By the FL Innovation News editorial team · Entrepreneurship
Written By: Pete Weishaupt
What a Cluster of Missing Scientists Reveals about Institutional Memory and Key-Person Risk
A peculiar pattern has entered the zeitgeist of the American news cycle recently: missing scientists, dead researchers, vanished administrators, a retired general up and vanished, and a widening public argument over whether any of it adds up to anything at all. The White House has acknowledged the issue is being reviewed. The Department of Energy’s National Nuclear Security Administration has said it is “aware of reports” involving employees at its labs, plants, and sites and is looking into the matter. At the same time, officials and experts interviewed by CBS News said they see no publicly established evidence of a single coordinated plot, and one government source told CBS the FBI was not treating the cases as one suspicious pattern.
But should they?
That question matters. There is tension between pattern recognition and pattern invention. And yet even if you reject the most dramatic explanations, there is still a serious strategic question staring back: what happens when a country loses a series of highly specialized people clustered around its space, nuclear, propulsion, advanced materials, and defense-research ecosystems? Even if every case ultimately proves unrelated, the concentration alone exposes a vulnerability modern states do not like to discuss: knowledge is not stored only in servers, vaults, and paper. It is stored in people.
That is the framework worth using here. Call it a knowledge scrub: not as a proven criminal finding, but as an analytic model for understanding the strategic effect of losing a small number of deeply embedded experts and connective personnel. In every serious enterprise there are people who function as institutional memory, the quiet repositories of what worked, what failed, whom to call, what not to repeat, and where the hidden skeletons are. Harvard Business Review describes subject-matter experts as keepers of critical information and institutional memory. Remove enough of those people, and the organization does not simply get smaller. It gets dumb. It slows down. It forgets. It relearns old lessons at new cost.
And this is why the recent cluster has drawn so much attention. Public reporting now places at least 11 names in the discussion. The list spans Michael David Hicks of NASA’s Jet Propulsion Laboratory, Frank Maiwald of JPL, Monica Reza of JPL, Carl Grillmair of Caltech/NASA work, Nuno Loureiro of MIT’s Plasma Science and Fusion Center, Jason Thomas of Novartis, retired Air Force Maj. Gen. William Neil McCasland, Steven Garcia tied to the Kansas City National Security Campus in Albuquerque, Melissa Casias and Anthony Chavez with Los Alamos ties, and Amy Eskridge in Huntsville. Officials have not confirmed a connection among them. But the accumulation of cases across the space, nuclear, and advanced research world is what has elevated the story from local tragedy to national security curiosity.
The New Mexico cluster is the sharpest example. KOB reported that at least four people with some level of access or proximity to classified materials, or who had that access at some point in their careers, disappeared in New Mexico within roughly 11 months: McCasland, Garcia, Casias, and Chavez. Geography alone does not prove conspiracy, but geography does matter in intelligence work. Albuquerque and Los Alamos are not random dots on a map. They sit inside one of the most sensitive research corridors in the country, where nuclear stewardship, defense science, laboratory administration, and retired senior leadership overlap in ways that create both value and vulnerability.
McCasland’s disappearance is one reason the story broke into the open. Fox reported that he vanished in late February from his Albuquerque home during roughly a one-hour window while his wife was out. His phone, glasses, and wearable devices were found in the house; his wallet, hiking boots, and a .38-caliber revolver were missing. Fox also reported that he had previously described experiencing a “mental fog,” though authorities said there was no indication he was disoriented at the time. His career included leadership roles at the Air Force Research Laboratory and the National Reconnaissance Office. Those are not incidental credentials. They place him in the category of people who do not merely know facts, but know architectures.
What stands out in several of the disappearance cases is not simply absence, but electronic absence. Public reporting has emphasized that McCasland left his phone and devices behind; other reports have described similar device-related anomalies in the cases of Garcia and Casias. Standing alone, that proves little. But in the aggregate it suggests a problem that investigators would be negligent to ignore: either a set of people under strain made unusually consequential decisions to sever their digital trails, or something more deliberate happened around them. Those are very different possibilities, but each is serious.
