order to find a minimally sufficient explanation for the outcomes.Further, he did not interact with the intricacies of the R&P mechanism.Jervis' conclusions can be mined down further to identify whether there are other, perhaps more deeply rooted causal mechanisms which impacted the outcomes.
To demonstrate its intricacies, the RPC model, the model is applied to elucidate elements of intelligence success and failure associated with two aspects of the US response to the Rwandan genocide: the mandate to evacuate American expatriates, and intervention to slow the genocide itself.A critical examination of existing evidence will provide an understanding of where Rwanda stood in terms of US priorities prior to the genocide, and then provide observable manifestations in the requirements and priorities process as events unfold.If no failures are visible in the R&P process, then it must be assumed that failure is likely the result of poor collection or analysis within the intelligence community, or poor policy input or response.

Rwanda, 1994: is intelligence to blame?
On 6 April 1994, a private plane carrying the Presidents of Rwanda and Burundi was shot from Rwandan airspace, killing both leaders.The crash was the spark that lit a wildfire; immediate and widespread violence broke out as Hutu-led militias and their extremist allies embarked on ethnic genocide that became the fastest, most efficient extermination in the 20 th century.Over the course of 100 days, it is estimated that up to one million Tutsis and moderate Hutu were killed. 29nticipating a renewed civil war, the US government had one immediate mandate: evacuate the Americans.The US quickly triggered a response to extract US nationals, but there was no immediate or urgent response to the overall genocide.The 100 days of genocide in Rwanda have been viewed as failure on the part of the international community. 30The US government in particular was accused of failing to initiate timely intervention.This in turn led to accusations of failure on the part of the US intelligence community.In the first days of the crisis, the US was guided by internal concerns and mischaracterizations of the violence, as the following sections will demonstrate.
Six years after the genocide, professor Alan Kuperman argued that President Clinton could not have known that genocide was occurring until around 20 April, two weeks after it began.Kuperman identified five factors to support his argument: intelligence depicted the fighting as a civil war; the violence was reported to be slowing; early death counts were grossly underestimated; focus was almost exclusively on Rwanda's capital, Kigali; and no credible observers raised the prospect of genocide until the end of its second week (The Guardian, 31 March 2004).These factors, he argued, were the fault of the US intelligence community, for failing to commit in-country resources to 'a tiny state in a region of little strategic value'. 31uperman is not wholly inaccurate.For instance, in the immediate aftermath of the 6 April crash, the violence was depicted as a civil war, and one report on 8 April indicated a 'glimmer of hope' that the fighting had slowed. 32However, Kuperman failed to acknowledge one key factor: the intelligence community was operating within guidelines set the by the administration.They could not commit resources in Rwanda without the consent of the policymakers.Kuperman's argument, as is commonly the case, reinforces the idea that responsibility for national security failure is compartmentalized, and ignores the impact of the systemic processes that bind the intelligence and policy communities.
At the start of the violence, a single mandate was escalated: evacuate Americans from Rwanda.The joint intelligence and policy community functioned to those goals, but when the operation was completed, there was no mandate to address the crisis in Rwanda.The distinction in the results is striking, but unlike Kuperman's assertion, this does not equate to failure on the part of the intelligence community.Rather, it is the result of an economically driven national agenda led by the Clinton administration, a change in the overall security landscape, and a fear of public backlash following the now infamous 'Black Hawk Down' incidents in Somalia.

