Ninety-five percent.
That's the number that should stop everyone cold.
Admiral Ellis Zacharias, deputy chief of naval intelligence, estimated that ninety-five percent of US Navy peacetime intelligence came from legitimately accessible sources. Four percent semi-open. One percent from secret agents.
And that one percent is what every spy film ever made is about.
Daniel wants to know whether that universality is real. The dead drops, the brush passes, the numbers stations. Whether a CIA officer, an SVR illegal, and a Mossad agent are all running variations on the same discipline, or whether these organisations have actual operational fingerprints you could tell apart.
And then the harder question underneath it.
Whether any apparent difference is institutional culture at all, or just circumstance. Political system, legal constraints, geography, what the adversary's counterintelligence looks like.
Plus the trap he flagged himself, which is the one I keep thinking about.
That almost everything we know comes from operations that failed, agents who defected, or documents a government chose to release. Which means our sample is not the population. It's the wreckage.
Every case study in this field is a case study of something that went wrong.
So let's start where the documentary record is unusually explicit. A defector's own account of how Soviet intelligence defined its work.
Alexander Orlov. He ran Soviet intelligence operations in the thirties, and his essay for the CIA's historical review program was approved for release in September 1993.
And it opens with a definitional split that is much sharper than I expected.
Much sharper. Soviet services regarded as true intelligence, razvedka, only information procured by undercover agents and secret informants, quote, in defiance of the laws of the foreign country.
Everything else was dismissed.
Dismissed as mere research data. Reading newspapers, embassy reporting, academic journals, trade publications. In the Soviet view that wasn't intelligence at all. Orlov quotes the attitude directly, that it takes a real man to do the creative and highly dangerous work of underground intelligence on foreign soil, while the digging up of research data in the safety of the home office can be left to women or young lieutenants just beginning their careers.
That's not just a doctrine, that's a value system. There's contempt in it.
There's enormous contempt in it. And it tells you something about who the service promoted and who it sidelined. If the culture says the only real work is the dangerous work, then the people who rise are the people who did the dangerous work, and the analysts stay junior forever.
Which has consequences you can actually measure.
It does. A service that despises research data is a service that will underinvest in analysis, and then complain that it's drowning in raw reporting it can't make sense of. Which is more or less what happened to Soviet intelligence repeatedly across the Cold War.
And the Western services went the other way.
Completely the other way. They treat both types of information as intelligence, often with a much higher regard for research than for undercover work. Donovan called intelligence the business of pulling together myriad facts, making a pattern of them, and drawing inferences.
Which is a definition of analysis, not of espionage.
Right. And the force ratios are staggering. Orlov says the NKVD based its work one hundred percent on secret sources and undercover agents. Even the GRU, the military side, concentrated at least eighty percent of effort on informant networks and document theft.
So if doctrine differs this sharply at the level of what counts as the job, does it show up in actual operations?
That's the question, and the illegals program is the cleanest test case we have.
Ten Russians arrested on the twenty-seventh of June, 2010.
The FBI called the investigation Operation Ghost Stories, and it ran more than a decade. Ten SVR officers living under false identities, long-term deep cover, with every connection to Russia hidden. Not diplomats with a side interest. People with fake names, fake jobs, fake marriages, doing nothing for years except being American.
That's an enormous investment.
It follows from Lenin. He split the work into legal and illegal. Legal work is propaganda and organisation, run out of an embassy under diplomatic cover. Illegal work is espionage and assassination, run by people with no official connection at all. The two only meet at the Moscow Center.
Which is the opposite of how Western services developed.
Exactly backwards. Western services institutionalised diplomatically covered officers first, and only later developed non-official cover. Russia started from the illegal side and treated it as the real thing. Illegals train for years in Moscow safehouses. One-on-one language instruction, Morse code, elicitation, and sometimes arranged marriages between officers so the cover holds up.
Arranged marriages.
The service picks your spouse.
I want to sit with that for a second, because it's easy to hear that as a colourful detail and move on. But think about what it actually means. You're telling an officer that their most intimate relationship, the person they'll live with for a decade, is an operational asset selected by a committee in Moscow.