Monica Reza’s case pushes the pattern westward and broadens its significance. Newsweek reported that Reza, director of materials processing at JPL, disappeared in June 2025 while hiking in California. It also reported that she had worked on a government-funded rocket materials project overseen by McCasland. If that is accurate, then her importance lies not in a sensational label but in what advanced materials science actually does for a modern state: it determines what engines survive, what systems endure heat, what payloads can be delivered, and how long an engineering lead remains an engineering lead.
The California cases do not stop there. Newsweek reported that Michael David Hicks, a longtime JPL scientist who worked on Deep Space 1 and the DART asteroid-deflection mission, died in July 2023, with no public cause of death disclosed in the Daily Mail report it summarized. The same Newsweek piece lists Frank Maiwald, another JPL principal researcher, as having died in July 2024, and Carl Grillmair as having died in February 2026. Separate reporting confirms Grillmair was fatally shot at his home in Llano, California, and a suspect was charged. Whatever the explanation for the full cluster, the accumulation of losses around the JPL-Caltech orbit is difficult to dismiss as irrelevant to institutional continuity.
Massachusetts adds another flank. MIT announced the death of Nuno Loureiro, director of its Plasma Science and Fusion Center, after he was shot in Brookline in December 2025. AP likewise reported that authorities opened a homicide investigation and, at the time, had no suspects in custody. Newsweek’s running list also includes Jason Thomas, a Novartis pharmaceutical researcher whose body was recovered from Lake Quannapowitt in Wakefield after he had been missing since December 2025. These facts do not, by themselves, prove a single campaign. But they do underscore that the public record now spans multiple states, multiple sectors, and multiple methods of disappearance or death.
Amy Eskridge is the most combustible name on the list and the easiest to mishandle. Newsweek reported renewed attention to her 2022 death in Huntsville after reports that she had expressed concerns about her safety. Fox separately noted there is no publicly available evidence linking Eskridge’s death to the other cases, and authorities have not indicated any connection between her work and the circumstances of her death. That is where a responsible analysis must pause. Eskridge may ultimately belong inside the same story, or she may belong in a different one entirely. At this stage, the honest answer is that the public record does not let us say one way or the other.
Why would a “knowledge scrub” deserve our attention at all?
Because the strategic effect of losing key people is not proportional to headcount. CBS reported that the facilities in question collectively employ more than 20,000 people, many in administrative or support roles without access to important information. That is a useful corrective against melodrama. But it is also the wrong denominator for the question that matters. The relevant population is not total payroll. It is the much smaller layer of people who hold rare tacit knowledge, cross-organizational memory, and authority over complex systems. That population may be low thousands, perhaps smaller still if one narrows it to principal investigators, senior engineers, niche materials specialists, senior administrators with sensitive access, and retired leaders who still carry unique historical understanding.
In such a network, losing 11 people is not a 0.05% problem. It can be a connectivity problem. Think of a specialized ecosystem of roughly 2,000 high-value nodes. Eleven losses would equal about 0.55% of the population. In a flat workforce, that would be unpleasant but survivable. In a hub-and-spoke knowledge network, it can be devastating if those losses cluster among bridges, repositories, and gatekeepers.
A bridge is someone who connects one domain to another: defense to academia, lab to acquisition, engineering to operations. Remove the bridge and information still exists, but no longer travels cleanly.
A repository is the veteran who remembers the failed tests, dead ends, informal workarounds, the unwritten reasons a promising line of effort was abandoned ten years ago. Remove the repository and the organization starts paying tuition to its own past.
A gatekeeper is the person who knows who can access what, which program offices matter, which teams can be trusted, where the administrative choke points sit. Remove the gatekeeper and coordination degrades even before capability does.
