From the 1960s and 1970s, ‘affirmative action’ for particular demographic groups (not Far Eastern Asians, for example) grew steadily among Western governments, educational institutions, and organizations generally. Since 2009, the expression ‘affirmative action’ has fallen out of favor in the United States and has been succeeded by ‘diversity, equity, and inclusion’ (DEI) initiatives. Ramping up especially during the years of the Obama presidency, DEI personnel policies have spread greatly. These DEI policies are aimed to grow the proportions of various designated demographic groups. The training, content, and programming policies of the DEI agenda also promote various favored ideas, in university curricula and activities, for example.
Whether the DEI initiatives are beneficial, on net, is a complicated question, but advocates say they are ethically desirable and beneficial to organizations, to the world at large, or both. DEI advocacy implies that, prior to DEI initiatives, the net benefits of enhanced demographic diversity had been neglected or forgone. When a university adopts an ‘anti-racism’ DEI initiative, for example, it also implicitly admits to having practiced racism, since, otherwise, there would be no problem to fix. DEI is a remedy to a problem—discrimination against DEI-privileged demographic identity groups—which has not plausibly existed for decades. When critics of DEI request evidence of instances from recent decades of racism or other forms of bigotry or impropriety against DEI-favored groups, the request is sometimes met with vague insinuations that the instances, though pervasive, are too subtle—‘unconscious’—to identify and evidence; more commonly the request is simply ignored. Advocates of DEI may present disparities in group-level variables such as income, wealth, life expectancy, or presence in various vocations as evidence of a ‘systemic’ problem, even though disparities might flow from differences in the quality of K–12 schools or causes other than racism.
Besides asking for evidence of a problem, critics often ask: What are the social mechanisms by which DEI policies generate net benefits? What problems do DEI policies cause? What is the total effect of DEI policies? What evidence is there that the total effect is positive?
The total effect of any course of action is difficult to estimate, because it consists of many hard-to-assess particular effects and because there is no definite way to ‘add up’ all of the particular effects. Often the sides focus on only some of the particular effects. Claims and evidence are made for particular effects, touted as ‘findings,’ which are then marshaled by one side or the other. Deirdre McCloskey (1990, 73) has noted that the methodology of some social scientists “resembles the method of the drunk who looks for his keys under the lamppost because the light is better there.”
In this article I treat DEI in U.S. intelligence services, generally known as the intelligence community (IC), and focus on the claim that DEI policies improve the operational performance of the agencies. The IC consists of 18 agencies, including major ones such as the Central Intelligence Agency (CIA), National Security Agency, and Defense Intelligence Agency. Like the rest of federal-government personnel policy, the IC has been a setting of burgeoning DEI policies and programs, especially since 2011, designed to favor privileged demographic groups in hiring, promotions, awards, and assignments. The topic is strategically important because, if critics of DEI policies are right, DEI has potentially significant negative ramifications for the national security of nations that adopt such policies (Gentry 2024a). Intelligence is the first line of national defense. Policymakers (and citizens) therefore need to know which perspective, that of the DEI advocates or that of the DEI critics, is more accurate.
I have a background in the U.S. military and the IC, and in academic research on intelligence-related issues. I have published extensively on intelligence issues, including assessments of the IC’s operational procedures and performance. From my experience and research, I know the challenges of conceptualizing, measuring, and publicly documenting the effects of reforms in the IC. Those challenges are far greater in intelligence studies than in research into most other areas of society. In private-sector activities, including economic enterprise, performance meets public life, in the marketplace and open discourse, and unfolds mainly within a context of voluntary choice. The government sector is different, and that difference is heightened within the IC given the context-sensitive and highly interpretive nature of much intelligence work and its inherent secrecy. The conceptualizing of ‘benefits’ or ‘good’ performance is almost as controversial as are the ends of the IC itself. Next, there is the problem of measuring performance. And then there is the matter of whether evidence, data, and testimony can be made public. Much is classified or otherwise unverifiable by external researchers. Readers of normal social science research need to understand that IC performance-focused research is different. Consider two contrasts, both in connection with the issue of DEI.
Many DEI advocates, following a series of prominent McKinsey & Co. reports (e.g., Hunt et al. 2015), have claimed that DEI policies are operationally effective. The McKinsey studies claim to find correlations between demographic diversity and positive outcomes for American businesses (such as industry-adjusted earnings before interest and taxes as a percentage of revenues). The studies involve measuring demographic diversity in a large number of organizations, measuring quantified outcomes, and doing a statistical analysis in search of correlations. These features make the investigation viable as quantitative empirical research. They also make the findings subject to direct replication, provided that the methods and data are provided (which McKinsey did not do). Those features also make criticism potentially powerful, as when McKinsey’s results were challenged by researchers who undertook a like investigation and found no such correlations (Green and Hand 2024). Now, think about the IC. For an investigation of particular effects of DEI in the IC, the number of organizations is not large, they do not aim at a particular tangible goal (like earnings), and, if there is any measurement of success, those measurements are vague, often assigned a confidence-level degree that is inherently subjective, and likely are classified. Quantitative methods are largely out of reach. But even qualitative methods are difficult.