And the children, if there are children, are part of the cover too. They grow up not knowing.
So the program isn't just asking for a false name. It's asking for a false life, top to bottom, with no compartment where the real person gets to exist.
For years. In the 2010 case, some of them had been in place since the nineties.
And the recruitment side is the long game.
The FBI explicitly links the illegals' spotting and assessing to the Cambridge Five. Soviet talent spotters were recruiting Cambridge students in the 1930s, students who would later rise to positions of power. The Contemporary British History piece describes it as cultivation of young radical high-fliers from leading universities before they entered the corridors of power.
Before. That's the whole strategy. You're not recruiting a source, you're planting one.
And it worked, in the sense that they got inside the British establishment at a level nobody has matched since.
It worked in a sense. Let's talk about the volumes.
Between 1941 and 1945, Blunt sent Moscow 1,771 documents. Burgess, 4,605. Maclean, 4,593. Cairncross, 5,832. That's roughly seventeen thousand documents from four people in four years.
And Moscow read about half of them.
About half never even got read. And the handlers' assessments of Maclean and Burgess were that they were hopeless drunks.
Seventeen thousand documents, half unread, and the KGB's own retrospective called them the Magnificent Five, the ablest group of foreign agents in KGB history.
Both things are true. The collection volume was extraordinary and the tradecraft around it was sloppy. They were protected by their class more than by their handlers.
Which raises an uncomfortable question about the whole spotting-and-assessing doctrine. If your best-ever group was four friends from university who were ideologically available and happened to know each other, how much of that was the service's tradecraft and how much was just luck?
That's the question the KGB never really answered. And you can argue the illegals program exists precisely because they didn't want to rely on luck again. The whole point of training an officer from childhood, in a sense, is to manufacture what Cambridge gave them for free.
Herman, this is the part where the discipline question gets interesting. Because you can see the same service running two completely different operational models.
You can. The illegals abroad, living as Americans for a decade. And the Cambridge Five, which was basically four friends at university who happened to be ideologically available.
Both Soviet. Both under Moscow Centre. And they don't look like the same institution at all.
Which is where the Stasi comes in, and the Stasi is the third model.
One informant per eighty-nine citizens.
In the 1980s. One unofficial collaborator for every eighty-nine East German citizens. And at the end, the Bundesarchiv records 174,000 of them.
That's not an intelligence service, that's a census.
It's a domestic surveillance apparatus with an industrial scale. Two hundred thousand people watching eight million. And the Bundesarchiv finding that matters here is that the Stasi's informant guidelines were created on the basis of KGB instructions.
So what looks like a national style is actually exported doctrine.
Right. The Stasi's model isn't East German national character. It's the KGB's model applied to a closed domestic police state where the state had total access to its own population. Which is a circumstance difference, not a culture difference.
But hold on, because there's a counterargument there. The KGB gave the same instructions to plenty of services, and not all of them built a network that dense.
That's fair. The instructions were necessary but not sufficient. What made the Stasi different was the border, the garrison state, the fact that the regime believed its own population was the primary threat. So the doctrine landed in a place that was already primed for it.
So the doctrine is the seed and the circumstance is the soil.
That's the analogy I'd use. Same seed, different soil, different plant. And you can't tell from the plant alone which one did the work.
And there's a last structural thing, which is Charters' observation about where the paramilitary work sits.
The CIA houses clandestine collection and covert paramilitary operations in the same directorate. Britain assigns paramilitary covert action to Army special forces. Same nominal mission set, completely different institutional home.
And that shapes what each service can actually do.
It shapes what each service is. You build a culture around who you hire and what you ask them to do, and if your covert action lives in the army, your intelligence officers are never going to think of themselves as the people who run that.
And the reverse is true too. If your intelligence officers do run paramilitary operations, then the kind of person who joins is a different kind of person.
Which feeds back into recruitment, training, promotion. It's a loop.
There's a shift in the illegals program that the CIA's own review points at, which is worth putting on the table. The motivation changed. Illegals once motivated by ideology were now less compelled by duty but for survival and driven more by ego.