This is not abstraction. It is how high-end technical systems fail in real life: not all at once, but through delay, duplication, hesitation, clearance bottlenecks, broken mentorship chains, and the quiet disappearance of context. HBR’s point about experts as custodians of institutional memory is mundane in ordinary business. In national security science, it becomes acute.
That is why the number “11” matters less than the type of 11. If the wrong 11 vanish from a network of 2,000, the damage can exceed the loss of hundreds of ordinary positions. A principal researcher in deep-space, a materials specialist, a fusion center director, a retired general who understands how old programs connect to current ones, an administrative employee with sensitive scheduling and access visibility, a contractor who knows the location and flow of high-value components: these are not interchangeable human resources. They are nodes of compression. They hold more than one person’s worth of memory.
History offers uncomfortable precedent for this way of thinking.
The most obvious example is Operation Paperclip. Britannica notes that the United States sponsored the immigration of roughly 1,500 German and Austrian scientists and technicians after World War II in order to exploit their knowledge for military and industrial purposes. The point was not merely to acquire documents. It was to acquire people before rival powers did. Paperclip was a recognition, in policy form, that advanced knowledge travels poorly without the humans who embody it. It was also an acknowledgment that denying such people to an adversary can matter almost as much as recruiting them for oneself.
A second precedent is the sustained campaign against Iranian nuclear scientists. Reuters has documented multiple attacks on Iranian scientists from 2010 onward and reported that Mohsen Fakhrizadeh was widely regarded by Western intelligence as a key figure in Iran’s alleged covert nuclear-weapons effort. Reuters also noted that he was the only Iranian scientist named in the IAEA’s 2015 final assessment of possible military dimensions of Iran’s program. That is what a key node looks like in the real world: not just a credentialed researcher, but a person whose role, history, and organizational centrality make him uniquely valuable. Whether one supports or opposes such operations is beside the point. The precedent is clear: states have long recognized that removing certain people can slow, disrupt, or scramble strategic programs.
A third example is Gerald Bull, the artillery and “supergun” engineer assassinated in Brussels in 1990. His death remains one of the classic illustrations of how a project can outlive its paperwork but not its architect. Bull’s work for Iraq had raised alarm precisely because he was not just another engineer on a large team; he was the intellectual center of gravity for a very specific and dangerous technical ambition. Remove him, and the project’s viability changed immediately.
Then there is the British Marconi deaths episode of the 1980s. That case remains deeply contested, and it should be treated cautiously. The “Marconi deaths” are as much a lesson in evidentiary humility as they are in Cold War unease. What makes them relevant here is not that they prove a hidden hand, but that they show how quickly clusters of deaths around sensitive technical work can become breeding grounds for speculation when authorities cannot provide satisfying clarity. In other words: even when a pattern is not provable, the strategic costs of ambiguity are real.
That brings us back to the present cluster.
There are three ways to read it.
The first is the null hypothesis: unrelated tragedies, accidents, homicides, and disappearances, amplified by social media and glued together by the public appetite for pattern. CBS quoted experts who saw no obvious link and emphasized that these cases are spread across years, organizations, and circumstances. That view deserves serious weight. It may turn out to be correct.
The second is the mixed-threat hypothesis: not one master plot, but a combination of ordinary tragedy, isolated criminal violence, and a smaller number of genuinely counterintelligence-relevant incidents. This is often how the real world looks. Some cases may be tragic but personal. Others may reveal security weaknesses. A few may expose something far darker. Bureaucracies dislike this explanation because it offers neither the comfort of coincidence nor the drama of total conspiracy.
The third is the knowledge-scrub hypothesis: that at least some of these cases, especially where they cluster around sensitive institutions and where device abandonment or unusual disappearance details recur, amount to a campaign aimed not primarily at stealing documents but at degrading continuity, memory, and confidence. This remains unproven. But it is the one hypothesis that takes seriously the strategic logic of human capital as a target set.