For a second contrast, consider the methods that a sociologist or social psychologist might take to studying the effects of DEI in some types of organizations. One way to find out how a reform affects people is to ask them. Researchers develop survey, questionnaire, and interview formats, and in their scholarship explain their methods. They describe the target population, whether counterparties, such as employers and employees, are being queried, how the interviewees were approached, whether gatekeepers affected selection of interviewees, and the response rate. They describe how the interview was conducted (e.g., structured or unstructured, thematic or particularistic), whether and how the conversation was recorded, whether notes are available, whether anonymity was optional, how informed consent was assured, and whether and what kind of prior approval was obtained. For the IC, however, many of the normal methods are complicated, ill-defined, or simply unviable. IC agencies do not allow external researchers access to their employees. Even when viable, the documentation may not be verifiable because of classification status and other sensitivities that do not exist in other settings. For example, see Rob Johnston (2005) on the analytic culture in the IC—one of the very few IC-produced studies of intelligence activities made public. Below I offer some description of the methods relied upon in my own research, including this article.
Next I provide some background, and then turn to the claims made in favor of DEI in the IC, which form the target of my criticisms. I argue that, contrary to the claims of intelligence agencies, DEI damages the operational performance of the agencies.
What explains
excellent intelligence performance?
The performance or effectiveness under discussion includes all aspects of intelligence operations, including the acquisition of information and its conversion into ‘intelligence’ through processes of analysis. The aim of intelligence analysis is to provide national leaders with data and judgments about security threats and the general geopolitical situation useful for improving national decision-making. In 1990, the CIA’s Studies in Intelligence published an anonymous piece on the Directorate of Intelligence (DI) titled “How to Succeed and Survive the DI’s Organizational Culture” (Anonymous 1990), which offers a useful circa-1990 description of the work of intelligence analysts.
Intelligence analysis is a prime task of the IC, but it is not the only one. The IC collects information and conducts various sorts of operations, most of which are highly classified. Also, the popular image of the IC is affected by how intelligence products are filtered and selected at the upper reaches of political leadership. There again, troubles and controversy abound. For the issue under discussion, however, the main focus here is on intelligence analysis, conducted by analysts.
Highly pertinent to any assessment of DEI policies are decades of scholarship and practical experience regarding effectiveness in the IC agencies. The intelligence-studies literature, going back to lessons from World War II, offers many insightful assessments of the traits that make good intelligence officers (Gentry 2023a, 575–587). The studies provide detailed descriptions of key characteristics of individual people who do well in a broad range of intelligence activities including military and civilian intelligence services and Western and Soviet bloc services. The key characteristics include personality traits, thinking styles, education, and experiences. These discussions address such matters as: What makes a good case officer and the specialties supporting human intelligence collection; key elements in various types of analysis, including strategic warning; and prominent traits of deception specialists. Well-recognized errors caused by cognitive biases cannot be addressed by demographic diversity; only intellectual preparation, self-awareness, and diverse educations and experiences can help (Gentry 2016b, 658, 665–666). CIA careerist Richards Heuer (1999) began to study cognitive biases in CIA analysts in the 1970s, compiling a series of initially classified articles published in Studies in Intelligence into a now-classic book, published in 1999, that identified methods, now called “structured analytic techniques” (SATs), that are designed to overcome primarily cognitive biases in both individuals and groups of people. These apply to the brains of human beings without differentiation into the identity groups that are the primary focus of DEI advocates. Such cognitive techniques have nothing whatever to do with overcoming biases associated with demographic group interactions.
The IC’s traditional concept of desirable diversity concerns, among other things, understandings, not of the population of the United States that is the focus of DEI advocates, but of the world outside of the United States—understandings that the mainly foreign-focused IC agencies try to comprehend. By traditional reasoning, intellectual diversity helps intelligence organizations collectively think and thereby perform better. If operating in Sri Lanka, for example, it makes sense for some U.S. intelligence officers to ‘look like’ the folks of Colombo, not those of Cleveland. It has long been clear to U.S. intelligence agencies, and to reputable external observers, that persons from all significant demographic groups can be, and are, capable intelligence officers (Gentry 2023a, 575–585). But it is individual characteristics, not those of large demographic identity groups, that are important. Moreover, there is a disconnect between the DEI emphasis on domestically defined demographic diversity as an aid to business performance in the United States and the different skills and backgrounds needed to effectively address foreign intelligence problems.
Nothing in the traditional intelligence studies literature suggests that large-group demographic diversity of employees is, per se, an important ingredient of agencies’ operational performance. Consider one of the finest intelligence operations of all time, the intelligence-enabled deception that convinced Hitler and the German military that the Western allies’ invasion of France would occur in the Pas de Calais region of northern France and not Normandy in 1944. The record documents that the operation was conducted virtually entirely by Caucasian British men from a wide range of cultural and professional backgrounds who were drawn together by the existential threat posed to Britain by Hitler’s Germany (Hesketh 2000). The shared sense of common purpose is perhaps the most important ingredient in teamwork of any kind. A key asset is different perspectives on how to achieve goals and how to assess weaknesses and dangers in any proposed course of action. The D-Day example illustrates that demographic diversity is not necessary to produce fine operational performances. Demographic diversity is not essential except sometimes in particular operational assignments.
