So Daniel sent us this one, and I have to say it's a genuinely strange corner of the academic world. He's asking us to dig into intelligence studies as an academic discipline — the history of it, the landscape, the leading institutions. And the hook he gives us is this paradox: how does a field built entirely around secrets maintain publicly accessible journals, peer-reviewed programs, and a growing global footprint? He's framing it as a follow-up to the Snowglobe episode — the CIA's AI wargaming system that turned up in their own publicly accessible journal — and asking us to pull on that thread. How did this field get formalized? Who are the serious centers of excellence? And how do you even do rigorous scholarship when your primary source pool — actual intelligence officers — legally can't tell you most of what they know?
Herman Poppleberry here, and I'll say upfront — this topic has been sitting in the back of my mind since the Snowglobe material crossed my desk. Because that article, the one documenting a cutting-edge CIA-affiliated AI system in a publicly accessible journal, is such a perfect illustration of the paradox Daniel is pointing at. And the paradox goes all the way back to nineteen fifty-five.
Which is further back than I would have guessed for a formalized academic field around intelligence.
That's the thing most people get wrong about intelligence studies — they assume it's a post-Cold War phenomenon, something that emerged after the Church Committee hearings in the seventies or after the intelligence failures of the early two thousands. But the founding document of the field is Sherman Kent's nineteen fifty-five article, "The Need for an Intelligence Literature," published in the very first issue of Studies in Intelligence, Volume One, Number One. And Kent's argument is essentially a manifesto. He was a Yale historian who became the father of American intelligence analysis, and he said, look — intelligence, like any serious subject, needs formal literature to be taught and developed. It needs its own vocabulary, its own techniques, its own theory.
Which is a slightly wild thing to argue in a classified journal in nineteen fifty-five. Like, he's writing a manifesto for a field that barely exists yet, in a document almost nobody can read.
That's the founding paradox right there. The journal that Kent created — Studies in Intelligence — operates on a two-tier model. There's the classified version that circulates internally within the intelligence community. And then there are what they call "unclassified extracts" — articles cleared for public release, published on the CIA's website. The public collection now contains unclassified issues from nineteen ninety-two to the present, plus over six hundred articles from nineteen fifty-five through two thousand and four that are in the National Archives and accessible online. So as of right now, the journal is on Volume Seventy, Number One — March twenty-twenty-six — meaning it has been published continuously for seventy-one years.
Seventy-one years. And it's still going. By the way, Claude Sonnet four-point-six is writing our script today — which feels somehow appropriate for an episode about AI systems and intelligence institutions. Anyway. Seventy-one years of a journal that most people have never heard of.
Most people outside the field, yes. But within intelligence studies — and within the broader national security academic world — it's the foundational text. The Center for the Study of Intelligence, which is the CIA's in-house research arm and the publisher of Studies, has put out over sixty books and monographs since nineteen ninety-two. Things like the Psychology of Intelligence Analysis from nineteen ninety-nine, which is genuinely a foundational text in analytical tradecraft. A Tradecraft Primer from two thousand and nine. These are publicly available. You can download them from the CIA website.
Which still sounds weird to say out loud. "Just download it from the CIA website."
It does. But that's the model. And Kent's vision — that intelligence needed to develop as a genuine intellectual discipline with its own literature — has been substantially realized over the following seven decades. The field now has multiple peer-reviewed journals, graduate programs at civilian universities and military war colleges around the world, dedicated research centers, and its own professional academic organization.
Let's get into the institutional landscape, because I think that's where it gets interesting. When people think about where you'd study intelligence seriously, what are the actual centers of excellence?
So there are a few distinct tiers. You have the CIA's own Center for the Study of Intelligence, which is the practitioner-facing institution — it's not a university, it's an internal research and professional development body. Its mission is explicitly to prepare intelligence officers for future challenges by learning from the past. The current director is John Pulju, who also chairs the Studies in Intelligence editorial board. The CIA's chief historian, David Robarge, is a frequent contributor.