There is another reason to resist dismissing the issue too quickly. Even if no coordinated foreign operation is eventually established, the United States has still learned something unflattering about itself: it has allowed too much critical knowledge to remain concentrated in too few people. The vulnerability exposed by this story may not be only that adversaries can target key personnel. It may be that the system is still too dependent on them once they are gone. That is not a conspiracy problem. That is a resilience problem.
If a modern national security enterprise is wise, it does not wait for certainty before reducing concentration risk. It maps the bridges. It identifies the repositories. It captures tacit knowledge before retirement, illness, disappearance, death, or simple attrition scrubs it from the record. It treats key-person risk as seriously as cyber risk, because in sophisticated technical systems the two can intersect: an intruder can steal data, but the loss of the wrong human being can still do more lasting damage than the theft of the files.
The same principle applies far beyond national security. Mid and Small companies live with their own version of the knowledge-scrub problem every day. A competitor does not need to steal your entire database to weaken you. Sometimes they only need to hire away the one operations manager who knows how the plant really runs, the estimator who understands your pricing logic, the salesperson who owns the customer relationships, the controller who knows where the margin leaks are, or the project lead who quietly holds the company together. In a smaller business, the loss of one or two critical people can erase years of institutional knowledge almost overnight.
Key-person risk is not just an HR problem. It is a corporate intelligence problem. It is a resilience problem. It is a competitive risk problem. Every business has bridges, repositories, and gatekeepers. The question is whether leadership knows who they are before a competitor, buyer, creditor, litigant, or disgruntled employee does.
The current public evidence does not prove that America is under a coordinated campaign of scientific decapitation. It does prove something narrower and still disturbing: that a small number of deaths and disappearances, concentrated around sensitive research domains, can trigger federal concern, public alarm, and an immediate debate over whether the country understands where its irreplaceable knowledge actually resides. Officials have not established a link. Experts caution against conspiratorial overreach. All of that can be true at once.
But if the United States waits for courtroom-level proof before confronting the strategic meaning of tacit knowledge loss, it will have misunderstood the problem. And the same is true for the private sector. A company does not have to wait for a lawsuit, a poaching campaign, a failed acquisition, a cyber incident, or a surprise resignation to ask where its most valuable knowledge lives. By then, the damage may already be done.
This is where corporate intelligence and investigations matter. At Weishaupt Strategic Group, we look at businesses the way intelligence professionals look at networks: who holds influence, who controls information, where the vulnerabilities sit, and what happens if a key node disappears. The goal is not paranoia. The goal is preparation. A company that understands its human knowledge network can protect it, document it, succession-plan around it, and reduce the chance that one departure becomes an institutional amputation.
The lesson of history, from Paperclip to Iran’s targeted scientist killings to the fate of Gerald Bull, is that states have always known the value of rare minds. The same lesson applies in business. Competitors know the value of rare minds, too. The open question in this case is not whether human capital can be targeted. It can. The open question is whether America, and the companies operating inside it, have become too comfortable assuming their own depth will always compensate for the loss of their most important nodes.
Depth matters. So does memory. And memory, unlike hardware, cannot be surged overnight.
About the Author
Pete Weishaupt is the founder of Weishaupt Strategy Group (WSG), a corporate intelligence and Weishaupt Investigations, LLC, a private investigation firm focused on AI-enabled due diligence, investigative research, and risk analysis.
With more than three decades of experience in the intelligence community, Pete brings structured analytic tradecraft, HUMINT discipline, and operational rigor to complex corporate and legal matters. At WSG, he is building the world’s first AI-native investigative firm that combines secure generative AI workflows with traditional investigative methodology; all while enhancing speed, clarity, and defensibility without compromising confidentiality or professional standards.
Pete advises corporate leaders, investors, and counsel on matters involving background investigations, enhanced due diligence, pre-transaction intelligence, and risk. His philosophy is simple: AI should augment investigator judgement, not replace it.
Learn more at weishaupt.ai
