At a casual glance, the recent cascades of American disasters might seem unrelated. In a span of fewer than six months in 2017, three U.S. Naval warships experienced three separate collisions resulting in 17 deaths. A year later, powerlines owned by PG&E started a wildfire that killed 85 people. The pipeline carrying almost half of the East Coast’s gasoline shut down due to a ransomware attack. Almost half a million intermodal containers sat on cargo ships unable to dock at Los Angeles ports. A train carrying thousands of tons of hazardous and flammable chemicals derailed near East Palestine, Ohio. Air Traffic Control cleared a FedEx plane to land on a runway occupied by a Southwest plane preparing to take off. Eye drops contaminated with antibiotic-resistant bacteria killed four and blinded fourteen. 

While disasters like these are often front-page news, the broader connection between the disasters barely elicits any mention. America must be understood as a system of interwoven systems; the healthcare system sends a bill to a patient using the postal system, and that patient uses the mobile phone system to pay the bill with a credit card issued by the banking system. All these systems must be assumed to work for anyone to make even simple decisions. But the failure of one system has cascading consequences for all of the adjacent systems. As a consequence of escalating rates of failure, America’s complex systems are slowly collapsing.

The core issue is that changing political mores have established the systematic promotion of the unqualified and sidelining of the competent. This has continually weakened our society’s ability to manage modern systems. At its inception, it represented a break from the trend of the 1920s to the 1960s, when the direct meritocratic evaluation of competence became the norm across vast swaths of American society. 

In the first decades of the twentieth century, the idea that individuals should be systematically evaluated and selected based on their ability rather than wealth, class, or political connections, led to significant changes in selection techniques at all levels of American society. The Scholastic Aptitude Test (SAT) revolutionized college admissions by allowing elite universities to find and recruit talented students from beyond the boarding schools of New England. Following the adoption of the SAT, aptitude tests such as Wonderlic (1936), Graduate Record Examination (1936), Army General Classification Test (1941), and Law School Admission Test (1948) swept the United States. Spurred on by the demands of two world wars, this system of institutional management electrified the Tennessee Valley, created the first atom bomb, invented the transistor, and put a man on the moon. 

By the 1960s, the systematic selection for competence came into direct conflict with the political imperatives of the civil rights movement. During the period from 1961 to 1972, a series of Supreme Court rulings, executive orders, and laws—most critically, the Civil Rights Act of 1964—put meritocracy and the new political imperative of protected-group diversity on a collision course. Administrative law judges have accepted statistically observable disparities in outcomes between groups as prima facie evidence of illegal discrimination. The result has been clear: any time meritocracy and diversity come into direct conflict, diversity must take priority. 

The resulting norms have steadily eroded institutional competency, causing America’s complex systems to fail with increasing regularity. In the language of a systems theorist, by decreasing the competency of the actors within the system, formerly stable systems have begun to experience normal accidents at a rate that is faster than the system can adapt. The prognosis is harsh but clear: either selection for competence will return or America will experience devolution to more primitive forms of civilization and loss of geopolitical power.

From Meritocracy to Diversity

The first domino to fall as civil rights-era policies took effect was the quantitative evaluation of competency by employers using straightforward cognitive batteries. While some tests are still legally used in hiring today, several high-profile enforcement actions against employers caused a wholesale change in the tools customarily usable by employers to screen for ability. 

After the early 1970s, employers responded by shifting from directly testing for ability to using the next best thing: a degree from a highly-selective university. By pushing the selection challenge to the college admissions offices, selective employers did two things: they reduced their risk of lawsuits and they turned the U.S. college application process into a high-stakes war of all against all. Admission to Harvard would be a golden ticket to join the professional managerial class, while mere admission to a state school could mean a struggle to remain in the middle class.

This outsourcing did not stave off the ideological change for long. Within the system of political imperatives now dominant in all major U.S. organizations, diversity must be prioritized even if there is a price in competency. The definition of diversity varies by industry and geography. In elite universities, diversity means black, indigenous, or Hispanic. In California, Indian women are diverse but Indian men are not. When selecting corporate board members, diversity means “anyone who is not a straight white man.” The legally protected and politically enforced nature of this imperative renders an open dialogue nearly impossible. 

However diversity itself is defined, most policy on the matter is based on a simple premise: since all groups are identical in talent, any unbiased process must produce the same group proportions as the general population, and therefore, processes that produce disproportionate outcomes must be biased. Prestigious journals like Harvard Business Review are the first to summarize and parrot these views, which then flow down to reporting by mass media organizations like Bloomberg Businessweek. Soon, it joins McKinsey’s “best practices” list and becomes instantiated in corporate policies. 

Unlike accounting policies, which emanate from the Financial Accounting Standards Board and are then implemented by Chief Financial Officers, the diversity push emanates inside of organizations from multiple power centers, each of which joins in for independent reasons. CEOs push diversity policies primarily to please board members and increase their status. Human Resources (HR) professionals push diversity policies primarily to avoid anti-discrimination lawsuits. Business development teams push diversity to win additional business from diversity-sensitive clients (e.g. government agencies). Employee Resource Groups (ERGs), such as the Black Googler Network, push diversity to help their in-group in hiring and promotion decisions.

Diversity in Theory and Practice

In police academies around the country, new recruits are taught to apply an escalation of force algorithm with non-compliant subjects: “Ask, Tell, Make.” The idea behind “Ask, Tell, Make” is to apply the least amount of force necessary to achieve the desired level of compliance. This is the means by which police power, which is ultimately backed by significant coercive force, can maintain an appearance of voluntary compliance and soft-handedness. Similarly, the power centers inside U.S. institutions apply a variant of “Ask, Tell, Make” to achieve diversity in their respective organizations.