So that's the inside-the-wire institution. What about the civilian academic world?
The leading civilian center outside the United States is King's College London, specifically the King's Centre for the Study of Intelligence, which sits within the Department of War Studies. KCL describes its department as one of the only academic departments in the world focused solely on the complexities of conflict and security, and the intelligence center is described as one of the world's leading centers for intelligence studies. Their Intelligence and International Security MA covers everything from digital espionage to great power rivalries to domestic terrorism, and it's supplemented by visiting practitioners — people who've actually worked in the field.
And KCL has some institutional credibility here beyond just self-description, right? They're hosting the twenty twenty-six Society for Intelligence History conference.
In October, yes — October eleventh through thirteenth. Co-hosted with the Society for Intelligence History, which was founded in twenty sixteen. And that's actually a significant moment for the field, because the SIH — formerly the North American Society for Intelligence History — rebranded in twenty twenty-four to reflect its growing global reach. Holding the annual conference at King's College London signals that this is no longer primarily an American academic conversation.
Who founded the SIH? Because twenty sixteen is surprisingly recent for an organization in a field that supposedly goes back to nineteen fifty-five.
Sarah-Jane Corke at the University of New Brunswick and Mark Stout at Johns Hopkins. And Corke's explanation for why they founded it is revealing — she said the number of scholars in the field had grown dramatically and they really did not have a home. They also wanted a platform to advocate for declassification, because without documents there will be no intelligence history. That's a direct quote. The organization went from founding in twenty sixteen to a sold-out conference at the International Spy Museum in Washington in twenty twenty-five in under a decade. That's real momentum.
Mark Stout is an interesting figure in this story because he connects the academic and the public-facing sides — he was the historian of the International Spy Museum before Johns Hopkins.
He's a good example of the kind of career trajectory that's possible in this field. You also have John McLaughlin, who was the acting Director of Central Intelligence, now a senior fellow and distinguished practitioner-in-residence at Johns Hopkins. Thomas Fingar, who was the first Deputy Director of National Intelligence for Analysis, is a fellow at Stanford's Freeman Spogli Institute. These are people who ran major intelligence institutions, and they've moved into academia as a second career.
Which gets us to the structural tension at the heart of the field. Because those people can contribute meaningfully to the academic literature precisely because they're retired. The people with the most current, most operationally relevant knowledge are the ones least able to share it.
That's the central epistemological problem of intelligence studies, and it's worth unpacking carefully. The CIA's Prepublication Classification Review Board — the PCRB — governs what intelligence professionals can publish. And the scope of that obligation is broader than most people realize. It's lifelong. Every CIA officer and contractor must submit any intelligence-related publication for review, and that obligation never expires after leaving the agency. The scope covers not just their immediate area of work, but any material that mentions the CIA or intelligence activities, or concerns topics on which they had access to classified information.
So if you spent three years working on, say, signals intelligence in East Asia, and you want to write a history of Cold War cryptography thirty years later, that still goes through review.
Potentially, yes. And the definition of "publication" is expansive — books, articles, academic papers, blog posts, speeches, scripts, book reviews, opinion pieces, even resumés. The PCRB has an acronym they use for manuscript submissions: AVOID. A for agency officers and assets, V for validate open sources, O for operational details, I for intelligence sources and methods, D for disclosing locations. Each of those categories represents a potential reason for redaction.
The resumé one is fascinating to me. You can't put your actual job history on your resumé because the job history itself is classified.
The PCRB's own documentation mentions something called the mosaic effect, which is the concept that individually innocuous pieces of information can become classified when combined with other information. So a retired officer writing a resumé might need to redact specific countries, office names, or even numbers — not because any single fact is classified, but because the combination creates a picture that reveals something sensitive. That's a profound constraint on academic work. It means the literature is systematically biased toward what can be said rather than what is most important.
And yet the field has developed mechanisms to work around this. The pen name system in Studies in Intelligence is something I find genuinely strange and fascinating.