In the US: changes at the end of the Cold War
Throughout the second half of the 20 th century, the requirements for the intelligence community were relatively static, centred around monitoring Soviet activity during the Cold War.The Cold War ended in late December 1991, and on 20 January 1993, President Bill Clinton became the first president to enter office in a post-Cold War era.With the omnipresent Soviet spectre abated, the Clinton administration faced the challenge of continuing to redefine American power and identifying new priorities for post-Cold War foreign policy, as indicated by the predecessor Bush administration's vision for a 'new world order'. 33As policymakers voiced an array of potential priorities, the implementation of an effective process to determine priorities became increasingly vital.
A secondary effect of the end of the Cold War was a reconsideration of multilateral peacekeeping efforts. 34The Clinton administration considered these efforts a means to advance US interests -but it was important to limit spending here as well.The US had committed forces to overseas deployments in Africa, the Caribbean, Europe and the Middle East, and the multiple engagements were cutting into the anticipated peace dividend at the end of the Cold War.
Clinton's administration, which ran on an internal slogan of 'the economy, stupid!' (New York Times, 31 October 1992) sought to curb spending and grow the economy.Every decision was made with a fiscal strategy in mind, and domestic policy was the administration's top priority.In terms of national security, focus was concentrated on consolidating efforts, improving efficiency, and cutting costs.
Upon taking office, Clinton tasked a team to identify actions and collaborations that could create a more cost-effective, unified intelligence community. 35The report called for a more centralized, integrated intelligence community with cross-agency collaboration.In order to curb spending on United Nations engagements, in March 1993 a presidential review identified the conditions which should govern US policy on multilateral engagements. 36The findings recommended a series of strict guidelines regarding US voting, support for, and engagement in multilateral peace operations with the UN. 37Together, the findings of these two reviews informed the strategy for national security and multi-lateral peacekeeping initiatives, and shaped the development of intelligence mandates for the first decade after the Cold War.