The first tactics for implementing diversity imperatives are the “Ask” tactics. These simply ask all the members of the organization to end bias. At this stage, the policies seem so reasonable and fair that there will rarely be much pushback. Best practices such as slating guidelines are a common tool at this stage. Slating guidelines require that every hiring process must include a certain number and type of diverse candidates for every job opening. Structured interviews are another best practice that requires interviewers to stick with a script to minimize the chance of uncovering commonalities between the interviewer and interviewee that might introduce bias. Often HR will become involved in the hiring process, specifically asking the hiring manager to defend their choice not to hire a diverse candidate. Because the wrong answer could result in shaming, loss of advancement opportunities, or even termination, the hiring manager can often be persuaded to prioritize diversity over competence.

Within specialized professional services companies, senior-level recruiting will occasionally result in a resume collection where not a single diverse candidate meets the minimum specifications of the job. This is a terrible outcome for the hiring manager as it attracts negative attention from HR. At this point, firms will often retain an executive search agency that focuses on exclusively diverse candidates. When that does not result in sufficient diversity, roles will often have their requirements diluted to increase the pool of diverse candidates. 

For example, within hedge funds, the ideal entry-level candidate might be an experienced former investment banker who went to a top MBA program. This preferred pedigree sets a minimum bar for both competence and work ethic. This first-pass filter enormously winnows the field of underrepresented candidates. To relax requirements for diversity’s sake, this will be diluted in various ways. First, the work experience might be stripped. Next, the role gets offered to MBA interns. Finally, fresh undergraduates are hired into the analyst role. Dilution works not just because of the larger field of candidates it allows for but also because the Harvard Admission Office of 2019 is even more focused on certain kinds of diversity than the Harvard Admission Office of 2011 was. 

This dilution is not costless; fewer data points result in a wider range of outcomes and increase the risk of a bad hire. All bad hires are costly but bad hires that are diverse are even worse. The risk of a wrongful termination lawsuit either draws out the termination process for diverse hires or results in the firm adjusting by giving them harmless busy work until they leave of their own volition—either way, a terrible outcome for the organizations which hired them.

If these “Ask” tactics do not achieve enough diversity, the next step in the escalation is to attach carrots and sticks to directly tell decision-makers to increase the diversity of the organization. This is the point at which the goals of diversity and competence truly begin displaying significant tension between each other. The first step is the implementation of Key Performance Indicators (KPI) linked to diversity for all managers. Diversity KPIs are a tool to embarrass leaders and teams that are not meeting their diversity targets. Given that most organizations are hierarchical and pyramidal, combined with the fact that America was much whiter 50 years ago than it was today, it is unsurprising that senior leadership teams are less diverse than America as a whole—and, more pertinently, than their own junior teams. 

The combination of a pyramid-shaped org chart and a senior leadership team where white men often make up 80 percent or more of the team means that the imposition of an aggressive KPI sends a message to the layer below them: no white man in middle management will likely ever see a promotion as long as they remain in the organization. This is never expressed verbally. Rather, those overlooked figure it out as they are passed over continually for less competent but more diverse colleagues. The result is demoralization, disengagement, and over time, departure. 

While all the aforementioned techniques fall into the broad category of affirmative action, they primarily result in slightly tilting the scale toward diverse candidates. The next step is simply holding different groups to different standards. Within academia, the recently filed Students for Fair Admissions v. President and Fellows of Harvard College lawsuit leveraged data to show the extent to which Harvard penalizes Asian and white applicants to help black and Hispanic applicants. The UC System, despite formally being forbidden from practicing affirmative action by Proposition 209, uses a tool called “comprehensive admission” to accomplish the same goal.

The latest technique, which was recently brought to light, shows UC admissions offices using the applicants’ high schools as a proxy for race to achieve their desired goal. Heavily Asian high schools such as Arcadia—which is 68 percent Asian—saw their UC-San Diego acceptance rate cut from 37 percent to 13 percent while the 99-percent-Hispanic Garfield High School saw its UC-San Diego acceptance rate rise from 29 percent to 65 percent.

The preference for diversity at the college faculty level is similarly strong. Jessica Nordell’s End of Bias: A Beginning heralded MIT’s efforts to increase the gender diversity of its engineering department: “When applications came in, the Dean of Engineering personally reviewed every one from a woman. If departments turned down a good candidate, they had to explain why.”

When this was not enough, MIT increased its gender diversity by simply offering jobs to previously rejected female candidates. While no university will admit to letting standards slip for the sake of diversity, no one has offered a serious argument why the new processes produce higher or even equivalent quality faculty as opposed to simply more diverse faculty. The extreme preference for diversity in academia today explains much of the phenomenon of professors identifying with a minor fraction of their ancestry or even making it up entirely.

During COVID-19, the difficulty of in-person testing and online proctoring created a new mechanism to push diversity at the expense of competency: the gradual but systematic elimination of standardized tests as a barrier to admission to universities and graduate schools. Today, the majority of U.S. colleges have either stopped requiring SAT/ACT scores, no longer require them for students in the top 10 percent of their class, or will no longer consider them. Several elite law schools, including Harvard Law School, no longer require the LSAT as of 2023. With thousands of unqualified law students headed to a bar exam that they are unlikely to pass, the National Conference of Bar Examiners is already planning to dilute the bar exam under the “NextGen” plan. Specifically, “eliminat[ing] any aspects of our exams that could contribute to performance disparities” will almost definitionally reduce the degree to which the exam tests for competency.