Traditionally, intelligence analysts have been drilled on the importance of thorough work. The sometimes-lengthy coordination and review processes of all IC analytic organizations, often done by floating groups of many analysts and managers with appreciably different intellectual and demographic backgrounds, long have been designed to help ensure the completeness of analysis and reduce bias in analytic products. Even in their better days, the organizations did so with incomplete success. But they generated bureaucratic incentives for good work (Gentry 2016a, 160–163). Rapid progress through the ‘review process’ by colleagues and diverse groups of managers that all U.S. intelligence analyses undergo before publication is known to be career-enhancing. The anonymous 1990 analyst mentioned above, writes:
Every DI analyst regularly experiences the pressure of not wanting to be wrong. Consequently, the good ones develop techniques for reducing the risks through research habits, collaborative analysis, alternative hypothesis exploration, retesting assumptions, and challenging the conventional wisdom. (Anonymous 1990, 24)
The larger point is that the IC has long been interested in understanding what makes for operational effectiveness, and there is no reason to suppose that the IC had, prior to 2011, erred by unduly discounting demographic diversity. To my knowledge, the DEI advocates have never given any evidence of an IC performance problem to be fixed, nor of any instance in which one of the periodic failures of U.S. intelligence has been attributed to insufficient demographic diversity. Indeed, no one has ever credibly attributed even one intelligence failure to a dearth of demographic diversity.
Policy background since 2011
The promulgation of DEI terminology was advanced markedly by President Barack Obama, during 2009–2017, who promised to “transform” America (Weigel 2011). Obama spoke extensively about the importance of preferential treatment for politically favored demographic categories, especially blacks, women, and LGBTQ+ people (link). He introduced what we now call DEI with Executive Order 13583 of August 18, 2011, “Establishing a Coordinated Government-wide Initiative to Promote Diversity and Inclusion in the Federal Workforce.” Although some of the major DEI policy documents published by IC elements give passing nods to the old virtues of diversity of outlooks, foreign experiences, language expertise, and so on, the new emphasis, following Obama, is domestic and demographic. The new emphasis is also group-oriented, in that, rather than considering the diversities that are present in a set of finalists for a position, the diversities that now count are the few placard diversities concerning race, gender, and so on. It is a form of identity politics. The old program of ‘equal opportunity employment’ has effectively been superseded. DEI became a core component of what Obama and his supporters frequently called “our values” (link).
Obama relied heavily on two senior intelligence leaders to advance his program in the IC—Director of National Intelligence James Clapper and CIA director John Brennan. DNI from 2010 to 2017, Clapper (2018, 301) insisted that DEI policies cause improvement in the operational performance of intelligence services. He said that it was “advantageous to employ openly transgender employees, who brought unique perspectives to mission challenges and contributed to successes” (ibid.). But, inconsistently with IC policy (link) that requires all analytic judgments to be supported by evidence to the extent possible, Clapper never explained how domestically defined demographic diversity improves the IC. He even noted that a major cause of the increased hiring of LGBTQ+ people in his years as DNI was the “tremendous outside pressure from LGBT groups that were seasoned from fighting for gay, lesbian, and bisexual Americans’ rights and had embraced transgender rights as the next battlefield” (Clapper 2018, 301).
John Brennan, CIA director from 2013–2017, similarly asserted without supporting evidence that “CIA simply must do more to develop the diverse and inclusive leadership environment that our values require and that our mission demands” (Brennan 2020, 7, 94, 132, 140, 283, 270, 286, 414; Gentry 2023a). Brennan was especially aggressive at politicizing the CIA. He imposed what some analysts among themselves called a form of ideology-driven “soft totalitarianism,” enforced by the diversity offices created by Obama’s 2011 executive order (Gentry 2023b, 194). Obama’s discriminatory hiring practices gradually altered the demography of the IC, which gradually came to have more women and minorities. He thereby created cadres supportive of DEI policies on ideological grounds and others who liked DEI policies because they gave privileged identity groups material advantages in forms such as preferential promotions and assignments. He stated quite candidly that he sought to change CIA’s organizational cultures in politically significant ways, especially by embedding traits consistent with what we now call DEI (Brennan 2020, 7, 94, 132, 140, 283, 270, 286, 414). When Donald Trump emerged in mid-2016 as a serious presidential candidate, Brennan pointedly directed CIA personnel to be politically active in defense of the DEI-related ‘progress’ he had achieved (Brennan 2020, 3, 94, 414).
As of this writing, the Office of the Director of National Intelligence (ODNI) website (link) emphasizes DEI and annually reports details of the IC’s demographic diversity, now required by law. The website tells of “IC’s struggle to recruit talented officers who mirror the diverse country they serve.” The mirror metaphor is apt. The IC, at the behest of Presidents Obama and Joe Biden, seeks to make the IC ‘look like’ America. Biden administration officials continue to tout DEI. For example, Secretary of State Antony Blinken cited racial statistics in claiming that fostering President Biden’s DEI policies is a national security priority (Gaouette 2021). He required each of the State Department’s regional bureaus to designate a deputy assistant secretary to focus on creating more diversity, and he said his goal “is to incorporate diversity and inclusion into the department’s work at every level” (ibid.). State’s Bureau of Intelligence and Research, its intelligence unit, now has its own dedicated diversity officer.
The old IC concept of trying to find and deploy diversity of outlook, education, and experience in individual people as tools of effective performance therefore is largely gone. Evidence of this change now is abundant. Officials effectively say so regularly by emphasizing only demographic diversity. Annual reports of the demographic diversity of the intelligence workforce contain no diversity-related performance metrics or even discussion about the desirability of developing such measures (ODNI 2023). Questioning the efficacy of any aspect of DEI policies or suggesting that claims about the performance implications of DEI policies should be tested is bureaucratically dangerous for federal employees and for scholars who study DEI issues at similarly ‘woke’ universities. Illiberal intolerance of such questioning is a serious and still-growing threat to the effectiveness of agencies whose mission is to make America more secure.