Tracing the requirements and priorities process
Clinton's Director of Central Intelligence, R. James Woolsey, quickly sought to improve the requirements process for the post-Cold War era.His strategy for the requirements and priorities process, called the 'National Intelligence Needs Process', was completed in late 1993.While this process was not formalized until President Clinton signed Presidential Decision Directive 35 in 1995, 38 by early 1994 the process was already in place.Woolsey's appointed Issue Coordinators were meeting with high-level policymakers and identifying their intelligence needs.The aggregated needs were evaluated against current collection plans, and then examined for intelligence gaps or other shortfalls. 39he full details of the process remain classified; however, it is possible deduce how the process functioned at the time of the Rwandan crisis.Under Woolsey, the intelligence community adopted a tiered prioritization scheme, where every nation would be assessed and ranked.Tier 1 was reserved for enemy states, and Tier 4 issues were reserved for countries required 'virtually no effort'. 40Woolsey also included a Tier 0 for issues required crisis coverage on a short-term (ad-hoc) basis.Priorities were addressed with a two-pronged approached, where issues were defined as 'support to military operations' and 'support to policy operations'. 41Guidance also addressed support to law enforcement and counterintelligence. 42 In the context of these categories, priorities were ranked as follows: • Tier 0 -Crisis coverage required (short term, 3-12 months) • Tier 1a -Countries that are enemies/potential enemies • Tier 1b -transnational topics of highest priority • Tier 2 -Other countries of high priority • Tier 3 -Low priority countries commanding some effort • Tier 4 -Low priority countries requiring less coverage. 43er 1 was reserved for nations that could become belligerents against the United States, including Cuba, North Korea, Iran, Iraq, and Libya. 44Tier 1a was reserved for transnational threats, including global narcotics rings, terrorism, organized crime, and the threat of proliferation of weapons of mass destruction. 45With the focus on support to military operations, areas where US troops were deployed were considered short-term emergencies placed at Tier 0. The bulk of resources was given to items at Tiers 0, 1, and 1a, leaving fewer resources for lower items.
Rwanda, neither great ally nor great threat, was assessed as low priority.The nation was not an essential to US security or interests.It was not a major trading partner (trade between the nations was under $10 million in 1992 and 1993). 46As a landlocked country in central Africa, the region held no strategic value in terms of military bases or naval ports, and was distant from areas of vital interest to the US.The relationship between the nations was largely diplomatic; a 1996 report from the House Permanent Select Committee on Intelligence reveals that Rwanda received very little intelligence coverage.Assistant Secretary of State for African Affairs George Moose contextualized Rwanda in terms of security concerns for US allies in Africa, particularly Zaire, Uganda, and Tanzania.When asked where Rwanda stood in terms of priority, Moose considered Rwanda 'maybe' a third-tier issue. 47A 1996 report from the House Permanent Select Committee on Intelligence reveals that Rwanda received very little intelligence coverage, indicating that it may have ranked in the Tier 4 category. 48s war in Rwanda and neighbouring Burundi escalated in the early 90s, the US called for calm in the region, but concentrated on greater emergencies such as enforcing no-fly zones in Iraq, the Bosnian conflict (which would later evolve into another genocide), and the fallout from the military coup in Haiti.
Apart from diplomatic engagement, the administration was particularly reticent to engage in Rwanda because of the recent past.In 1992, the George H. W. Bush administration sent military forces on a humanitarian and peacekeeping mission to Somalia.By autumn of 1993, mission creep expanded the engagement into a plan to restore order in the war-torn nation.Under Clinton, a key mission for the forces was Operation Gothic Serpent -a strategy to capture warlord Mohamed Farrah Aidid and his lieutenants, who controlled Mogadishu.On 3 October, intelligence from Joint Special Operations Command had identified to location of Aidid and his lieutenants.The capture operation was expected to take under an hour, however Aidid's men were prepared for the invasion, and close combat ensued.Over the course of two days, the Battle of Mogadishu had left 84 US troops injured and 18 killed. 49o the American public, Somalia did not take centre stage until video emerged showing the body of an American soldier being dragged through the streets.The video was met with public outcry and depleted political will within Congress.Clinton announced that US troops would be withdrawn from Somalia within six months.The final troops left the theatre in March 1994, two weeks before the Rwandan genocide began.
On the heels of Somalia, the emerging events in Rwanda were out of bounds.The government became disinclined to get actively involved in foreign interventions, particularly in Africa.Richard Clarke, Chairman of the Counterterrorism Security Group, feared that an attempt to engage in Rwanda would generate even more backlash.To this end, assisted in the development of a Presidential Directive to reform multilateral peace operations.The directive would close any gaps left open in the national security strategy, and set rigorous standards for US approval of, and involvement in, UN efforts.
Backlash from Somalia was not the only consideration factored into the development of the new directive.In keeping economic concerns at the fore, the administration was acutely aware that the US footed nearly one third of UN peacekeeping costs.Clinton had voiced concerns about management inefficiencies and aimed to reduce US funding to 25 per cent. 50Thus, before any support was given to UN security operations, the US government assigned 16 considerations to determine whether to support peacekeeping activities.Seven of these guided US voting on operations that did not involve American troops; six more factors were added if the US was to deploy troops, and a further three if the troops would be engaged in combat. 51Presidential Decision Directive 25 was not formalized until May 1994, the doctrine it contained was already set in motion.The directive set strict guidelines for approving peacekeeping operations, limited US involvement, cut spending on UN peacekeeping, and required Congressional and public support for future operations. 52hus, when the plane crash killed the presidents of Rwanda and Burundi, the question of engagement was moot.Only weeks after the failed intervention in Somalia and the return of the last remaining troops, it was highly unlikely that Congress or the American public would look favourably upon a new deployment.The focus instead was to identify any Americans at risk, and escort them to safety.