Similarly, standards used to select doctors have also been weakened to promote diversity. Programs such as the City College of New York’s BS/MD program have eliminated the MCAT requirement. With the SAT now optional, new candidates can go straight from high school to the United States Medical Licensing Examination Step 1 exam in medical school without having gone through any rigorous standardized test whose score can be compared across schools. Step 1 scores were historically the most significant factor in the National Residency Matching Program, which pairs soon-to-be doctors with their future residency training programs. Because Step 1 scores serve as a barrier to increasing diversity, they have been made pass/fail. A handful of doctors are speaking out about the dangers of picking doctors based on factors other than competency but most either explicitly prefer diversity or else stay silent, concerned about the career-ending repercussions of pointing out the obvious. 

When even carrot and stick incentives and the removal of standards do not achieve enough diversity, the end game is to simply make decision-makers comply. “Make” has two preferred implementations: one is widely discussed and the other is, for obvious reasons, never disclosed publicly. The first method of implementation is the application of quotas. Quotas or set-asides require the reservation of admissions slots, jobs, contracts, board seats, or other scarce goods for women and members of favored minority groups. Government contracts and supplier agreements are explicitly awarded to firms that have acronyms such as SB, WBE, MBE, DBE, SDB, VOSB, SDVOSB, WOSB, HUB, and 8(a). 

Within large employers and government contractors, quotas are used for both hiring and promotions, requiring specific percentages of hiring or promotions to be reserved for favored groups. During the summer of 2020, the CEO of Wells Fargo was publicly shamed after his memo blaming the underrepresentation of black senior leaders on a “very limited pool” of black talent was leaked to Reuters. Less than a month later, the bank publicly pledged to reserve 12 percent of leadership positions for black candidates and began tying executive compensation to reaching diversity goals. In 2022, Goldman Sachs extended quotas to the capital markets by adopting a policy to avoid underwriting IPOs of firms without at least two board members that are not straight white men.

When diversity still refuses to rise to acceptable levels, the remaining solution is the direct exclusion of non-diverse candidates. While public support for anti-discrimination laws and equal opportunity laws is high, public support for affirmative action and quotas is decidedly mixed. Hardline views such as those expressed in author Ijeoma Oluo’s Mediocre: The Dangerous Legacy of White Male America—namely that “any white man in a position of power perpetuates a system of white male domination”—are still considered extreme, even within U.S. progressive circles. 

As such, when explicit exclusion is used to eliminate groups like white men from selection processes, it is done subtly. Managers are told to sequester all the resumes from “non-diverse” candidates—that is, white males. These resumes are discarded and the candidates are sent emails politely telling them that “other candidates were a better fit.” While some so-called “reverse discrimination” lawsuits have been filed, most of these policies go unreported. The reasons are straightforward; even in 2023, screening out all white men is not de jure legal. Moreover, any member of the professional managerial class who witnesses and reports discrimination against white men will never work in their field again. 

Even anonymous whistleblowing is likely to be rare. To imagine why, suppose incontrovertible evidence was produced that one’s employer was explicitly excluding white male candidates, and a lawsuit was filed. The employer’s reputation and the reputation of all the employees there, including the white men still working there, would be tarnished. That said, we can expect to see more lawsuits from men who feel they have little to lose.

This “Ask, Tell, Make” framework, under various descriptions, is the method by which individuals with a vested interest in more diversity push their organizations toward their preferred outcome. Force begins requesting modest changes to recruiting to make it “more fair.” Force ends with the heavy-handed application of quotas and even exclusion. The American system is not a monolith, however, which means that the strength of the push and its effects on competency is not distributed evenly.

Competency Is Declining From the Core Outwards

Think of the American system as a series of concentric rings with the government at the center. Directly surrounding that are the organizations that receive government funds, then the nonprofits that influence and are subject to policy, and finally business at the periphery. Since the era of the Manhattan Project and the Space Race, the state capacity of the federal government has been declining almost monotonically. 

While this has occurred for a multitude of reasons, the steel girders supporting the competency of the federal government were the first to be exposed to the saltwater of the Civil Rights Act and related executive orders. Government agencies, which are in charge of overseeing all the other systems, have seen the quality of their human capital decline tremendously since the 1960s. While the damage to an agency like the Department of Agriculture may have long-term deadly consequences, the most immediate danger is at safety-critical agencies like the Federal Aviation Administration (FAA). 

The Air Traffic Control (ATC) system used in the U.S. relies on an intricate dance of visual or radar observation, transponders, and radio communication, all with the incredible challenge of keeping thousands of simultaneously moving planes from ever crashing into each other. Since air controlling is one of the only jobs that pays more than $100,000 per year and does not require a college diploma, it has been a popular career choice for individuals without a degree who nonetheless have an exceptionally good memory, attention span, visuospatial awareness, and logical skills. The Air Traffic Selection and Training (AT-SAT) Exam, a standardized test of those critical skills, was historically the primary barrier to entry for air controllers. As a consequence of the AT-SAT, as well as a preference for veterans with former air controller experience, 83 percent of air controllers in the U.S. were white men as of 2014. 

That year, the FAA added a Biographical Questionnaire (BQ) to the screening process to tilt the applicant pool toward diverse candidates. Facing pushback in the courts from well-qualified candidates who were screened out, the FAA quietly backed away from the BQ and adopted a new exam, the Air Traffic Skills Assessment (ATSA). While the ATSA includes some questions similar to those of the BQ, it restored the test’s focus on core air traffic skills. The importance of highly-skilled air controllers was made clear in the most deadly air disaster in history, the 1977 Tenerife incident. Two planes, one taking off and one taxiing, collided on the runway due to confusion between the captain of KLM 4805 and the Tenerife ATC. The crash, which killed 583 people, resulted in sweeping changes in aviation safety culture. 

Recently, the tremendous U.S. record for air safety established since the 1970s has been fraying at the edges. The first three months of 2023 saw nine near-miss incidents at U.S. airports, one with two planes coming within 100 feet of colliding. This terrifying uptick from years prior resulted in the FAA and NTSB convening safety summits in March and May, respectively. Whether they dared to discuss root causes seems unlikely.