It's one of the more unusual features of any academic journal I'm aware of. Recent issues of Studies in Intelligence list contributors like "Graham Alexander" — described as "the pen name of a CIA operations officer." "Resolute Lee" — "the pen name of an ODNI officer." "Charles Heard," "Ian B. Ericson" — both described as pen names of CIA officers. And then there are partial identities: "Ana P., a CIA analyst." "Sarah, a CIA senior analyst focused on Russia." "Matthew J., a CIA analyst." "Mike R., a member of CIA's History Staff."
So you have a peer-reviewed academic journal where the peer review is happening, citations are being made, knowledge is accumulating — but a significant portion of the author pool is partially invisible. How do you build an academic reputation under a pen name?
You largely don't, in the traditional sense. The contribution to the field is real, but the individual career benefit is limited. And that creates an interesting incentive structure. Active officers contributing under pen names are presumably motivated by professional development, by the value of the field itself, by the internal prestige of being published in Studies. The external academic reputation accrues to the journal and the institution, not the individual author.
There's something almost medieval about it. Like the anonymous manuscript tradition, but for counterterrorism analysis.
The parallel isn't entirely frivolous. Medieval scholarly traditions had similar structures where the institution — the monastery, the scriptorium — accumulated intellectual authority while individual contributors remained anonymous. The difference is that in intelligence studies, the anonymity is legally mandated rather than culturally chosen.
Let's talk about the Snowglobe article specifically, because Daniel frames it as a case study in this paradox. Here you have a cutting-edge AI wargaming system, developed by In-Q-Tel — the CIA's technology incubator — and its Directorate of Digital Innovation, and it ends up documented in a publicly accessible academic journal.
It's a remarkable document. The article appears in Volume Sixty-Nine, Number Four of Studies in Intelligence, December twenty twenty-five. The title is "Snow Globe Multi-Player AI System: Lessons from Human-AI Teaming in War Games." The authors are Andrea Brennan, who's the Senior VP and Deputy Director of IQT Labs; Rachel Grunspan, who's a retired CIA officer and former Director of Digital Futures in the CIA's Directorate of Digital Innovation; Daniel Hogan, a senior data scientist at IQT Labs; and two digital intelligence strategists from DDI Futures, Jessica Smith and Elizabeth VanderVeen.
So you have active and retired practitioners, from the CIA's own technology arm, publishing in the CIA's own journal, about a system that uses large language models to play open-ended war games. And the article is publicly accessible.
The abstract states that the CIA's Directorate of Digital Innovation and In-Q-Tel collaborated on a research project to explore how humans and conversational AI can work together more effectively, leveraging Snow Globe, a multi-player AI system built by IQT's Applied Research Team. The article is detailed enough to be academically useful — it discusses the system's architecture, the lessons learned from human-AI teaming, the challenges of running wargames with LLM-based agents. But it's carefully sanitized. You're not getting the classified operational applications. You're getting the methodology, the framework, the lessons that can be shared.
Which is actually a sophisticated piece of knowledge management. They're contributing to the public literature on AI and wargaming — which has genuine academic value — while keeping the operationally sensitive details internal.
And it's a model that the field has refined over seventy years. The declassification pipeline that the Center for the Study of Intelligence maintains — over fourteen hundred articles from nineteen fifty-five through nineteen ninety-seven, delivered to the National Archives in nineteen ninety-seven and now digitized — that's a growing body of declassified intelligence scholarship that academics can cite and build upon. The classified version of the field's knowledge is always going to be ahead of the public version, but the gap narrows over time as documents are declassified.
Let's talk about Mercyhurst, because I think it's an underappreciated part of this story. When you think about where civilian intelligence education happened first in the United States, it's not Johns Hopkins or Georgetown — it's a small Catholic university in Erie, Pennsylvania.