The mandate shift: evacuate the Americans
Evidence of a mandate shift can be seen in reports from the US intelligence community and statements made by President Clinton.Shortly after the plane crash on 6 April, Prudence Bushnell, acting Assistant Secretary of State for African Affairs, sent an email reporting the deaths of the presidents of Rwanda and Burundi to US Secretary of State Warren Christopher.The memo was written before official announcements of the crash. 53At the time, US embassies in Kigali and Bujumbura reported relative calm, with a noted 'increase in sporadic gunfire and grenade explosions' in Kigali'. 54Bushnell concluded that Americans were believed safe, but warned that if the crash was deliberate, it could result in widespread violence.
Over the next 12 hours, further reports contextualized the situation.The National Intelligence Daily, a report sent out by the US intelligence community, featured an update on the crash, warning that Hutus may seek revenge on the Tutsis, and civil war could resume and spill into neighbouring Burundi. 55The Bureau of Intelligence and Research assessed that 'rogue elements of the Hutu military, possibly the elite presidential guard', were responsible for shooting down the plane. 56urther, they may have been responsible for the execution of Prime Minister Agathe Uwilingiyimana and several other Rwandan cabinet officials. 57By 1300 on 7 April, a report from the US defence attaché in Kigali warned that the presidential guard in Kigali was 'out of control' in the streets. 58ithin 22 hours of the crash, the State Department mobilized a 24-hour Task Force to monitor the situation and coordinate the US response. 59Under the guidelines in place, the rapid creation of this working group indicates escalation from higher US authorities.The priority was the immediate evacuation of American nationals.By the nightfall, a plan emerged to transport citizens overland into safety.The Kigali Embassy reported that a group of 24 American expats were organizing a convoy to travel to neighbouring Burundi. 60 report by HPSCI also described Rwanda as 'a country that had little, if any, intelligence coverage suddenly becoming a top tier priority'. 61The rapid escalation of priority, and the resultant mobilization of intelligence and policy, functioned smoothly and rapidly.

Release of resources
Prior to the genocide, intelligence spending in early 1994 was stratified mainly across the top tiers of priorities and focused heavily on support to military operations.Because of this weighting, considerably fewer resources were in place for nations or issues at lower tiers.Further, because of the failure in Somalia, a US official recalled, 'Anytime you mentioned peacekeeping in Africa, the crucifixes and garlic would come up on every door'. 62he US embassy in Kigali was small.There were no CIA representatives, no defence attaché, or political officers.Present in the region were Ambassador David Rawson, his deputy chief Joyce Leader, a few other embassy officers, and a representative for the US Agency for International Development (USAID).Together, they provided the bulk of intelligence and information on the ground.
When the evacuation of US nationals escalated to a Tier 0 priority, resources were immediately made available in the form of a working group a 24-hour situation room to field constant calls and ensure rapid evacuation.Secretary of State Christopher tasked the embassies in Rwanda and Burundi to ensure that all necessary resources were made available for contact with the working group in Washington. 63In Kigali, Rawson and Leader maintained constant communication to plan the evacuation.Despite the limited resources in Rwanda, support from resources in Washington and Burundi helped speed the process.

Interagency cooperation
As a low-tier nation, intelligence assets dedicated to Rwanda were minimal and, interagency cooperation prior Rwanda was negligible.Apart from State Department involvement in the 1992 Arusha accords, there was no significant need to create working groups for matters in the region.However, because the intelligence collection and analysis became highly centralized under the Clinton administration, the community could quickly establish interagency collaborative efforts.The working group assembled under Bushnell included representatives from eight departments within the government, including African Affairs, the Bureau of Intelligence and Research, and the Bureau of Political and Military Affairs.
By 7 April, the working group had received enough information to identify where the Americans were and to establish evacuations by overland convoy.Collaborative efforts in Washington, Kigali, and Bujumbura allowed for immediate planning and execution.The Government General in Rwanda promised to provide military escort for personnel from the American Embassy. 64They would leave for Bujumbura the next morning.