Given the sheer size of the U.S. military in both manpower and budget dollars, it should not come as a surprise that the diversity push has also affected the readiness of this institution. Following three completely avoidable collisions of U.S. Navy warships in 2017 and a fire in 2020 that resulted in the scuttling of USS Bonhomme Richard, a $750 million amphibious assault craft, two retired marines conducted off-the-record interviews with 77 current and retired Navy officers. One recurring theme was the prioritization of diversity training over ship handling and warfighting preparedness. Many of them openly admit that, given current issues, the U.S. would likely lose an open naval engagement with China. Instead of taking the criticism to heart, the Navy commissioned “Task Force One Navy,” which recommended deemphasizing or eliminating meritocratic tests like the Officer Aptitude Rating to boost diversity. Absent an existential challenge, U.S. military preparedness is likely to continue to degrade. 

The decline in the capacity of government contractors is likewise obvious, with the largest contractors being the most directly impacted. The five largest contractors—Lockheed Martin, Boeing, General Dynamics, Raytheon Company, and Northrop Grumman—will all struggle to maintain competency in the coming years. 

Boeing, one of only two firms globally capable of mass-producing large airliners, has a particularly striking crisis unfolding in its institutional culture. Shortly after releasing the 737 MAX, 346 people died in two nearly identical 737 MAX crashes in Indonesia and Ethiopia. The cause of the crashes was a complex interaction between design choices, cost-cutting led by MBAs, FAA issues, the MCAS flight-control system, a faulty sensor, and pilot training. Meanwhile, on the defense side of the business, Boeing’s new fuel tanker, the KC-46A Pegasus is years behind on deliveries due to serious technical flaws with the fueling system along with multiple cases of Foreign Object Debris left inside the plane during construction: tools, a red plastic cap, and in one case, even trash. Between the issues at ATC and Boeing, damage to the U.S.’s phenomenal aviation safety record seems almost inevitable.

After government contractors, the next-most-affected class of institutions are nonprofit organizations. They are entrapped by the government whose policies they are subject to and trying to influence, the opinions of their donor base, and lack of any profit motive. The lifeblood of nonprofits is access to capital, either directly in the form of government grants or through donations that are deemed tax-deductible. Accessing federal monies means being subject to the full weight of U.S. diversity rules and regulations. Nonprofits are generally governed by boards whose members tend to overlap with the list of major donors. Because advocacy for diversity and board memberships are both high-status positions, unsurprisingly board members tend to voice favorable opinions of diversity, and those opinions flow downstream to the organizations they oversee. 

Nonprofits—including universities, charities, and foundations—exist in an overlapping ecosystem with journalism, with individuals tending to freely circulate between the four. The activities of nonprofits are bound up in the same discourses shaped by current news and academic research, with all four reflecting the same general ideological consensus. Finally, lacking the profit motive, the decision-making processes of nonprofits are influenced by what will affect the status of the individuals within those organizations rather than what will affect profits. Within nonprofits, the cost of incompetent staffers is borne by “stakeholders,” rather than any one individual.

While all businesses subject to federal law must prioritize diversity over competency at some level, the problem is worse at publicly-traded corporations for reasons both obvious and subtle. The obvious reason is that larger companies present larger targets for EEOC actions and discrimination lawsuits with hundreds of millions of dollars at stake. Corporations have logically responded by hiring large teams of HR professionals to preempt such lawsuits. Over the past several decades, HR has evolved from simply overseeing onboarding to involvement in every aspect of hiring, promotions, and firings, seeing them all through a political and regulatory lens. 

The more subtle reason for pressure within publicly-traded companies is that they require ongoing relationships with a spiderweb of banks, credit ratings agencies, proxy advisory services, and most importantly, investors. Given that the loss of access to capital is an immediate death sentence for most businesses, the CEOs of publicly-traded companies tend to push diversity over competency even when the decline in firm performance is clear. CEOs would likely rather trade a small drag on profits margins than a potentially career-ending scandal from pushing back. 

Whereas publicly-traded corporations nearly uniformly push diversity, privately-held businesses vary tremendously based on the views of their owners. Partnerships such as the Big Four accounting firms and top-tier management consultancies are high-status. High-status firms must regularly proclaim extensive support for diversity. While the firms tend to be highly selective, partnerships whose leadership is overwhelmingly white and male have generally capitulated to the zeitgeist and are cutting standards to hit targets. Firms often manage around this by hiring for diversity and then putting diversity hires into roles where they are the least likely to damage the firm or the brand. Somewhat counterintuitively, firms with diverse founders are often highly meritocratic, as the structure harnesses the founder’s desire to make money and shields them from criticism on diversity issues.

The most notable example of a diverse meritocracy is Vista Equity Partners, the large private equity firm founded by Robert F. Smith, America’s wealthiest black man. Robert F. Smith is one of the most vocal advocates for and philanthropists to historically black U.S. colleges and universities. It would be reasonable to expect Vista to prioritize diversity over competency in its portfolio companies. However, Vista has instead been profiled for giving all portfolio company management teams the Criteria Cognitive Aptitude Test and ruthlessly culling low-performers. Given the amount of value to be created by promoting the best people into leadership roles of their portfolio companies, one might imagine this to be low-hanging fruit for the rest of private equity, yet Vista is an outlier. Why Vista can apply the CCAT without a public outcry is obvious. 

The other firms that tend to still focus on competency are those that are small and private. Such firms have two key advantages: they fall below the fifteen-employee threshold for the most onerous EEOC rules and the owner can usually directly observe the performance of everyone inside the organization. Within small firms, underperformance is usually obvious. Tech startups, being both small and private, would seem to have the right structure to prioritize competency.