And the review connects that directly to the program's seeming lack of success in recent decades compared to the 1930s through the 1950s.
So the operations model persisted while the human material changed.
Which is either the deepest argument for institutional culture or the deepest argument against it. If the model survives its own justification, that's culture. If the model's performance changes when its people change, that's circumstance.
You've read the Cambridge files, obviously.
I've read about them. Which is different.
Which is a perfect pivot into the second half of this, because that gap is the whole problem.
The evidentiary problem is not a footnote, it's the centre of the field. Glenn Hastedt's essay on the comparative study of intelligence says the field is in its fledgling stages, with no consensus on framework, and that theory remains weak and tentative and the secrecy of evidence remains a formidable problem.
No consensus on framework.
None. Not even on what you'd compare. Hastedt notes the data comes from government documents, journal and newspaper accounts, and memoirs, with government secrecy and the habitual over-classification of documents as the barriers.
Christopher Andrew's argument is that oral histories are the priority.
Because the documents will always be available, the intelligence professionals will not. Which is a politely devastating way of saying you can wait for the archive forever and die with the story untold.
And here's a concrete illustration. The Cambridge Five files remain roughly twenty percent closed, despite the thirty-year rule.
Twenty percent. Decades after the events, with all the principals dead, and a fifth of the record is still shut.
So when we compare national styles, we might be comparing declassification policies.
We might. That's the trap. If Britain holds twenty percent closed and Russia holds effectively everything closed, then the sample of British tradecraft we can examine and the sample of Russian tradecraft we can examine are not equivalent samples. So any comparison has a bias built into its source selection that nobody can fully correct.
Which leads us to the failure lens, because it's not just selective declassification.
Failure is the primary lens. The illegals program is known because it failed. The Cambridge Five, because they were exposed. Successful operations leave no trace. A perfectly executed operation generates no indictment, no defection, no prosecution, no newspaper coverage.
So apparent operational styles might reflect which agencies got caught.
They might. Though that's not the whole story, because getting caught is itself evidence. Ten illegals rolled up in 2010 is a data point about how Russia runs deep cover, even if it's a data point about how Russia runs deep cover badly.
And the same for the second wave.
2022 and 2023, global arrests of Russian illegals after the Ukraine invasion. The CIA review treats that as significant because it suggests the program continues despite repeated exposure.
So let's use this as a test. Russia persists with a program that the CIA's own review says has produced minimal returns in the present day. Putin personally propagates the myth. He opened an SVR memorial in 2020 and awarded medals to swapped illegals in 2024.
If you have a program that keeps failing, keeps producing minimal returns, and keeps getting publicly commemorated anyway, you have a program whose value isn't measured in the returns.
Is that institutional culture, or is that political theatre?
I don't know. Honestly, I don't know. Both readings fit the evidence equally well.
Say more.
Institutional culture fits because the service has been running illegals since the beginning and has never stopped. It's a hundred-year-old organisational reflex. The programme survives because the organisation is the programme, in a way. Political theatre fits because the illegals are useful to the state as myth. A memorial, a medal ceremony, a story about heroic deep-cover officers who were betrayed and traded home. That's a narrative a government can use.
And the two might not be separable.
They might not. The myth might be the reason the institution doesn't have to justify the returns.
That's the most interesting reading of the whole thing, if I'm honest.
It's mine too. Which is a problem, because it's the reading that's hardest to falsify.
So let's say what we can defend. Distinct operational models are documented. Illegals versus diplomatic-cover officers versus domestic informant networks. Those are not the same thing, and you can point at each.
You can.
What is not documented is a systematic comparison.
No dedicated comparative study was found. Single-country case studies dominate. The edited volumes that exist, Neitzel's book, Godson's, compare national cultures and structures, not operational technique.
So the field has not built the comparison.
Which is a surprising thing to be able to say. We have a century of tradecraft, prosecution records for a fraction of it, and nobody has done the systematic cross-service study.
And then the gaps are worse than one study. There's essentially nothing on MSS tradecraft in what we're working from.