Claims that DEI helps intelligence work
The assertions, shown above, by Clapper, Brennan, Blinken, and others, that DEI is good for IC performance are little more than assertions. The IC DEI advocates provide very little material for skeptics to engage. The single most significant item specifically on why DEI is good for IC performance is a 985-word popular piece by the former deputy chief of the CIA’s analysis directorate Carmen Medina. The piece is titled “Want a Sharper, Stronger CIA? Hire Folks Who Look Like America,” and it appeared in an online magazine called Overt Action. That magazine is now defunct, but Medina’s article has been preserved by the Internet Archive (link).
Medina’s conclusion is captured by the article’s title: “Hire folks who look like America” to improve the CIA. She writes: “So my bottom line is that the CIA’s recommitment to an aggressive Diversity and Inclusion Strategy is not only good for society—which in a democracy must factor into government decision-making—but also great for the mission” (Medina 2016).
Medina asserts that she emphasizes causal mechanisms, writing, “I need to understand how diversity leads to better analysis and decisions.” The main mechanism, she maintains, is that “simply interacting with individuals who are different forces group members to prepare better, to anticipate alternative viewpoints and to expect that reaching consensus will take effort.” She writes that CIA “managers will need to learn new skills to deal with the higher levels of tension characteristic of healthy, diverse teams. As the research shows, it’s that friction that contributes to better outcomes” (Medina 2016).
For support, Medina cites a 2014 article in Scientific American by Katherine Phillips called “How Diversity Makes Us Smarter.” Clapper, Brennan, and others less clearly have used Medina’s arguments, often without citing Medina, let alone Phillips. As of August 13, 2024, Phillips’s paper had 808 Google citations.The figure of 808 includes both the citations for “How Diversity Makes Us Smarter” and those for “How Diversity Works” which packages the Phillips article interspersed with three brief pieces by others. Phillips clarifies what kind of diversity she is speaking of by asking, “What good comes from diversity of race, ethnicity, gender and sexual orientation?” The article is generally about diversity along such DEI lines; it never mentions the intelligence agencies but offers a causal mechanism intelligence officials have borrowed, claiming it also applies to intelligence work. To show the intuition, she poses a thought experiment:
Consider the following scenario: You are writing up a section of a paper for presentation at an upcoming conference. You are anticipating some disagreement and potential difficulty communicating because your collaborator is American and you are Chinese. Because of one social distinction, you may focus on other differences between yourself and that person, such as her or his culture, upbringing and experiences—differences that you would not expect from another Chinese collaborator. How do you prepare for the meeting? In all likelihood, you will work harder on explaining your rationale and anticipating alternatives than you would have otherwise.
This is how diversity works: by promoting hard work and creativity; by encouraging the consideration of alternatives even before any interpersonal interaction takes place. The pain associated with diversity can be thought of as the pain of exercise. You have to push yourself to grow your muscles. The pain, as the old saw goes, produces the gain. In just the same way, we need diversity—in teams, organizations and society as a whole—if we are to change, grow and innovate. (Phillips 2014)
In other words, competition with demographic ‘others’ drives better personal, and thence organizational, performance. Phillips relates findings of a number of studies; Medina refers to one of the studies mentioned by Phillips and also refers to two others. I now turn to some critical thoughts about both Medina (2016) and Phillips (2014).
Critical thoughts on
Medina (2016) and Phillips (2014)
Phillips and Medina may be correct in saying that demographic and background differences among interactors sometimes are associated with each being less confident that the others will accept their own suppositions and interpretative frameworks, and consequently they may have to work harder to justify those suppositions and interpretive frameworks. However, preparing for and conducting such conflict is rarely important in the IC, as the British deception of 1944 and many other successful intelligence operations illustrate. At the beginning of her article, Phillips writes:
Research has shown that social diversity in a group can cause discomfort, rougher interactions, a lack of trust, greater perceived interpersonal conflict, lower communication, less cohesion, more concern about disrespect, and other problems. So what is the upside? (Phillips 2014)
Neither Phillips nor Medina argue that organizations systematically err in reckoning the mix of their personnel. Phillips (2014) notes that diversity of expertise—“engineers, designers and quality-control experts”—is vital to organizational effectiveness. She never explains why organizations would be getting that kind of diversity right while they get demographic diversity wrong. Both Phillips and Medina merely highlight one possible effect of greater demographic diversity, without ever making an argument about why that effect would be systematically misestimated. As with universities imposing “anti-racism” programs, they promote DEI policies on the basis of identity politics without ever establishing that we should think that there is a performance problem to be fixed.
Phillips and Medina report that a number of studies show that diversity is beneficial. I will offer a few remarks about some of the studies, but some general remarks are in order. First, again, a necessary (but not sufficient) condition for the imposition or promotion of DEI as a useful performance-enhancing tool is that systematic erring that cannot be overcome by learning and competition among organizations had been occurring, and none of the studies provide evidence of systematic erring by organizations. Second, published research studies are myriad, and Phillips and Medina mention only a few. Third, there is publication bias, particularly on a topic like DEI. Besides the bias toward statistically significant findings (the ‘file-drawer problem’), the vast majority of journals in the social sciences are left-wing in political orientation. DEI is an ideological agenda as well as a set of policies and programs.