The American System Is Cracking

Promoting diversity over competency does not simply affect new hires and promotion decisions. It also affects the people already working inside of America’s systems. Morale and competency inside U.S. organizations are declining. Those who understand that the new system makes it hard or impossible for them to advance are demoralized, affecting their performance. Even individuals poised to benefit from diversity preferences notice that better people are being passed over and the average quality of their team is declining. High performers want to be on a high-performing team. When the priorities of their organizations shift away from performance, high performers respond negatively.

This effect was likely seen in a recent paper by McDonald, Keeves, and Westphal. The paper points out that white male senior leaders reduce their engagement following the appointment of a minority CEO. While it is possible that author Ijeoma Oluo is correct, and that white men have so much unconscious bias raging inside of them that the appointment of a diverse CEO sends them into a tailspin of resentment, there is another more plausible explanation. When boards choose diverse CEOs to make a political statement, high performers who see an organization shifting away from valuing honest performance respond by disengaging.

Some demoralized employees—like James Damore in his now-famous essay, “Google’s Ideological Echo Chamber”—will directly push back against pro-diversity arguments. Like James, they will be fired. Older, demoralized workers, especially those who are mere years from retirement, are unlikely to point out the decline in competency and risk it costing them their jobs. Those who have a large enough nest egg may simply retire to avoid having to deal with the indignity of having to attend another Inclusive Leadership seminar.

As older men with tacit knowledge either retire or are pushed out, the burden of maintaining America’s complex systems will fall on the young. Lower-performing young men angry at the toxic mix of affirmative action (hurting their chances of admission to a “good school”) and credentialism (limiting the “good jobs” to graduates of “good schools”) are turning their backs on college and white-collar work altogether. 

This is the continuation of a trend that began over a decade ago. High-performing young men will either collaborate, coast, or downshift by leaving high-status employment altogether. Collaborators will embrace “allyship” to attempt to bolster their chances of getting promoted. Coasters realize that they need to work just slightly harder than the worst individual on their team. Their shirking is likely to go unnoticed and they are unlikely to feel enough emotional connection to the organization to raise alarm when critical mistakes are being made. The combination of new employees hired for diversity, not competence, and the declining engagement of the highly competent sets the stage for failures of increasing frequency and magnitude.

The modern U.S. is a system of systems interacting together in intricate ways. All these complex systems are simply assumed to work. In February of 2021, cold weather in Texas caused shutdowns at unwinterized natural gas power plants. The failure rippled through the systems with interlocking dependencies. As a result, 246 people died. In straightforward work, declining competency means that things happen more slowly, and products are lower quality or more expensive. In complex systems, declining competency results in catastrophic failures. 

To understand why, one must understand the concept of a “normal accident.” In 1984, Charles Perrow, a Yale sociologist, published the book, Normal Accidents: Living With High-Risk Technologies. In this book, Perrow lays out the theory of normal accidents: when you have systems that are both complex and tightly coupled, catastrophic failures are unavoidable and cannot simply be designed around. In this context, a complex system is one that has many components that all need to interact in a specified way to produce the desired outcome. Complex systems often have relationships that are nonlinear and contain feedback loops. Tightly-coupled systems are those whose components need to move together precisely or in a precise sequence. 

The 1979 Three Mile Island Accident was used as a case study: a relatively minor blockage of a water filter led to a cascading series of malfunctions that culminated in a partial meltdown. In A Demon of Our Own Design, author Richard Bookstaber added two key contributions to Perrow’s theory: first, that it applies to financial markets, and second, that regulation intended to fix the problem may make it worse. 

The biggest shortcoming of the theory is that it takes competency as a given. The idea that competent organizations can devolve to a level where the risk of normal accidents becomes unacceptably high is barely addressed. In other words, rather than being taken as absolutes, complexity and tightness should be understood to be relative to the functionality of the people and systems that are managing them. The U.S. has embraced a novel question: what happens when the men who built the complex systems our society relies on cease contributing and are replaced by people who were chosen for reasons other than competency? 

The answer is clear: catastrophic normal accidents will happen with increasing regularity. While each failure is officially seen as a separate issue to be fixed with small patches, the reality is that the whole system is seeing failures at an accelerating rate, which will lead in turn to the failure of other systems. In the case of the Camp Fire that killed 85 people, PG&E fired its CEO, filed Chapter 11, and restructured. The system’s response has been to turn off the electricity and raise wildfire insurance premiums. This has resulted in very little reflection. The more recent coronavirus pandemic was another teachable moment. What started just three years ago with a novel respiratory virus has caused a financial crisis, a bubble, soaring inflation, and now a banking crisis in rapid succession.

Patching the specific failure mode is simultaneously too slow and induces unexpected consequences. Cascading failures overwhelm the capabilities of the system to react. 20 years ago, a software bug caused a poorly-managed local outage that led to a blackout that knocked out power to 55 million people and caused 100 deaths. Utilities were able to restore power to all 55 million people in only four days. It is unclear if they could do the same today. U.S. cities would look very different if they remained without power for even two weeks, especially if other obstructions unfolded. What if emergency supplies sat on trains immobilized by fuel shortages due to the aforementioned pipeline shutdown? The preference for diversity over competency has made our system of systems dangerously fragile.

Americans living today are the inheritors of systems that created the highest standard of living in human history. Rather than protecting the competency that made those systems possible, the modern preference for diversity has attenuated meritocratic evaluation at all levels of American society. Given the damage already done to competence and morale combined with the natural exodus of baby boomers with decades worth of tacit knowledge, the biggest challenge of the coming decades might simply be maintaining the systems we have today. 