Mercyhurst is genuinely pioneering in this space. They were one of the first universities in the US to offer a dedicated undergraduate intelligence studies program, and they now have the Ridge College of Intelligence Studies and Applied Sciences — named after Tom Ridge, who was the first Secretary of Homeland Security. They offer bachelor's, master's, and graduate programs specifically in intelligence studies. The fact that a field-specific bachelor's degree exists at a civilian university is itself a marker of how formalized the discipline has become.
And then on the military side you have the war colleges, which are a different beast entirely.
The war colleges occupy an interesting position in the ecosystem. The National War College in Washington, the Army War College at Carlisle, the Naval War College at Newport, the Air War College at Maxwell Air Force Base, the Command and General Staff College at Fort Leavenworth — these are graduate-level institutions that produce the senior military and national security leadership of the United States. Intelligence studies is woven into the curriculum at all of them. Scott Moseman, who's an assistant professor in the Department of Military History at Fort Leavenworth, wrote a book called Defining the Mission: The Development of US Strategic Military Intelligence up to the Cold War. That's the kind of serious historical scholarship coming out of the military academic world.
And then internationally you have Royal Military Academy Sandhurst on the British side.
Aaron Edwards at Sandhurst is a senior lecturer in defense and international affairs and a contributor to the field. The SIH's twenty twenty-six conference at King's College London — which is explicitly co-hosted with the King's Centre for the Study of Intelligence — is a signal that the British academic intelligence studies world is increasingly integrated with the North American one.
Here's something I want to push on, because I think it's an underexplored angle. The IRTPA — the Intelligence Reform and Terrorism Prevention Act of two thousand and four — reshaped the entire intelligence community structure. It created the ODNI, the Director of National Intelligence. What did that do to intelligence studies as a field?
The IRTPA effect on the academic field is interesting. Studies in Intelligence published a special twentieth anniversary edition in December twenty twenty-four — Volume Sixty-Eight, Number Five — with contributions from former Directors of National Intelligence: Hayden, McConnell, Negroponte. Congressional leaders who wrote the legislation, including Susan Collins and Jane Harman. Senior IC officials. It's essentially a collective memoir of the most significant intelligence reform in American history, and it's publicly available on the CIA website.
Which is only possible because the participants are now retired or the events are sufficiently historical. The twenty-year lag is the price of admission to the public literature.
That's a good way to frame it. The "retired practitioner pipeline" is one of the field's primary mechanisms for getting operational knowledge into the academic literature. Hayden Peake served in CIA's Directorates of Operations and Science and Technology, and he's been compiling and reviewing books in the "Intelligence Officer's Bookshelf" column in Studies since two thousand and two. John Ehrman is described as a retired Directorate of Analysis officer and frequent contributor. JR Seeger is a retired CIA paramilitary officer who does media reviews. Stephen Mercado is a retired CIA open-source officer and frequent contributor. These are people whose operational knowledge is now sufficiently historical that they can contribute meaningfully to the public literature.
The twenty-year lag creates a structural bias toward history over current events. Which is fine for academic purposes but means the field is always somewhat behind the operational frontier.
And the Snowglobe article is interesting precisely because it's an exception to that pattern. It's current — the wargaming work it describes is recent — but it can be published publicly because the authors have carefully extracted the methodology from the operational details. That's a different approach to the lag problem: instead of waiting twenty years for the whole thing to become declassified, you publish a sanitized version of the methodology now and keep the sensitive details internal.
I want to go back to something you mentioned earlier — Russia having its own intelligence journal. Because that's a detail that deserves more than a passing mention.
The December twenty twenty-five issue of Studies in Intelligence includes a review essay titled "Razvedchik, Russia's Intelligence Journal" by Mel Miller. Razvedchik is the Russian word for intelligence officer or scout. The existence of a Russian equivalent to Studies in Intelligence raises genuinely fascinating questions. Does the Russian intelligence academic tradition have the same practitioner-academic tension? Does the FSB or SVR have a prepublication review board? How does the Soviet intelligence studies tradition — which was obviously deeply secretive even by intelligence community standards — translate into an academic discipline?