Intelligence functions
In the years leading up to the genocide, the CIA had issued warnings of potential instability in Rwanda.A January 1993 report warned of possible widespread ethnic violence, 67 and in December the CIA reported that up to 40 million tons of small arms had been transferred to Rwanda from Poland, via Belgium. 68By January 1994, analysts predicted that if violence were to erupt in Rwanda, the worst-case scenario would involve half a million deaths. 69Those most familiar with the region expected that renewed violence would take the form of civil war.
The intelligence community was highly centralized under the Clinton administration.When the plane crash occurred, the collaborative environment allowed them to work swiftly and effectively.Information was collected from US embassies in Rwanda, Burundi, Tanzania, Uganda, and Brussels.A picture of the events rapidly emerged.The plane was possibly shot down by the Hutu presidential guard, 70 who were also responsible for the killing of the Prime Minister and several government officials. 71Further, the extremists had killed two Belgian civilians and 10 Belgian peacekeepers.Most importantly, the intelligence community found that foreign nationals, including about 300 US citizens, remained at risk in Kigali. 72n addition to embassy sources, the National Security Agency monitored the hate-speech on Rwandan national radio, and the National Reconnaissance Office would later be tasked with obtaining satellite imagery of the devastation.Director Woolsey or his deputy, Admiral William Studeman kept Clinton and his Assistant for National Security Affairs, Anthony Lake, apprised of the situation with morning briefs, detailing information from working group meetings. 73

Policymaker response and outcome
Armed with this information flowing from the intelligence community, decision makers could rapidly organize and authorize an overland convoy.On 9 April at 0700, five convoys departed from Kigali and other points, for Bujumbura, Burundi.By 1410, the US Embassy in Bujumbura reported that the final vehicle had safely crossed the border. 74Hours later, a cable stated that the last convoy had arrived at their location at 1712, carrying Ambassador Rawson. 75That same day, Secretary of State Warren Christopher appeared on television, stating, 'In the great tradition, the ambassador was in the last car, so that evacuation has gone very well'. 76In a letter to Congress on 12 April, Clinton said 'I am pleased to report that these operations were successful, that no hostilities were encountered, and that no casualties were suffered by U.S. forces in this operation'. 77he centralized intelligence structure allowed the smooth escalation of priority.Mandate-level functions, including resources, interagency cooperation and authorizations were effective.This allowed rapid collection and analysis of intelligence and rapid decisions by policymakers.When viewed in a table 2, it becomes possible to see that each mandate level function was effectively and rapidly deployed.