The path of least resistance will be the devolution of complex systems and the reduction in the quality of life that entails. For the typical resident in a second-tier city in Mexico, Brazil, or South Africa, power outages are not uncommon, tap water is probably not safe to drink, and hospital-associated infections are common and often fatal. Absent a step change in the quality of American governance and a renewed culture of excellence, they prefigure the country’s future.

PALLADIUM Magazine is the first magazine of the 21st century. To get our beautiful quarterly print edition, subscribe here.

Demis Hassabis co-founded Google Deep Mind. Carlotta Cardana for WSJ

Alphabet’s top scientist, Demis Hassabis, held discussions with government officials and leaders of other artificial-intelligence labs about forming a new independent industry safety entity in the weeks before he relinquished his role as chief executive of Google DeepMind, according to people familiar with the matter.

Hassabis, who co-founded Google DeepMind and is considered a pioneer in the field of machine learning, has been spending increasing amounts of his time advancing the idea of an organization that would codify safety guardrails and best practices for developing artificial general intelligence, or AGI, these people said. In discussions with peers, he has likened the proposed entity to the International Atomic Energy Agency, a nongovernmental watchdog that promotes cooperation on nuclear energy.

Earlier in the summer, Hassabis discussed aspects of his plan with high-ranking Trump administration officials, including Treasury Secretary Scott Bessent and Michael Kratsios, one of President Trump’s top tech advisers, people familiar with the matter said.

Hassabis didn’t respond to requests for comment. A Google spokesperson said the company sees ensuring the safety of advanced AI models as “vitally important to society” and that Hassabis is uniquely positioned to work toward that goal.

“Demis chose to step into a new role at Google and is focusing his efforts more strategically on scientific breakthroughs and shaping the future of AGI,” the spokesperson said. 

The world’s largest AI labs are locked in a race to reach AGI—computers that can achieve humanlike intelligence—but many of the top minds in tech, including leaders of the companies themselves, have expressed fears that the spread of the technology will disrupt society or that its power will be difficult to control.

Hassabis has discussed the need for the U.S. to lead a globally coordinated effort to establish AI standards. Carlotta Cardana for WSJ

Leaders at both OpenAI and Anthropic have called for establishing new mechanisms that would allow AI companies to coordinate on safety matters. OpenAI CEO Sam Altman last month proposed “a U.S.-led international forum” to determine standards and provide governance over labs, while Anthropic endorsed a global agreement to slow development and verify compliance. Hassabis’s efforts represent some of the most extensive and concrete steps undertaken by an official representing a leading lab.

Recent high-profile instances of AI bots going rogue and carrying out cyberattacks have added to widespread anxieties that the development of the technology is moving too fast and without enough regulation. The Trump administration has increased scrutiny of the security risks posed by AI but kept model evaluations for the most powerful tools voluntary.

In mid-July, Hassabis published an essay on X arguing that the U.S. should create an industry-funded “standards body” to help assess the safety of AI models. The entity would operate like the Financial Industry Regulatory Authority, a self-regulatory body established during the Great Depression that oversees brokerage firms and is funded by the sector, he said.

Under his plan, federal agencies and the Energy Department’s national laboratories would work with AI companies to test models “in areas relevant to national security,” and would establish qualifications that would label a model “frontier-class” if it met certain standards. 

In the weeks leading up to the publication of the essay, Hassabis previewed his idea to Bessent and Kratsios. Bessent has helped shape the administration’s AI strategy, warning that recent models like Anthropic’s Mythos pose cyber threats to the financial system.

Last week, Google announced a dramatic management shake-up at DeepMind, its AI research unit, that saw Hassabis step down as CEO to be replaced by his deputy, Koray Kavukcuoglu. Hassabis was named to the joint roles of chairman of DeepMind and chief scientist of Alphabet GOOGL -0.13%decrease; down pointing triangle.

“I’ve been working towards AGI my whole life and now, like many of you, I feel it is close at hand,” he wrote in a memo. “With this backdrop, I’ve decided that now is the right time for me to hand over my day-to-day operational responsibilities at GDM, so that I have the time and space to focus on the big picture and help influence what is to come to the best of my ability.”

Jeff Dean, the company’s longtime chief scientist and a co-founder of Google Brain who was key to the creation of several of Google’s signature products, stepped down to found an AI-research startup.

In a blog post last week, Alphabet CEO Sundar Pichai wrote that he and Hassabis “have been long discussing a role that allows him to put his full attention on actively shaping the future of AGI.”

Hassabis, a Nobel laureate, discussed the need for the U.S. to lead a globally coordinated effort to establish AI standards at a June Group of Seven summit in France that included Trump, other world leaders and AI executives including Altman and Anthropic CEO Dario Amodei.

Critics of the White House’s approach and Hassabis’s idea say that voluntary efforts and self-regulatory concepts aren’t adequate to address the security and economic threats posed by AI. Many lawmakers are calling for mandatory model reviews and increased government oversight.

Corrections & Amplifications
Jeff Dean was a co-founder of Google Brain. An earlier version of this article incorrectly said he co-founded DeepMind. (Corrected on Aug. 13)

Copyright ©2026 Dow Jones & Company, Inc. All Rights Reserved. 87990cbe856818d5eddac44c7b1cdeb8

Appeared in the August 14, 2026, print edition as 'DeepMind’s Hassabis Pitched AI Safety Unit'.

Cami Clark and Anthropic CEO Dario Amodei at a summit in New Delhi in February. Ludovic Marin/AFP/Getty Images

When Indian Prime Minister Narendra Modi invited AI leaders to a meeting in New Delhi earlier this year, security protocols allowed each executive to bring one additional person with them. Most brought colleagues, but Anthropic CEO Dario Amodei brought his wife, Cami Clark.

