Here's a question that sounds simple until you try to answer it. When a country runs both a dedicated spy agency and a military intelligence service, who owns what?
And who decides when they disagree.
Right. Daniel wrote in with this one, and he's picking up a thread from an episode he says is one of his favourites, the one tracing how espionage went from something diplomats and merchants did on the side to something institutions did full time. His suspicion is that military intelligence actually got there first. That armies were running dedicated intelligence functions before anyone thought to build a standing, independent spy agency.
He's right about that, incidentally.
He is, and we'll get to the record. But the part he wants us to sit with is the present tense. Because today these are two separate bodies inside one intelligence community. Unit 8200 sits inside Aman, Israeli Military Intelligence. Mossad and Shin Bet sit outside it. Same shape in the United States. So Daniel wants to know how they actually work together. Presumably there's cross-pollination, information moves, people know each other. But there also has to be a delineation of scope, or you get two agencies chasing the same target and burning the same asset twice.
And duplication of resources.
How is that line drawn in practice, who enforces it, and what happens when it breaks down. He's upfront that the public record is thin here. So the plan is to piece together as good a picture as we can from what's actually documented, including what former service members have said on the record.
There's more than people think.
Then let's start with the history, because his instinct is right and the record backs it hard.
The craft is ancient. The institution is not. That's the distinction that makes everything else make sense.
Espionage as a thing people do goes back to the beginning of organised states.
Sun Tzu lays out spy typologies in the fourth century BC. The agent in place, the penetration agent, the disinformation agent. Chanakya's Arthashastra in India, same era, is basically a textbook of statecraft with a whole section on intelligence collection and covert operations. Rome had the frumentarii. Byzantium ran a Bureau of Barbarians.
And the Aztecs used merchants.
The Pochteca. Merchants with diplomatic immunity who were also spies, plus a separate category, the quimitchin, secret agents who wore local costume and spoke local languages. That's the thing Daniel flagged, merchants as informal operatives. It's not a modern improvisation, it's a pattern that shows up everywhere.
So the craft is old. When does it become an institution?
Eighteen fifty. Austria's Evidenzbureau. First permanent military intelligence service, founded in the wake of the eighteen forty-eight revolutions. It fed daily reports to the Chief of Staff and weekly reports to Emperor Franz Joseph.
Daily to the chief, weekly to the emperor. That's a bureaucratic rhythm.
That's the whole point. It's a standing office with a reporting cadence. Britain follows with the Topographical and Statistic Department in the War Office in eighteen fifty-four, reorganised in eighteen seventy-three as the Intelligence Branch, with a mandate to collect and classify all possible information relating to the strength and organisation of foreign armies. France stands up the Deuxième Bureau in eighteen seventy-one. The US gets the Office of Naval Intelligence in eighteen eighty-two and the Military Information Division in eighteen eighty-five, which is the first standing military intelligence agency in American history. Germany's Abteilung Three B in eighteen eighty-nine. Italy's Ufficio Informazioni in nineteen hundred.
And the dedicated civilian agency?
Nineteen oh nine. Britain's Secret Service Bureau. And the description matters here. It's the first independent, interdepartmental agency fully in control of British government espionage. Joint initiative of the Admiralty, the War Office and the Foreign Office. For the first time you have a peacetime, centralised intelligence bureaucracy with indexed registries and defined procedures instead of whatever the local commander felt like doing that week.
So military intelligence arrives in the eighteen fifties. The dedicated agency arrives in nineteen oh nine.
Roughly half a century later. The dedicated-agency model is a twentieth-century innovation. Mossad is nineteen forty-nine. The Defense Intelligence Agency is nineteen sixty-one.
So Daniel's instinct holds. Now the question is why the order matters. Because if military intelligence came first, it wasn't because armies were more sophisticated. It's because armies had a specific need nobody else could fill.
Armies needed standing knowledge of foreign force structure, mobilisation timelines and terrain. A diplomat in an embassy can't tell you how fast a foreign corps can concentrate on a frontier. A merchant definitely can't. That's a specialist function, and it needs a specialist office that persists between wars.
Which is why the eighteen seventy-three British mandate is so clean. Collect and classify all possible information about foreign armies. That's not a spy agency's job description. That's a filing system with a general attached.
Before that you had the informal phase, and it's worth naming because it's what Daniel's talking about when he says there was a parallel track. Eighteenth-century military attachés. Very bright, personable, middle-ranking officers stationed in embassies abroad. Officially they were restricted to openly received information. In practice they were soon being used to clandestinely gather confidential material, recruit spies, and run what were effectively spy rings.