Phillips cites two studies that show that diversity pays. One study “found that [U.S.] companies that prioritized innovation saw greater financial gains when women were part of the top leadership ranks.” The other “found that companies with one or more women on the board delivered higher average returns on equity, lower [debt to equity], and better average growth.” These results are analogous to the results of the McKinsey studies, which Jeremiah Green and John R. M. Hand (2024) did not find upon replication. But, at any rate, the relationships are correlations, as Phillips notes, and do not show the direction of causality. Phillips also cites a survey of bank executives, finding that “for innovation-focused banks, increases in racial diversity were clearly related to enhanced financial performance” (Phillips 2014), and a study of 1.5 million scientific papers, finding that “papers written by diverse groups receive more citations and have higher impact factors than papers written by people from the same ethnic group” (Phillips 2014). Once again, Phillips’s presentation suggests zero evidence of systematic erring and no basis for ascertaining the direction of causation or other causal explanations.
The four other studies referred to by Phillips are experiments. One asked subjects to solve a murder mystery, another again asked subjects, now either Republican or Democratic, to solve a murder mystery, another asked subjects to discuss either child labor practices or the death penalty, and one asked subjects to act out mock jury trials. Thus, none of them are about how organizations function; as noted, intelligence analytic work is a corporate process in which analysts’ drafts are thoroughly reviewed by colleagues and managers whose often significant intellectual diversity and evolving memberships (generally) improve analytic products. In addition, none of the subjects of these experiments is close to most of the analytic work of foreign-focused intelligence agencies.
Phillips tells us that in the four experiments the groups with diverse subjects produced outcomes more positive than the outcomes of the nondiverse groups. The diverse groups “significantly outperformed the groups with no racial diversity” in solving the murder mystery. In the variant in which Democrats and Republicans were sometimes mixed and sometimes not, subjects in unmixed groups “prepared less well.” In the discussion experiment, presentations by black subjects to white listeners were “perceived as more novel and led to broader thinking and consideration of alternatives.” In mock jury-trial experiments, “the diverse juries were better at considering case facts, made fewer errors recalling relevant information and displayed a greater openness to discussing the role of race in the case” (Phillips 2014). Such are the positive outcomes reported.
Medina (2016) follows Phillips but brings matters specifically to the matter of DEI diversity in the IC. McKinsey and Phillips offer related arguments that Medina makes (allegedly) relevant to the IC by assuming that the asserted benefits of domestically defined demographic diversity similarly improve the performance of foreign-focused intelligence agencies. She refers to only three studies, one of which is the murder-mystery experiment. Her presentations of the other two studies are highly flawed. (The studies are identified by hyperlinks provided in Medina’s article.)
Medina discusses an article in the Proceedings of the National Academy of Sciences titled “Ethnic Diversity Deflates Price Bubbles” (Levine et al. 2014). Here is Medina’s complete presentation and discussion of the study:
In another study, this time of market traders, published in the Proceedings of the National Academy of Sciences in 2014, researchers found that markets populated with diverse traders suffered fewer and less catastrophic price bubbles than more homogeneous markets. As was the case for the solvers of murder mysteries, diversity facilitated the kind of healthy friction among traders that led to better decisions and upended conformity. The researchers discovered that price bubbles arose not only from individual errors or financial conditions, but also from “the social context of decision making.” This example is particularly compelling because the traders can be considered experts in their field. In other words, even when everyone is an expert, diversity leads to better results. (Medina 2016)
This presentation of the study is misleading in two ways. First, it makes it sound as though the study examined diversity and trading in the real world. In fact, the article (Levine et al. 2014) reports on experiments conducted at universities in Singapore and Texas. Second, Medina stresses that “This example is particularly compelling because the traders can be considered experts in their field. In other words, even when everyone is an expert, diversity leads to better results.” In fact, the experimental subjects were not actual traders with expertise. They were university students and staff “trained in business or finance” (Levine et al. 2014, 18525). The study and its published appendix (Levine et al. 2014), appear to say nothing about the mixing and performance of staff members within the experimental groups. The appendix notes, “we sampled only from university students and staff” (Levine et al. 2014, appendix p. 517). Nowhere else is the word “staff” found in either the article or the appendix. One would think that the inclusion of staff as experimental subjects would be reported upon. The age, status, and savvy of staff-member subjects, as opposed to student subjects, is itself a kind of diversity, and the effects of that diversity would be a factor to control for, track, and report on.
Medina refers to the Good Judgment Project, run by University of Pennsylvania psychologists Philip Tetlock and Barbara Mellers under the sponsorship of the Intelligence Advanced Research Projects Activity (IARPA), a unit of the Office of the Director of National Intelligence. The Good Judgment Project (GJP) assessed characteristics of people who did particularly well in a series of forecasting tournaments conducted between 2011–2015. Without saying so plainly, Medina (2016) seems to imply that the GJP findings support the notion that DEI hires can be trained up, since, as she says, “forecasting is a skill that that can be learned independently of a functional or regional expertise.” She writes: “The Good Judgment Project, funded by IARPA and led by University of Pennsylvania professor Philip Tetlock, has found that experts do no better and often worse than non-experts in predicting future international events.”
But the GJP made no formalized comparison between non-experts and experts. The GJP recruited volunteers from the general public to participate in forecasting tournaments. During the course of the tournaments, the GJP “stratified forecasters into groups on the basis of performance,” dubbing the best of the best “superforecasters” (Mellers et al. 2015b, 269). The GJP focused on the superforecasters, a group of proven top performers. Also, the GJP compared the effects of different forms of prepping and different settings of working on forecasting performance (Ungar et al. 2012; Mellers et al. 2014; 2015a; 2015b). The GJP did not publish superforecasters’ demographic characteristics.