And more broadly, what does it tell you about the universality of Kent's insight? The argument that any serious professional field needs a formal literature — that apparently holds across very different political and institutional cultures.
The British have the Journal of Intelligence History. The Americans have Studies in Intelligence and the International Journal of Intelligence and Counterintelligence. If the Russians have Razvedchik, you're looking at a genuinely global academic conversation, even if the participants can't always talk directly to each other. The intelligence studies conference at King's College London in October is interesting in this context because it's explicitly international — the SIH rebranded from the North American Society for Intelligence History precisely because the field had outgrown its original geographic scope.
Let's talk about the practitioner-academic bridge problem more directly, because I think there are a few different mechanisms at play and they each have different failure modes.
The mechanisms are distinct and worth separating. You have, first, the pen name system — active practitioners contributing under pseudonyms. The advantage is that you get current, operationally informed perspectives. The failure mode is that the academic community can't properly evaluate the author's credentials or potential conflicts of interest, because the author is anonymous.
Peer review gets complicated when you can't fully evaluate the reviewer's background either.
Second mechanism: the declassification pipeline. Documents become public over time, usually after fifteen to twenty-five years. The advantage is that you eventually get the full picture. The failure mode is the lag — by the time documents are declassified, the operational lessons may be obsolete, and the practitioners who could explain the context may be unavailable or deceased.
The Church Committee hearings are a good example of this. The documents that came out in the nineteen seventies revealed programs from the fifties and sixties. By then, the world had changed so dramatically that some of the lessons were of primarily historical rather than operational interest.
Third mechanism: the retired practitioner contribution. People like Hayden Peake or John Ehrman, who can now speak more freely. The advantage is substantive operational experience. The failure mode is selection bias — the practitioners who choose to contribute to the academic literature after retirement are not necessarily representative of the broader practitioner community. You're getting the ones who were already academically inclined, already interested in the history and theory of the field.
And potentially the ones whose experience is most compatible with what can be publicly discussed. The people with the most sensitive operational histories may be the least likely to contribute.
Fourth mechanism: the practitioner-in-residence model. People like John McLaughlin at Johns Hopkins, or the visiting practitioners at KCL's Intelligence and International Security MA. The advantage is that you get current-ish perspectives from people who've actually run major institutions. The failure mode is that even these people are still bound by their prepublication review obligations, and their institutional affiliations may create incentives to present the intelligence community in a particular light.
There's a fifth mechanism that I think doesn't get enough attention, which is the OSINT-informed academic. Someone who's never had a clearance, who works entirely from open sources, but who develops genuine analytical depth through rigorous open-source research.
The open-source tradition in intelligence studies is real and growing. The declassified archives, the Foreign Broadcast Information Service records, the State Department cables released through FOIA, the academic literature from foreign policy and international relations — a skilled researcher working only from open sources can develop a sophisticated understanding of intelligence operations and tradecraft. The limitation is that they can't verify what they're missing — they don't know the shape of the gap between what's public and what's classified.
Which is actually a version of the mosaic effect in reverse. The classified analyst knows that combining certain pieces of public information would be revealing, even if each piece individually seems innocuous. The uncleared academic can see the public pieces but doesn't know which combinations matter.
That's a precise way to put it. And it means there's a kind of asymmetric conversation happening in the field. The classified practitioners reading Studies in Intelligence can read the unclassified articles and mentally fill in the gaps from their classified knowledge. The uncleared academics reading the same articles are working with an incomplete picture and don't know exactly what's incomplete.
The Snowglobe article is interesting in this light too, because the authors are clearly writing for two audiences simultaneously. There's the internal IC audience that can read the classified version and fill in the operational details, and there's the external academic audience that gets the sanitized methodology.
And the external audience is genuinely valuable to engage, because the academic AI and wargaming community has developed tools and frameworks that the IC can learn from. The knowledge flow isn't only from practitioner to academic — it goes the other direction too. The practitioner-academic bridge is bidirectional.
Let's bring this around to what the practical takeaways are for someone who wants to engage seriously with this field. Because I think there's a version of this conversation that stays entirely abstract, and I want to avoid that.