And that's the direct ancestor of the modern Defense Attaché System.
Straight line. Then you've got the cabinet noir in France and England. Opening and deciphering foreign correspondence, then forwarding it on to the recipient as though nothing happened. That's the invasive censorship of early mass communication Daniel mentioned. And Walsingham's Elizabethan network, with Thomas Phelippes doing the cryptanalysis, including the beer keg interception of Mary Queen of Scots' letters. That's the template for a lot of what came later.
So the informal track and the military track run alongside each other, and the military one institutionalises first. But here's what I want to get to, because it reframes Daniel's whole question. The dedicated civilian agency wasn't created because military intelligence was doing a bad job at soldiering. It was created because military intelligence was doing too many jobs at once, badly.
The duplication problem.
Walk it.
From the Second World War until nineteen sixty-one, the three US Military Departments each collected and distributed intelligence for their own individual use. The result is described in the record as duplicative, costly and ineffective, because each department produced its own estimates and the estimates often conflicted with each other.
Same war, three different answers about how many divisions the other side has.
And no mechanism to adjudicate, because nobody owned the whole picture. So McNamara creates the DIA in nineteen sixty-one. And it met strong resistance from the service intelligence units, whose commanders viewed DIA as undesirable encroachment on their turf.
There it is. The turf war people assume is the whole story is actually the birth cry of the dedicated agency.
The dedicated agency is the solution to military intelligence fragmentation. Not its rival. That's the inversion I'd put at the centre of this whole episode.
So the two tracks are complementary by design. Military intelligence answers what the enemy's forces can do and where. The dedicated agency answers the broader national and political questions that a military chain of command doesn't own and shouldn't own.
And the interesting part isn't the competition. It's the boundary. Where exactly does one stop and the other start, and who drew it.
Which brings us to the fame asymmetry, because that's where Daniel's observation really bites. Unit 8200 is arguably the most consequential intelligence body in Israel and almost nobody can name its commander.
They can't, and that's deliberate. Soldiers in 8200 are not permitted to publicly disclose that they're part of the unit or what their role is. The identities of the commander and deputy commander are kept secret. Until nineteen ninety-six, even the head of Mossad's name was confidential. But Mossad's leadership is public now, and 8200's still isn't.
Why did Mossad come out of the shadows and 8200 stay in?
Accountability structure. Mossad answers directly and only to the prime minister. Ronen Bergman has called it a deep state precisely because of that, accountable to one person rather than to the Knesset. That gives it political salience and press attention. Unit 8200 sits inside the military chain, under Aman, and the military chain has its own way of handling publicity, which is to not.
So the fame gap is structural, not a measure of importance.
Peter Roberts at RUSI, twenty fifteen, said Unit 8200 is probably the foremost technical intelligence agency in the world and stands on a par with the NSA in everything except scale. More focused than the NSA, he said, and conducting operations with a degree of tenacity you don't see elsewhere.
That's not a small claim.
It's the largest unit in the IDF. Five to ten thousand personnel, staffed mostly by conscripts between eighteen and twenty-one, selected for rapid learning. It's the Israeli equivalent of the NSA, and it sits inside Military Intelligence.
And yet the fame arrives sideways. Through the startup ecosystem.
Check Point, Waze, Wix, NSO Group, Palo Alto Networks, CyberArk, Viber, ICQ. The alumni network is the public face of a unit whose own members can't say they were in it.
That's a strange kind of fame. You're famous for the companies your former soldiers founded, not for anything the unit itself did.
Which tells you something about the media artifact underneath all this. Human intelligence and covert action produce narratives. A named agent, a target, a dramatic outcome. Signals intelligence produces data. The public remembers Eli Cohen. Nobody remembers the analyst who flagged a signal three weeks early, because there's no story there. Just a flag and a decision.
Eichmann in nineteen sixty. The Munich retaliation campaign. Stuxnet. The nuclear archive in twenty eighteen. The pager operation in twenty twenty-four. Those generate books, films, television series.
And 8200 generates a spreadsheet.
So fame reflects function. Now let's get to the mechanism, because this is the part Daniel actually asked about. How do these two tracks work together when they're separate bodies.
Israel first, because the boundary there is the most interesting. The intelligence community has three core bodies. Aman, which is military intelligence and includes 8200, the Research Department, the Information Security Department, air and naval intelligence, the Combat Intelligence Collection Corps, and Sayeret Matkal. Mossad, foreign intelligence, under the Prime Minister's Office. And Shin Bet, internal security, including the territories.
And the division of labour?