While the forecasting tournaments were still ongoing, the journalist David Ignatius (2013) claimed, based on confidential information unseen by the GJP about actual IC officials, that the GJP’s “top forecasters…performed about 30 percent better than the average for intelligence community analysts who could read intercepts and other secret data.” At the conclusion of the forecasting tournaments, commenting on Ignatius’s claim, Tetlock stated that he was “willing to make a big reputational bet that the superforecasters beat the intelligence analysts in each year in which such comparisons were possible” (Tetlock and Gardner 2015, 301 n.11). But the claim that Tetlock’s top forecasters outperformed the IC average has never been substantiated, and, at any rate, did not come from the GJP, even if Tetlock finds it plausible.
Besides selecting the best and focusing on them, the GJP found that some forms of training provided by the GJP—such as probability training, scenario training, and cognitive debiasing training—led to more accurate forecasts (Mellers et al. 2015a, 8; Mellers et al. 2014, 1109–1110; Mellers et al. 2015b, 269). The GJP also developed methods of measuring forecast success and studied the traits of superforecasters, and then proposed methods to improve forecasting skill. The main takeaway from the GJP is that some people forecast better than others and that one’s ability to make accurate forecasts about important world events can be improved—good forecasting is primarily about developing a series of skills, skills that the Project’s superforecasters show deftness in (Tetlock and Gardner 2015, 18–19). In 2017 (hence, after Medina 2016), in a new Preface to his book Expert Political Judgment, Tetlock remarks that benefit can come from forecast-team members holding “different knowledge bases and points of view,” but he seems to mean diverse intellectual viewpoints, not demographic differences (Tetlock 2017, xxvii). In this way, he supports the traditional view of sources of intelligence excellence. Meanwhile, what Medina promotes for the IC is not the sort of training the GJP found to be helpfulIt is important to note that the trainings appear not to have been the decisive factor in forecasting success even though the Project did find they helped improve forecasting. See Mellers et al. 2014, 1110; Mellers et al. 2015b, 269. but rather DEI training. DEI training is surely not the sort of training the GJP provided to participants.In discussing the results of the Good Judgment Project, Mellers et al. state, for example, that “Forecasters in Year 1 performed better when they received cognitive-debiasing training” (2015b, 269). It is important to underscore that cognitive debiasing training is not the same as DEI training. DEI training—for example, unconscious (or implicit) bias training—typically aim at addressing purported racial, sexual, and gender biases from a stylized victim-oppressor lens. DEI trainings are distinct from cognitive debiasing training. Itamar Shatz (n.d.) defines cognitive biases as “systematic patterns of deviation from rationality, which occur due to the way our cognitive system works.” Shatz gives the example of the so-called “ostrich effect” as one form of cognitive bias whereby “people…avoid information that they perceive as potentially unpleasant,” such as skipping a checkup at the doctor’s office for fear of inauspicious news. Likewise, the study conducted “scenario training.” Consider Mellers et al.’s description of scenario training: “Scenario training taught forecasters to generate new futures, actively entertain more possibilities, use decision trees, and avoid biases such as overpredicting change, creating incoherent scenarios, or assigning probabilities exceeding 1.0 to mutually exclusive and exhaustive outcomes” (2014, 1107). Suffice it to say, cognitive debiasing training, scenario training, and probabilistic reasoning training are not the same as DEI training. Nowhere does it appear that the Good Judgment Project gave something akin to DEI training to the Project’s participants.
To sum up this review of the argumentation that has been given for pushing DEI diversity in the IC as a performance-enhancing tool, we see that it is nothing more than bald assertions by officials such as Clapper, Brennan, and Blinken, and the article by Medina (2016), drawing on work such as Phillips (2014). Given the longstanding U.S. intelligence norms and policies that require that evidence supports intelligence analyses, it is striking that no U.S. intelligence officer has ever credibly explained that there is a problem to be fixed, and why the fix is DEI. None have explained why we should think that promoting of DEI would improve the performance of U.S. intelligence agencies or the IC as a whole.
The reason such casual reasoning is used to make such sweeping claims can only be ideology and politics. That important point said, it is crystal clear that capable people of all demographic groups can be, and are, effective intelligence officers. The failure to justify the DEI agenda in the IC arguably violates the law and established government policy, including section 1019 of the Intelligence Reform and Terrorism Prevention Act of 2004 and Intelligence Community Directive 203 “Analytic Standards,” first issued in 2007, which mandates that all intelligence analyses be supported by factual evidence to the extent possible and is a policy directive of the director of national intelligence.ICD 203, third edition, December 2022 (link).
Now, I put aside the semblance of argumentation for DEI and offer some argumentation of my own on the fundamental question, namely: Is the DEI agenda good for the IC?
Damage caused by DEI policies
Politically driven advocates of DEI not only fail to understand that domestically demographic diversity does not improve the performance of foreign-focused intelligence services, they have significantly damaged the operational performance of the agencies. “Operational” means characteristics that affect mission accomplishment of core tasks such as collection of valuable information and conversion of it, via analysis, into intelligence.