Comparing the US evacuation to the genocide non-mandate
The primary concern for the Clinton administration at the start of the fighting was the immediate evacuation of Americans at risk in Rwanda, and the escalation to Tier 0, applied only to the evacuation of Americans.In the broader scope of the Rwandan crisis, the doctrine set forth to limit engagement in peacekeeping operations was in place, and there was no escalation made regarding the genocide.The decision to put economic concerns first would come back to haunt the US.
Once the safety of Americans was assured, the mandate was rapidly de-escalated, and the urgency among senior officials waned.The interagency working groups continued daily meetings on Rwanda, but no longer did they garner the interest of top-level leadership.Anthony Lake stated, 'I was obsessed with Haiti and Bosnia during that period, so Rwanda was, in William Shawcross' words, a 'sideshow', but not even a sideshow -a no-show'. 78Echoing this sentiment, on the morning of the final convoys' arrival in Burundi, Senate minority leader Bob Dole said, 'I don't think we have any national interest there.The Americans are out, and as far as I'm concerned, in Rwanda, that ought to be the end of it'. 79he political will to engage in Rwanda (although not militarily) did not manifest until later into the crisis.Thus, a mandate shift requiring immediate action to address the genocide was never put into place.In part, the decision stemmed from an expectation of violence in the region, and characterization of that violence as a civil war.War in Burundi between ethnic Hutus and Tutsis had broken out in 1993, and the perception was that the same could occur again in Burundi or Rwanda.A 14 March Africa Review from the Directorate of Intelligence warned that the exiled Tutsi members of the Rwandan Patriotic Front (RPF) had threatened to return to war if progress was not made toward a coalition government. 80The civil war in Burundi had resulted in the deaths of nearly 50,000 civilians, and the same was expected in Rwanda.
However, the US was not without warning of possible genocidal intentions in Rwanda.On 11 January 1994, Romeo Dallaire, Commander of United Nations Assistance Mission for Rwanda, sent a fax to the UN headquarters in New York, detailing information from an informant.'Jean-Pierre', a top-level trainer for the Interahamwe, a Hutu paramilitary organization, warned that the group sought to protect Kigali, from the RPF.He revealed that since the start of the UN mission in Rwanda, all Tutsis in Kigali had been registered with the government, likely for the purpose of their extermination.He estimated that 'in 20 minutes his personnel could kill up to 1000 Tutsis'. 81The fax, now known as the 'genocide fax', was largely ignored as a histrionic message from a wellmeaning neophyte who was 'didn't know what he was talking about'. 82part from the genocide fax, there had been other indications that the events of 1994 would devolve into genocide.In early 1992, Johan Swinner, Belgian ambassador to Rwanda, warned that the 'Akazu', a secret extremist group that included President Habyarimana's wife and family, was planning to exterminate the Tutsis and internal Hutu opposition in Rwanda. 83In October 1992, Professor Filip Reyntjens revealed to the Belgian senate a 'Zero Network' organization in Rwanda, which 'constituted a death squad on the Latin American Model'. 84In August of 1993, B.W. Ndiaye, Special Rapporteur for the UN, reported ethnic massacres among the civilian populations. 85This information was shared with the US intelligence community.
After the plane crash, the US intelligence community warned of a civil war that could impact Rwanda and spill into Burundi. 86Similar warnings were made on 8 April. 87Beyond the US, a report from the UN mission in Rwanda dated 12 April characterized the situation as a civil war. 88However, reports from the Red Cross and other local sources to indicate a different nature to the violence.On 10 April, Herve Le Guillouzic, medical coordinator of the International Committee of the Red Cross, stated that the death toll had risen dramatically: 'Yesterday, we were talking about thousands of dead.Today we can start with tens of thousands' (The New York Times. 10 April 1994).
By 13 April, the tenor of warnings from the US intelligence community changed.There is no longer any reference to civil war, rather reports now state Tutsis moving into Kigali were doing so to protect kinsmen from 'rampaging government troops'. 89An 18 April NID identifies 'rogue elements of the military and Hutu civilians intent on killing Tutsis' (emphasis added). 90The 20 April intelligence report uses the phrase 'slaughtering Tutsi civilians'.Finally, the word genocide appears for the first (and only) time on 23 April.By 26 April, the NID states that 'an accurate death toll may never be available'. 91As the urgency of the situation became apparent, there was still no political will to reescalate Rwanda to crisis level.In fact, the events spurred Anthony Lake to hasten the process of formalizing the Presidential directive to limit multi-lateral engagements.
Thus, when compared to the evacuation, there were significant distinctions at each stage in the mandate-level functions.Rwanda had dropped from a Tier 0 issue.Many of resources that were made available at the start of the crisis disappeared when the last car arrived in Bujumbura.In terms of interagency collaboration, the working group on Rwanda did continue daily meetings, largely addressing humanitarian and diplomatic efforts.Senior-level officials did not attend the meetings.
National Intelligence Daily reports continued to cover Rwanda, but Ambassador Rawson and Deputy Ambassador Leader, who had returned from Kigali and were valuable resources of information, were never fully utilized.Upon Leader's return, she was tasked with drafting the daily State Department summaries on Rwanda.However, she was told to rely only media reporting and US intelligence reports.She was not allowed to contact her sources in Rwanda.This limiting of firstperson information was deliberate; it kept the crisis abstract as the government was reluctant to acknowledge that genocide was occurring.
A declassified discussion paper from the Rwanda working group lists short-term policy objectives, with feedback comments largely cautioning to avoid language that could lead to engagement.For example, a suggestion for a genocide investigation was met with the comment, 'Be Careful.Legal at State was worried about this yesterday -genocide finding could commit [the US government] to actually "do something".' 92 Acknowledgment of genocide would have triggered US action in accordance with the Geneva Convention, and forced a mandate escalation.The government was extremely cautious to avoid this terminology and a new mandate in Rwanda.In observing the absence of the mandate shift, we can identify the differences to mandate-level structures that inform decision making and outcomes, as demonstrated in table 3.
On 17 May, the United Nations Security Council issued a resolution saying that 'acts of genocide may have been committed' and agreed to send 5,500 troops to the region.However, troop deployment was delayed by an inability to obtain national contributions of troops and equipment.Further, there were disagreements between the Pentagon and UN over the financing of the operation. 93It was not until 25 May that the US government began to use the phrase 'acts of genocide'.An exchange between State Department Spokeswoman Christine Shelly and reporter Alan Elsner on June 10 shows that the media was aware of the semantics: REPORTER: How many acts of genocide does it take to make genocide?CHRISTINE SHELLY: Alan, that's just not a question that I'm in a position to answer. 94r from sending troops, the US government was also reluctant to spend money on other nonmilitary options.On 5 May, a request came from Deputy National Security Advisor Sandy Berger for US engagement to jam the radio station, Radio Televisión Libre des Mille Collines, which was integral in spreading anti-Tutsi propaganda. 95The request was denied by Frank Wisner, Under Secretary of Defense, who claimed that ground-based operations in mountainous terrain would be ineffective, and airborne operations would cost $8500 per flight hour.Wisner urged focus on relief efforts in neighbouring countries. 96he Clinton administration relented in July, when Clinton went before Congress to request $320 million emergency aid and announced the deployment of 4000 troops to Zaire. 97The troops would assist in relief efforts only -not peacekeeping.Ultimately, the US, so intent on maintaining a budget, spent $237 million in aid.Had the administration approved a mandate escalation, or UN reinforcements when they were requested, the cost would have been only $30 million. 98