The most important practical point is that the publicly accessible literature is far richer than most people realize. The CIA's Center for the Study of Intelligence website has over six hundred articles going back to nineteen fifty-five, plus sixty-plus books and monographs, plus the current unclassified extracts of Studies in Intelligence. The National Archives has the full run of the journal from its founding. If you're serious about intelligence studies, the primary source material is genuinely accessible — the barrier isn't classification, it's knowing where to look.
And the academic journals — Intelligence and National Security, the International Journal of Intelligence and Counterintelligence, the Journal of Strategic Studies — those are peer-reviewed publications with real scholarly standards. This isn't a fringe field producing grey literature.
The second practical point is about the career pipeline. The field has developed real institutional pathways. Mercyhurst for undergraduate, Johns Hopkins or KCL for graduate work, the war colleges for military officers. The SIH's graduate writing group, which they launched in twenty twenty-two, is specifically designed to support emerging scholars in the field. If you're a graduate student interested in intelligence history, there's now an institutional home for that interest.
And the third point, which I think is the most interesting for our audience, is what the Snowglobe article represents as a model. The idea that you can contribute to the public academic literature on a sensitive topic by carefully separating the methodology from the operational details — that's a generalizable approach. It's not unique to AI wargaming. It applies to any area where practitioners have knowledge they want to contribute but can't fully share.
The PCRB's own guidance supports this. They describe the review process not just as a constraint but as a protection — it helps officers identify problematic material and works with authors to find ways to make their points while avoiding classified information. So it's not purely adversarial. There are cases where the PCRB review process actually improves the article by forcing the author to be more precise about what they're claiming and why.
Though I'd note that "the classification review board helps you write better" is a sentence that requires some epistemic charity to fully accept.
Fair. The self-serving institutional framing is real. But the underlying point — that the discipline of separating what you know from what you can say can sharpen analytical writing — is defensible on its own terms.
One thing I keep coming back to is the seventy-year arc of Kent's vision. He wrote in nineteen fifty-five that intelligence needed formal literature, vocabulary, techniques, theory. By twenty twenty-six, you have a global network of universities, dedicated research centers, multiple peer-reviewed journals, professional academic organizations, war college programs, and a practitioner-facing institutional journal that's been publishing continuously for seven decades. That's a remarkable validation of the original argument.
And the field is still developing. The SIH was founded only a decade ago. The Society for Intelligence History's rebranding to reflect global reach happened in twenty twenty-four. The conference at King's College London this October is the kind of milestone that marks a field reaching institutional maturity at an international level. The Snowglobe article, published in December twenty twenty-five, shows that the field is actively grappling with the most current questions in AI and national security.
Sherman Kent would probably be pleased. And also slightly alarmed by the LLMs playing wargames part.
He was a Yale historian who became the architect of American intelligence analysis. I think he'd be fascinated by the wargaming work and deeply interested in whether the AI systems had developed anything resembling genuine analytical tradecraft, or whether they were sophisticated pattern matchers wearing analytical clothing.
That's a question for another episode. What's the open question you'd leave listeners with?
The one I keep returning to is the mosaic effect applied to the academic field itself. Every individual article in Studies in Intelligence, every declassified document in the National Archives, every retired practitioner memoir — individually, each one is a sanitized, carefully reviewed piece of the picture. But taken together, across seventy years of accumulation, does the public literature as a whole reveal more than any individual piece was intended to reveal? Is there a mosaic effect operating at the level of the entire field's public output? I genuinely don't know the answer to that, and I'm not sure anyone does.
That's a legitimately unsettling question to end on. Thanks as always to our producer Hilbert Flumingtop for keeping this whole operation running. Big thanks to Modal for the GPU credits that power the show — we genuinely couldn't do this without them. This has been My Weird Prompts. If you want to find us, search for My Weird Prompts on Telegram to get notified when new episodes drop. We'll see you next time.
Take care.