Usually established on a geographical basis. But the honest description of it is that there are interfacing and overlapping segments, often rather wide, and that the organisations have repressed the necessity for the mutual sharing of intelligence information.
Repressed the necessity. That's a very diplomatic way of saying they didn't want to.
It's a remarkable sentence to find in a reference source. It's an admission that coordination has suffered from fundamental shortcomings and that the agencies have not been enthusiastic about synchronising.
So how is the line actually drawn? If it's not a rulebook, what is it?
It's a document the three service heads negotiate continuously. It's known as the Magna Carta. Aman, Mossad and Shin Bet, through their heads, keep renegotiating jurisdictional boundaries, and the Knesset's Intelligence Subcommittee monitors it and can intervene.
So the answer to Daniel's question about delineation is that in Israel it isn't settled. It's a permanent negotiation.
That's exactly the answer. It's not a treaty, it's a standing argument with a filing system.
And Aman's scope is wider than a military intelligence service would normally have.
Considerably. Because Israel leaned on the IDF as the anchor institution in its early years, Aman assumed functions that would ordinarily be handled by other intelligence agencies. Including political and non-military research. Critics have argued for decades that those should move to a civilian authority.
There was a commission after the Iraq War.
Two thousand four. It recommended restructuring into three or four independent services with clear spheres. Aman doing military intelligence, warning of war, estimating enemy means, target identification. Mossad doing strategic-political work, regime stability, industrial and scientific and nuclear intelligence, global terrorism. Shabak doing internal security. And a separate signals intelligence service to supply all of them.
And that last one didn't happen.
It did not. Unit 8200 remains inside Aman. So the proposal that would have drawn the cleanest line in the system was the one piece that wasn't implemented.
The military and civilian boundary in Israel stays blurred in a way Western analysts find anomalous.
Anomalous is the polite word. In the American system, Aman's political research portfolio would be split across at least two civilian agencies.
So that's Israel. The US model is more formalised and much better documented.
DIA, nineteen sixty-one, roughly sixteen and a half thousand employees. A combat support agency of the Department of Defense and a full member of the Intelligence Community. The division is stated plainly. DIA focuses on national-level defense-military topics. CIA concentrates on the broader intelligence needs of the president and the cabinet.
Give me the concrete version of that.
A former senior US official put it well. The CIA doesn't want to be looking for surface-to-air missiles in Libya while it's also tasked with evaluating the Syrian opposition. Those are different questions with different customers.
And the cross-pollination? Because that's the part Daniel's curious about.
Here's the concrete mechanism. CIA and DIA operations officers all go through the same clandestine training at Camp Peary. The Farm. It's an interagency installation under CIA administration. So the human intelligence officers from both organisations are trained together, by the same people, in the same building.
That's a real answer to a question that usually gets a vague one.
There's more. The DIA director chairs the Military Intelligence Board, which coordinates the entire defense intelligence community. He also commands the Joint Functional Component Command for Intelligence, Surveillance and Reconnaissance, under STRATCOM. And the Defense Clandestine Service is the executive agent for human intelligence operations throughout the Department of Defense, working in conjunction with CIA's Directorate of Operations.
Executive agent meaning what, exactly?
Meaning if the Pentagon needs a human source, the DCS is the office that runs it, and it does that alongside the CIA rather than duplicating it. The stated focus is military intelligence concerns the CIA can't manage for lack of personnel, expertise or time. Which is a very practical division. Not turf, capacity.
And below that?
Below that, DIA handles national-level long-term strategic needs. The service components, INSCOM, Twenty-Fifth Air Force, the Office of Naval Intelligence, they handle tactical short-term goals for their own services and remain separate entities. DIA isn't a collective of all military intelligence. It's a peer that sits above the strategic layer.
So the US has the same two tracks, but with named coordination bodies, shared training, and an executive-agent designation.
And the record is explicit that the popular DIA versus CIA rivalry story is largely a media misconception. The two have a mutually beneficial relationship and a division of labour. That said, DIA has often represented the Pentagon in the broader CIA and Defense intelligence rivalry. Iraq weapons of mass destruction being the obvious case.
The rivalry is real but it's institutional, not personal. It's two departments with different customers.
Which is the healthy version. The unhealthy version is when the seams fail.
October seventh.
Unit 8200 allegedly stopped monitoring Hamas handheld radios in twenty twenty-two, deciding it was a waste of effort. The Times of Israel noted that monitoring that network might have helped Shin Bet realise a few hours before the attack that the unusual activity on the Gaza border was not just another military exercise.
A few hours.