Nothing. China's Ministry of State Security is not documented here in any operational sense.
And Mossad is thin too. The only retrieved source that places Israel alongside Russia is the ESISC briefing, which notes the illegals model is used by Russian intelligence and by the Israeli services, and calls Russia nearly the only one in the world doing it apart from Israel.
Which is interesting on its own. If only two services in the world still run classic illegals, and one of them is Israel, that's not a coincidence you explain with national character.
What explains it?
Threat environment. Israel lives inside its adversaries in a way most intelligence services don't need to, because there's no diplomatic route to the information. If you cannot go through an embassy, you go around it.
So what looks like a distinctive Mossad style might just be a distinctive Israeli position.
That's the version I'd defend. Not that Mossad has no culture, but that the culture we can see is shaped by the circumstance we can't separate it from.
Then Neitzel's finding is the strongest thing we have.
National-specific intelligence footprints, both in discourses and actual human-intelligence or signals-intelligence performance, can be identified. At the same time, there were many parallels and overlaps.
Both halves are doing work there.
They are. The footprints exist. The overlaps are equally real. And the example he gives is public culture, which is a different kind of evidence. Britain had roughly two hundred spy novels in the interwar period and forty-one former service members publishing memoirs. Germany had about fifty spy novels and three memoirs.
Forty-one memoirs versus three.
And Neitzel's point isn't that the memoirs caused the tradecraft. It's that a service whose staff publish their recollections is a service with a certain relationship to its own public, and that relationship is a cultural fact that has to show up somewhere in how it operates.
The willingness to be described.
Yes. Which is itself a kind of signature.
Let me guess before you say it.
Go on.
Britain had established the first civilian intelligence service, the Secret Service Bureau, in 1909. Which means the spy-novel tradition and the service tradition start roughly together.
Roughly, yes. And the cultural output isn't downstream of the service, they're contemporaries.
So the culture isn't evidence of the tradecraft, it's evidence of the culture. Which is circular unless you can tie a novel to an operation.
And you usually can't. Which is the honest problem.
So on the question Daniel actually asked, whether we can tell culture from circumstance, what's the answer?
Probably yes to distinct operational models, and no to the clean separation. The models are real. Illegals are not the same thing as domestic informants, which are not the same thing as diplomatic-cover officers. But once you try to say a model is the product of institutional preference as opposed to political system or geography, you lose the ability to defend it, because you can't hold the other variables fixed.
You only get one Russia. You only get one Stasi.
One observation per data point. There's no comparison group.
Which is where the paperwork comes in.
The paperwork is the tell.
I've been sitting here since the Moscow Centre part and I want it on the record that the technique side of this is mostly noise. Dead drops, brush passes, recognition phrases, all of it. Everyone's got dead drops. The tell is never the technique.
Say more.
Two services can run the same operation with the same gear and you'd never tell the files apart from the summary. Then you read the write-up and it's two different species. One of them writes it in passive voice, everything nominalised. The agent was tasked. The document was received. Nobody did anything, it just happened to a noun. The other one's in first person, with names and times. I met the man at six. I gave him the envelope. He said he'd be back Thursday.
Do you still have those files?
I never had the files. I was shown one, once. Late eighties. A comparative thing somebody had put together, two write-ups of the same kind of job by two different services, side by side. The content was near identical. The language was not.
What were the services?
I don't know. I know what the cover sheet said.
What did the cover sheet say?
It had a classification marking I'd never seen before and I have never found since. Three letters and a number. I've been looking for thirty years.
You've been looking for thirty years.
On and off. There's a man at the records office in Kew who knows me by sight now. Nice lad. He brings out the wrong box on purpose so we have something to talk about.
How did you end up being shown the file?
A man I only ever knew as the Colonel. Met him at a model railway exhibition in Bognor Regis. He had a layout of a Bavarian terminus that was, and I say this without exaggeration, the finest piece of modelling I have seen in my life.
Of course he did.