Discussion
The US responses to the Rwandan genocide are a good expression of both the efficiencies and inefficiencies that can occur within the R&P process.The functions of the requirements and priorities process tend to be most visible during mandate shifts.Prior to the genocide, Rwanda was a low priority nation, and the early moments of the genocide caused an urgent, ad-hoc escalation of priority and rapid realignment of mandate-level functions.Thus, the Rwandan genocide provides a snapshot of strengths and weaknesses of the process.It is possible to observe how each of the mandate-level components functioned and displays nested strengths and weaknesses related to each component.Rwanda represents a dichotomy of the best and worst aspects of the Clinton administration's tiered system of requirements and priorities.The successful evacuation of US expats benefitted from the speed and acuity of being a Tier 0 operation.However, the genocide itself received sluggish and reluctant escalation of priority.In some respects, the delay in response can be measured by the loss of lives.
A 1996 report by House Permanent Select Committee on Intelligence staff highlighted two critical factors: first, Tier 0, which was established to address short-term crises, had become a holding place for prolonged conflicts.This was a double-edged sword; resources were dedicated to current intelligence, while long-term estimates suffered. 99Further, the top-heavy system meant that only the top five or six requirements were provided with the bulk of resources, while lower priority issues were left to 'languish with leftovers at best'. 100 Second, under the directive, adjustments to any priorities required presidential sign-off.But as Clinton became embroiled in personal and political trials, presidential review of priorities tapered off.As a result, emerging issues remained at a lower tier, while issues that no longer commanded urgent attention still maintained high priority.
Looking at the US response in Rwanda through the lens of requirements and priorities highlights the strengths and weaknesses of the mechanism.First, because priority escalation is the first stage of the mandate, the requirement of presidential review should have been addressed earlier, particularly as it became more apparent that Clinton was unable to conduct regular reassessments.
Second, the Clinton administration was heavily focused on economic concerns, and the reluctance to enter Rwanda was a decision based considerably on cost-saving measures.However, the cost of after-the-fact aid was nearly eight times greater than the expected cost of UN reinforcement.Benjamin Franklin's adage becomes appropriate: 'an ounce of prevention is worth a pound of cure'.Nowhere is this truer than in the failure to review the emerging issue of the growth and movements of Al Qaeda; the US is still paying the costs.
No actions were taken to rectify weaknesses in the requirements and priorities during the Clinton administration.By 2000, a report warned that the process was failing, the priorities had not been reviewed, and there was no mechanism to communicate to decision makers the problems caused by assigning military force protection to Tier 0. 101 By the time George W. Bush took office in 2001, the National Security Agency had a list of 1,500 formal requirements and 20,000 'essential elements of information'. 102Consequently, when the September 11 th attacks happened, Afghanistan was considered a Tier 3 issue, and the US military had not ordered a new map of the nation for four years. 103