Clark doesn’t work at Anthropic, but she is often seen sitting in the front row while Amodei talks at events such as Davos or can be found chatting up investors at gatherings such as the Allen & Co. conference in Sun Valley.

She acts as a sounding board and strategic adviser for Amodei, according to people close to the company. She also brought Anthropic a key early investor, former Google CEO Eric Schmidt—whom she had dated—as it was getting off the ground in early 2021, some of the people said. 

Despite her influence, there are scant details about Clark online​—and efforts have been made to remove references to her, according to a Wall Street Journal analysis and a person familiar with the matter.

The pair married in 2022, but Amodei’s Wikipedia page didn’t say he was married until this summer, and still doesn’t say to whom. Searches for “Dario Amodei’s wife” on Google often turn up a photograph of his sister, Daniela Amodei, who helps run the company.

Even Claude, Anthropic’s AI chatbot, responds to queries by saying “Dario Amodei’s marital status doesn’t seem to be clearly confirmed.”

Amodei spoke at a Wall Street Journal event at Davos this year. Clark watched from the audience. Maurizio Martorana for WSJ

As Anthropic hurtles toward an IPO that could top $2 trillion as soon as this fall, Clark is one of the most influential voices shaping the decisions of a CEO at the forefront of the AI industry—a revolution remaking all aspects of work and life that simultaneously thrills and terrifies most of the public. 

It’s a remarkable rise for a woman who didn’t graduate from college, had a previous brief marriage, at age 20, to a man more than 40 years older and once declared bankruptcy. Her career has included co-founding a women-focused “free luxury porn” company that she unsuccessfully tried to persuade convicted sex offender Jeffrey Epstein to invest in, and starting a women’s “social dieting app” that morphed into a women’s healthcare AI company. 

She made connections with powerful people, including Schmidt, whom she brought into Amodei’s orbit. The billionaire left his roles at Google and parent Alphabet in 2020 and transitioned into investing in tech startups, among other things. She also tried to persuade Schmidt to back a venture fund she would run with him to invest in Anthropic and other AI companies, although that didn’t move forward.​ 

Claude, the company’s chatbot, is now the chief rival of OpenAI’s ChatGPT, and its Claude Code tool has been embraced by millions of engineers and everyday consumers for tasks such as software writing and data analysis. 

The company’s powerful Mythos model, capable of finding and exploiting vulnerabilities, rattled markets and stoked calls by some White House officials for more federal oversight of new AI models, and Amodei himself has been a prominent voice in warning the public about the safety risks of AI.

Early partner

Clark, born in Reno, Nev., in 1979, was an entrepreneur and often interacted with the rich and powerful, as well as up-and-comers in Silicon Valley.

The most important would turn out to be Amodei, whom she started dating in 2014. Just previously, beginning in 2011 and lasting for three years, she was in a relationship with Schmidt, who was then executive chairman of Google.

Clark spent time at the group house in San Francisco where Amodei, his friend Holden Karnofsky and his sister, Daniela—who would later marry Karnofsky—lived with other key figures from the AI safety and Effective Altruism worlds. The movement, centered largely in Silicon Valley, aims to use data and reason to determine how to do the most good possible.

When they got together, Amodei was still an academic, working as a postdoctoral scholar at Stanford University School of Medicine, having earned his Ph.D. in computational neuroscience from Princeton. He would go on to stints at Baidu and Google before joining OpenAI in 2016. 

Amodei quickly climbed the ranks, helping lead key research into “scaling laws”—the notion that intelligence increases proportionally with the increase in computing power and data used to train it.

As Amodei ascended within OpenAI, Clark joined him at company events. People who worked with him and knew the couple said she made sure his ideas were given the credit she felt they deserved. 

She introduced Mira Murati, one of her closest friends, to OpenAI co-founder Greg Brockman, according to people familiar with the matter. Clark had worked with Murati at Leap Motion, a company that made a virtual-reality hand-tracking device. Murati joined in 2018 as VP of applied AI and partnerships and later became the company’s chief technology officer.

Clark also introduced Amodei to Schmidt, who became taken with the young scientist. In 2018, Schmidt visited Amodei and Clark’s apartment in San Francisco and was impressed with their ideas, Schmidt told Bloomberg.​ 

Eric Schmidt at the 2011 Digital-Life-Design conference in Munich. Miguel Villagran/Getty Images

Eric Schmidt at the 2011 Digital-Life-Design conference in Munich. Miguel Villagran/Getty Images

When tensions arose between Amodei and OpenAI co-founders Sam Altman and Brockman over who wielded power at the startup, Amodei, his sister and five other employees left to found Anthropic, in December 2020. 

Anthropic in May 2021 announced it had raised $124 million for its Series A round of fundraising, led by Skype engineer Jaan Tallinn and including Facebook co-founder Dustin Moskowitz, Jane Street founder James McClave and the Center for Emerging Risk Research, as well as Schmidt.  

Clark, who wasn’t yet married to Amodei, wanted equity in the new company as well, according to documents reviewed by the Journal. In February 2021, Clark pitched Schmidt on creating a new venture fund called the Mother of AGI Fund. Its goal was to be “an elegant solution to formalize Cami’s involvement in Anthropic (Dario’s company), manage Eric’s investment,” and invest in “the AGI ecosystem,” according to a 40-page proposal. 

Others at Anthropic, including Daniela Amodei and other co-founders, didn’t support the plan, according to people familiar with the matter, and it didn’t move forward. 

By the end of that year, associates said they recall seeing what looked like an engagement ring on Clark’s hand, and not long afterward, the couple wed in Italy. 

Clark, known for being social and outgoing, has interacted with politicians and potential investors on behalf of Amodei, chatting with them and bringing them over to meet her bookish husband at conferences and events.