My study of the impact of DEI policies on the operational performance of IC agencies, exposited more fully in Gentry (2023a; 2023b; 2024a), concludes that DEI policies damage U.S. intelligence in at least five major, related ways: (1) politicization; (2) damage to collegiality; (3) damage to collection; (4) damage to analysis; and (5) reduced confidence in the impartiality and value of intelligence (Gentry 2024a). These build on each other, leading to serious damage to the ways in which intelligence is conducted, and thence to the operational performance of the agencies. This damage is similar in origins, although somewhat different in effects, to the DEI-caused damage other observers have found in the Defense Department (Lohmeier et al. 2024; Cauthen 2024) and the Federal Bureau of Investigation (A National Alliance 2023, 37–52, 67–69, 80–96).
Before turning to these five aspects of DEI-induced dysfunctions, it is appropriate to return to the question of methods and the special sensitivities and difficulties of researching performance and effectiveness in intelligence work. My research is based in part on interviews with several dozen former and some currently serving intelligence officers from throughout the IC, but most prominently from the CIA. After considering the claims of Clapper, Medina, and others and finding them unsatisfying, I faced the challenge of designing a research effort that would lead to persuasive conclusions about the actual effects of DEI policies on the IC’s operational performance. I decided that a process-tracing approach would be effective and relied on my many contacts in the IC and among former intelligence officers as well as publicly available information sources to provide relevant DEI-related evidence. Some of my contacts were willing to talk on the record, but most would talk only anonymously, often for well-justified fears of retribution if their cooperation with my project were known. I worked alone, as opposed to in a team, speaking to people one-to-one except in a few occasions. The research effort (Gentry 2023b; 2024a) supports the conclusion that DEI is damaging functional effectiveness, and the damage has occurred in five areas or aspects, summarized briefly here.
Politicization
Above, in the background discussion, I provided an account of growing politicization, arguing that DEI is a form of identity politics. It is politics in a domain that, by norms that in the past were widely accepted, should not be politicized, both by not playing partisan domestic politics and not injecting political views into intelligence analyses. Historically, the IC strongly emphasized apolitical public service. Because the impetus for DEI is widely perceived to be political, the policies themselves are widely perceived to insert a political motivation into day-to-day procedures and activities. Politicization permeates and exacerbates the following four headings discussed.
Importantly and tangibly, senior IC leaders, especially CIA director Brennan, invoked the possible threat to DEI policies that candidate and then President Donald Trump posed, to encourage political activism by intelligence officers, which among serving intelligence officers primarily took the form of leaks of both accurate and purposefully incorrect information (known as disinformation) designed to damage Trump and his administration (Gentry 2023b, 158–160, 253). This was a massive violation of the longstanding CIA ethic of apolitical public service (Gentry 2020).
Damage to collegiality
The heavy-handed orthodoxy of DEI is causing significant self-censorship by government personnel who do not support the DEI agenda, just as it is in academia. For example, a now retired very senior CIA manager of analysts wrote while he was still working:
As the workforce has gotten larger, so has diversity. Political correctness rules. With the increased political divide, it is hard to be outspoken. We choose our close colleagues carefully. (anonymous, quoted in Gentry 2023b, 412–413)
An immediate consequence of such divisions is that teamwork is damaged. Employees do not trust each other the way they once did. A State Department official noted that employees are now divided in a new way—people who accept Biden’s DEI agenda, an even stronger version of Obama’s, and those who do not. Due to the heightened DEI-induced politicization of agencies’ workspaces, people “look over their shoulders” and worry about perceptions of “compliance with ideological dictates.” An analyst at a major IC agency described himself as a “coward” for not speaking out against DEI and what he sees as a related rise of Marxist influences in academic intelligence studies. He is in fact a very decent person who has shown great courage in critically but professionally assessing intelligence organizations on many topics, but as a mid-career professional with a family he is understandably careful concerning DEI given its political and bureaucratic toxicity and the widely reported intimidation that surrounds it. Still others say quietly that they oppose the prominent support for LGBTQ+ issues on moral or theological grounds but say nothing; they too hide their beliefs in environments of intolerant ideological orthodoxy (Gentry 2024b).
According to a retired senior female CIA operations officer, the preferential treatment for women in hiring and promotions that Brennan initiated is controversial and divisive within the CIA. While some women like it, men feel discriminated against, and many senior women who succeeded in traditional ways, such as herself, feel their accomplishments and reputations are diminished (Gentry 2023b, 149). There are chronic hints, she said, that people wonder if senior women actually earned their promotions, as opposed to being given them for political reasons. There is no way such suspicions can be definitively disabused.
Similarly, a retired senior manager of analysts who retains close ties to many CIA people said white men at the CIA feel institutionally discriminated against. He estimated in 2019 that human resources offices’ staffs were about 80 percent female (Gentry 2024b). Indeed, at most IC elements, diversity and personnel offices were and continue to be staffed overwhelmingly by women and minorities, many of whom explicitly state their personal gender- and race-oriented political agendas. A common pejorative quip that is not politically incorrect, given the shifting ideological tone of the federal workforce, is that federal employees collectively are still “too male, pale, and stale.” Brennan, in his memoir noted that some CIA personnel did not like some of his policies (Brennan 2020, 286–287), but he did not acknowledge how they have created the deep divisiveness and destroyed trust.