Conclusion
This research has argued the utility of observing national security events through a standardized model which centralizes the requirements and priorities process.The goal was not to add another category of failure to the arsenal of existing descriptors, but to examine outcomes at a macroscopic level.Based on the findings observed, the mandatelevel functions did in fact play a role in the outcomes of urgent events and could be useful in after-action evaluations.
There are several key benefits to applying RPC model.First, it creates a common, system-wide method for the examination of outcomes.As stated, there is currently no single method of investigation utilized to observe outcomes in the aftermath of a national security or foreign policy event.In absence of this, various examinations contain different (often isolated) areas of focus without observing the systemic functions that link them.This is not to suggest that investigations should not be conducted at the department or agency level.However, in the absence of a systemic investigation, recommendations for reform can (and have) been based on only part of the picture.A second benefit in establishing a common model of examination is that over time, it would result in a catalogue of 'like-for-like' assessments.This might allow examinations to pinpoint common threads of concern and identify consistent weaknesses and strengths and can be useful in supporting a strategy of evidence-based policy.
Additionally, the RPC model can address accountability issues and prevent an overload of ad-hoc escalations.Historically, stakeholders have neglected properly use and update the requirements and priorities process, leading to an accumulation of unchanged and outdated requirements.The current iteration R&P model, the National Intelligence Priorities Framework 104 (introduced in 2003) has been the most recent attempt to resolve the issues associated with the prioritisation process.However, well intended, there is evidence that it is not yet optimised as we observe surprises such as the Arab Spring, the fall of the Afghan government, and the Russian invasion of Ukraine.
Using the RPC model as a post-action review tool and an audit trail can account for whether needs are presented and updated regularly.This causes an enforcement of accountability and mitigates against the escalation of too many ad-hoc issues, preventing the rapid depletion of supplemental funds. 105Holding intelligence and policy leadership simultaneously accountable for the upkeep of the requirements and priorities mechanism fosters dialogue and provides an opportunity to discuss the capabilities and limitations of intelligence.
Finally, because the model was not developed specifically for these cases, its general application has transitive properties which allow it to be tested against other cases of national security.As stated earlier, a version of the requirements and priorities process exists in varying forms, and under various names, across any areas where various missions and strategies exist.For example, implementation of increased support for veterans, addressing the COVID-19 pandemic, or modifying a national education system would all require the establishment of priorities, and the subsequent provision of resources, authorizations, and collaboration.In terms of audits and isolating root causes for outcomes, the model could be modified to address functions as they relate to other departments.The RPC model can inform a less myopic view by presenting a more holistic observation of systemic functions.This could potentially influence the way that government thinks about reform.

Disclosure statement
No potential conflict of interest was reported by the author(s).

Notes on contributor
Dr Neveen Shaaban Abdalla is a lecturer of Intelligence, Defense and Security at Brunel University London, and a member of the Brunel Centre for Intelligence and Security Studies.Her research explores the role of the requirements and priorities processes in cases of early warning success and failure, putting forth a model for holistic investigations into national security outcomes.

Table 2 .
RPC model tested against US response to evacuation of American expats.

Table 3 .
RPC model tested against US response to intervention during Rwandan genocide.CHRISTINE SHELLY: [. ..]We have every reason to believe that acts of genocide have occurred in Rwanda.