Anthropic drew the ire of the Trump administration earlier this year after Amodei refused to drop restrictions on how Claude could be used by the Pentagon. The government labeled Anthropic a supply-chain risk, and Pentagon partners were effectively barred from using its technology in their work with the Defense Department, a designation the company has challenged in court. The administration added restrictions on additional AI models from Anthropic in June that were later dropped. The standoff brought unwelcome attention to the company ahead of its IPO.

Clark has recently told people involved in politics that the company sees its mission as protecting America and isn’t as “woke” as its detractors might think it is, according to people familiar with the matter. 

At the Sun Valley conference in July, where she spent much of the week by her husband’s side, she had lunch with Ivanka Trump, a friend, and chatted with Jared Kushner, whom Amodei had approached earlier in the year to invest, people familiar with the matter said.

‘Revolutionary porn company’

Clark’s early years show a young woman hustling to succeed at a range of businesses, many she founded, including a porn company targeting women. 

In 1999, at the age of 20, she married 64-year-old Reno architect Waldemar Eklof III, who had designed buildings including the city’s Atlantis Casino Resort. They divorced three years later. On a now-defunct personal website, Clark said she dropped out of architecture school in 1999. 

Property records show she moved to San Francisco in 1999 and spent a decade there before spending time in New York and Los Angeles. In that period she worked for a mortgage brokerage company called OLG Financial. While raising money for a later venture, she would describe herself as the “co-founder of OLG Financial Corp.” The company’s 2005 incorporation documents don’t include her name.

A lender foreclosed on her San Francisco apartment in 2007, and two years later she filed for bankruptcy, according to property records and court filings.

Around 2010, she and Michelle Capocefalo started Eddice, which described itself as a “revolutionary porn company.” Named for Eddice Munson, Clark’s maternal grandmother, it aimed to emphasize sex positivity in a male-oriented porn industry, and carried the tagline: “intellectually promiscuous.” 

Michelle Capocefalo and Clark at a gallery event in New York in 2010. RYAN MCCUNE/Patrick McMullan/Getty Images

It included a blog that gathered content with titles such as “Lady Porn Day” and “Orgasm, Inc.” It also highlighted the work of intellectuals such as pundit Clay Shirky and marketing expert and author Seth Godin.

At one point, after alerting readers that its PayPal account had been frozen, Clark’s company announced on its website that it was “crowdsourcing” funding. “We are raising $350,000 to produce EDDICE.COM and our first four films, AMERICAN GIRL IN PARIS.”

In one of her pitches, part of which was reviewed by the Journal, Clark said her resume included “real estate + entrepreneurial investments” as well as “distribution + brand strategy consultant, interior design/architectural background.”​

Jeffrey Epstein’s world

Clark and Capocefalo attended the 2011 Digital-Life-Design conference in Munich, a tech, media, science and arts salon that executives often stop by on their way to the World Economic Forum in Davos. That year, the keynote speech was given by Schmidt, then still CEO of Google.

Little is publicly known about the three-year relationship between Clark and Schmidt, who remains married to another woman. Schmidt was also an investor in one of Clark’s startups, which focused on women’s health, according to people familiar with the matter.

At the Munich conference, Clark and Capocefalo met John Brockman, a literary agent of scientific books, who introduced Epstein to academics and Silicon Valley elites, according to Epstein documents recently released by the Justice Department.  

John Brockman introduced the women to Epstein that March.​ Writing that he was headed to a Wired magazine party where Brockman was the guest of honor, he told Epstein, “You should connect for dinner with my girls—Cami and Michelle. They’re in LA raising money for porn movie aimed at women’s market.” He then shared the website for Eddice.​

John Brockman at the Munich digital conference in 2011. Tobias Hase/DPA/ZUMA PRESS

Brockman concluded the March 3 email by noting a birthday party being thrown for him that Saturday, saying Clark and Capocefalo would be there. Clark followed up the email soon after, writing to Epstein, “Hello Jeffrey! We would love to have dinner with you this evening.” She sent him her cellphone number. John Brockman didn’t respond to requests for comment.

The next week, Clark sent Epstein an email with an attachment. “We thought you and the ladies might enjoy the script/treatment for our first 4 films, American Girl in Paris. A little nsfw,” she wrote.

A year later, she reached back out to Epstein. At first he didn’t remember her. She replied that they had been introduced by John Brockman a year earlier. “We have the free luxury porn company. Does that ring a bell?”

Epstein replied: “Yes, a loud gong.”

Clark joked back. “Haha, I was going to say, you would be the first person that didn’t remember us.” 

She pitched him on investing in her company, saying they had just hired someone from Oprah Winfrey’s company and Ashton Kutcher’s Katalyst, the media company behind shows like “Punk’d.” “Are you still interested?”

Epstein, then a registered sex offender, replied, “Can’t do sex TV.”

Emails between Clark and Jeffrey Epstein, released by the Justice Department. Justice Department

She pivoted to pitching him another business, a “social dieting app + website geared toward women” that would combine data from Fitbits and other health monitoring devices, user-entered data about nutrition, menstrual cycles, mood and location, and the ability to share it all, Facebook-style.

Their interaction continued over two years, the government released Epstein files show. Clark invited Epstein to her and Capocefalo’s housewarming party in Manhattan, and connected with him as friends on LinkedIn.

In 2013, Epstein received an automated note from LinkedIn that he should congratulate Clark for her new role as CEO of Female Algorithm Technologies. Traces of it online are scant, except for a Tumblr account with the edgy aesthetic that combined photos of nude women, a portrait of Steve Jobs with an early Macintosh, and all-caps mottos like Great Minds F— Each Other, using the full expletive, and “Let’s Make Lots of Money.”

Copyright ©2026 Dow Jones & Company, Inc. All Rights Reserved. 87990cbe856818d5eddac44c7b1cdeb8