Even former CIA operations officer Marc Polymeropoulos, who achieved considerable infamy for co-writing the open letter, signed by 51 former intelligence officers, that deceptively insinuated in October 2020 that the contents of Hunter Biden’s abandoned laptop computer, which contained emails suggesting that he and his father had engaged in political corruption, was Russian disinformation,On this see, e.g., House of Representatives Judiciary Committee 2023. sees a problem. Polymeropoulos (2024) laments that cultural changes have taken place among new operations officers that generate work habit differences and “woke” political attitudes sufficiently different from long-time perspectives as to warrant “counseling” about proper, disciplined attitudes and activities of operations officers. Polymeropoulos echoes comments of former chief of CIA counterintelligence Mark Kelton (2024), who observed that widespread worry within the CIA even in 2015, when Brennan was director, about the “fraying professional discipline” of the workforce turned into a “tsunami” of leaks in the Trump years. A longtime, currently serving intelligence analyst at an agency other than the CIA understated things when he said in 2024: “DEI policies have been a distraction from our core mission.” Another, senior officer at an agency other than the CIA assessed the operational implications of DEI policies concisely by saying “it negatively affects our mission” (Gentry 2024b).
Damage to human intelligence collection
Numerous current and former CIA operations officers have described many ways in which DEI policies have hampered field operations. These include weaker recruits, bad management, tolerance of weak operational performance in the name of protecting diversity, and the ideological divisions described above (for details, see Gentry 2024a).
Damage to analysis
Biases of many sorts creep into organizations and their leadership teams for reasons wholly independent of the demographic compositions of their members. For this reason, teams at different organizations in different countries often view things very differently, sometimes generating analytic errors. Such biases sometimes generate major intelligence failures. For example, Luke Benjamin Wells (2017) has shown how British and American intelligence officers, working from virtually identical raw data, but making different collective assumptions about Soviet intentions concerning Britain and the USA, respectively, reached very different conclusions in the 1950s about the size and purpose of Soviet strategic bombing forces. For example, the Americans worried that the Soviets were primarily targeting the United States, a fear the British did not have, leading them to overestimate Soviet bombers’ capacities to reach North America. In addition, U.S. Air Force officers evidently wanted to emphasize the Soviet threat to help rationalize their requests for larger budgets. Similarly, former chairman of the National Intelligence Council John Gannon (1997–2001) has recounted how he periodically was frustrated by the close-mindedness of analytic offices in the IC composed of personnel of many demographic identity groups and asked his national intelligence officer for warning, Robert Vickers (1996–2004), to both prod line units and provide alternative analyses (Gentry and Gordon 2019, 175, 220). As noted, former CIA officer Richards Heuer documented the importance of cognitive biases in analysis by both individuals and groups and offered ways to minimize them (Heuer 1999).
Heuer mentioned “motivated biases” like DEI only in passing, presumably because when he wrote in the 1970s to 1990s he knew that the agency’s quality control mechanisms then successfully excised most such biases (Gentry 2016). But times change and things do not always improve. Important biases in analysis appeared in the Obama years (Gentry 2023a, 376–379). DEI-related managerial problems of the sort afflicting CIA operators also affect the analysis directorate, albeit in different ways. While the CIA history staff has been good about recounting analytic successes and failures of the distant past, nothing in the public sphere addresses the implications for analysis of DEI policies or the organizational cultural changes they have wrought. It seems to be a taboo subject. A major purpose of DEI policies has been to alter the organizational cultures and resultant activities of IC agencies.
Reduced confidence in intelligence
The political activism of intelligence officers—overtly, via leaks, and by tailoring analysis in partisan ways—has generated polling that shows that Americans think increasingly poorly of intelligence, especially of the political activism of intelligence officers. These include perceptions of intelligence by its users, including decisionmakers. No small number of Republicans, especially, think the IC tried to conduct a “coup” against former President Trump (Chaffetz 2018). Intelligence that is not believed or trusted is not used operationally, damaging the perceived usefulness of intelligence. A lack of use of good intelligence is likely to damage the quality of national decision-making.
The IC’s main clients, or ‘customers’ in current parlance, are senior government decision-makers, but the intelligence agencies rely on taxpayer funding and ought to serve the public by improving national security decision-making. The agencies have long sought to both inform citizens in general terms about a secretive part of their government and to woo public opinion as an aid to furthering their own parochial interests (Moran 2016). Another operational result of DEI-motivated political activism is diminished confidence in the trustworthiness of the agencies, which means diminished faith in the accuracy and integrity of intelligence assessments that presidents and other senior decisionmakers use to make and rationalize their foreign policy judgments. Given the strong evidentiary basis for politicization and worse, it is no surprise that public confidence in intelligence is declining (Gentry 2024c). A Rasmussen poll released in October 2023 found that only 36 percent of American voters believed that intelligence agencies behaved in an impartial manner, while 51 percent said the agencies have their own political agendas, and 65 percent believed it likely that the agencies are influencing corporate media’s coverage of political issues (link). Another Rasmussen poll released in March 2024 showed that most Americans think the IC is trying to influence the 2024 presidential election (link).
Conclusions
Evidence is strong that the strikingly casual, superficial assertions about the performance implications of DEI that emerged in the IC after Obama’s perspectives on “diversity and inclusion” policies became U.S. government policies are little more than attempts to rationalize a political agenda. The weakness of these arguments is striking when compared to the generally much more sophisticated language of actual intelligence analyses. And, since Medina wrote, no other prominent current or former intelligence officer has tried to explain how and why demographic diversity improves the performance of American intelligence agencies. This political orthodoxy continues to cause and mask the degradation of U.S. intelligence. We now can see much damage, but not yet their full consequences. U.S. adversaries undoubtedly also see the problems noted herein and will exploit such weaknesses and vulnerabilities at times of their choosing.