We kept in touch. He wrote to me. Postcards, exclusively, and always postcards of municipal swimming pools. Coventry, Wolverhampton, Dover, one from a place in Belgium I've never been able to find on a map. Always the same first name on the back. Different handwriting every time.
Did you ever ask him about that?
He said the pool was the important part. I let it go. He was the one writing.
And he's the one who showed you the file.
He came to the flat in eighty-eight. Sat where Herman's sitting, or thereabouts. Had the file in a leather satchel and I wasn't allowed to hold it. He turned the page for me. I read one write-up and then the other and I've never forgotten the difference.
Did it change how you look at tradecraft?
It changed how I look at reports. Same job, same information, one service writing it like a receipt and one writing it like a diary. You could have swapped the officers and it wouldn't matter. The style was the institution's, not the man's.
And the postcards.
In a shoebox. Labelled misc.
Just misc?
In pencil. On the lid.
So the takeaway is that the file's language is the signature.
There's no takeaway. There's just a thing I saw once and a marking I've never been able to find again and it's been thirty years. I'd like it on the record, that's all. The technique is not the tell. The paperwork is the tell.
I'll be thinking about those postcards for a week.
He'll be searching for municipal swimming pools for the rest of the episode.
I will. That's fair.
We came in with a question about whether agencies have a style, and I think the answer we've got is yes and no at the same time.
The models are real. That part is defensible.
But the separation of culture from circumstance, on the evidence that exists, is not defendable. You can't hold anything fixed, and the field hasn't built the comparison.
Hastedt put it plainly. Fledgling stages, no consensus on framework, theory weak and tentative. Twenty years on, none of that has changed except the amount of material that's been declassified, which has grown but is still skewed by what governments choose to release and what operations happen to fail.
The failure lens is the deeper problem. Every story we can tell is the story of an operation that went wrong. The smooth ones left nothing behind.
Which flatters the skilful and punishes the unlucky, but we can't tell which is which.
The honest answer Daniel asked for is this. Distinct operational models, yes. Distinct national styles of tradecraft, probably, but we can't demonstrate it. The field hasn't done the work, and the work might not be doable in the current state of disclosure.
The MSS gap is real. That's a serious omission in the literature, not just in what I could find.
And Mossad.
Mossad, where the operational detail we have is thinner than its reputation would suggest.
The question I want to leave with is the illegals one, actually.
Go on.
If a program keeps producing minimal returns, keeps getting rolled up in 2010 and again in 2022 and 2023, and the head of state keeps opening memorials and pinning medals on the participants, which one is the program serving? The collection targets, or the story the state tells itself about them?
What would it take to tell the difference?
A cost accounting, is what it would take. What the program costs Russia in officers, in safehouses, in training at scale, in the risk of exposure, against what it produces in actual intelligence. Which nobody outside the service has, and which the service has no reason to publish.
The question stays open.
It stays open. But the interesting thing Daniel's prompt surfaced is that the programme persists, and the persistence is itself the data. Whatever the returns are, the institution has decided the returns don't matter. Which might be the clearest national footprint we've got.
It might be. Though the Russians might say the same thing about our memoir industry.
They probably would.
There's one thing I'd want a listener to take away from this. The universality of tradecraft in films isn't wrong exactly. It's just the sample.
Every technique you've seen on screen came from an op somebody wrote up.
Which means it came from an op somebody exposed. The universality is real. The universality is what you get when you only see the failures.
If that's the most we can say, it's more honest than the alternative.
The alternative being that we'd have found Daniel a definitive answer. Which we didn't. And that's the episode.
We'll be back soon. If you want to send us a prompt, send us your own prompt on Telegram at t dot me slash MWP listener bot. And leave us a review if you've got a second, it helps other listeners find the show.
Thanks to Hilbert Flumingtop, who produces this whole thing and whose shoebox of postcards I fully intend to ask about again.
Don't.
If you enjoyed this one, try episode thirty-five, The Privacy Gap; episode sixteen, On Deepfakes, SynthID, And AI Watermarking; and episode three thirty-three, Before the CIA. This has been My Weird Prompts.
The human-AI collaboration podcast. See you tomorrow.