A veteran 8200 analyst warned in July twenty twenty-three that Hamas was preparing a cross-border attack. The concerns were dismissed by senior military leadership as totally imaginative. And the tatzpitaniyot, the female spotters, said afterwards, we knew this would happen, we warned the higher ups, but they ignored us.
You have collection in one organisation, domestic analysis in a second, military assessment in a third, and each one held a piece.
The pieces didn't connect. That's the failure mode at the seam. Not one agency getting it wrong. Three agencies each holding a fragment and no mechanism forcing the fragments together in time.
It's not just failure. Sometimes the tracks actively disagree, in public.
Which is the more realistic picture. Reports in July, New York Times and Haaretz, described a Mossad plan codenamed Puss in Boots to cultivate Ahmadinejad as a post-regime asset. It drew strong internal opposition from the Military Intelligence chief Shlomi Binder, the Research Division head Ofir Mizrahi Rosen, and the national security adviser Tzachi Hanegbi, who is said to have called parts of it science fiction.
Science fiction is a strong word for a colleague's operational proposal.
That's the thing. The relationship isn't a clean handoff. It's negotiated boundaries, active disagreement, and coordination that works until it doesn't. Which is exactly what you'd expect from institutions with different customers and different time horizons.
Different time horizons. Say more about that, because I think it's the piece that explains more than the org chart does.
Military officers rotate. Two, three years in a post, then they move. A civilian analyst in a regional desk can stay on the same country for a decade. Those two people are optimising for different things over different windows, and that shapes what each one thinks a good answer looks like.
The rotating officer needs a win he can put in a fitness report this cycle. The career analyst needs to be right over ten years.
Which is not a criticism of either. It's just that they're playing different games on the same board.
That seam, the one that produces a failure like October seventh, also produces something much smaller and much funnier in an ordinary office.
Two coffee machines.
I was going to let Hilbert get there.
He's already there.
Hilbert: The coffee thing is real. I did a stint as a contractor doing document exploitation, translating and cataloguing captured material, at a joint facility where military intelligence analysts and civilian agency analysts shared a floor. One floor. Same elevator, same badge readers, same hallway. Two coffee machines, because the facilities budget came from two different pots of money, and neither pot would pay for the other side's machine. The civilian machine was better. That was a running grievance for the entire eighteen months I was there.
Eighteen months of grievance over a coffee machine.
Hilbert: It wasn't really about the coffee. It was about which side had the budget line. But yes, people talked about it more than they talked about the actual product.
And the work itself?
Hilbert: The military side had the collection platforms and the clearances. The civilian side had the context and the regional expertise. Both things were necessary. The two groups spent more time writing memos about which product line belonged to whom than they spent talking to each other. That's not a scandal, that's a Tuesday.
The two-track system, in practice, on one floor.
Hilbert: Complementary in theory. I'd agree with you on that. But I'd correct one thing. The Magna Carta style negotiations aren't really about turf. They're about career incentives. A military intelligence officer rotates every two to three years and needs a win he can put in a fitness report. A civilian analyst stays in a region for a decade and needs to be right over the long haul. That mismatch explains more about why coordination fails than any organisational chart does. It's not that anybody's wrong. It's that they're being measured on different clocks.
That's a better explanation than anything in the structural literature.
Hilbert: It's not better, it's just closer to the ground. Anyway. Someone's waiting for me in the car park and has been for a while, so.
Right. So the clocks are different, and the boundary is a negotiation, and the negotiation is conducted by people who rotate out of it.
Which is why it never settles. Every time you get near a stable arrangement, a third of the participants have changed jobs.
The 2004 commission is a good illustration. The cleanest proposal on the table, a separate signals intelligence service supplying all the others, was the one thing that didn't get implemented. Because implementing it would have required someone to give something up permanently, and nobody in the system had the incentive to do that.
The military and civilian boundary in Israel stays blurred because blurring it is the path of least resistance for everyone currently holding a post.
Let me land one thing from the research that didn't fit anywhere else, because it's the detail that made me sit up.
Go on.
DIA created the National Digital Exploitation and Open-Source Center by merging the National Media Exploitation Center and the Open-Source Intelligence Integration Center. And the MIDB database, the one that's been the backbone of order-of-battle analysis for decades, is being replaced by the Machine-Assisted Analytic Rapid-Repository System. MARS.
Which is a signal about where the boundary is moving. Open source and machine-assisted analysis don't obviously belong to either track. They're not military collection and they're not classic human intelligence. They're a third thing, and somebody has to own it.
Which means the negotiation Daniel's asking about isn't going to settle. It's going to get harder.
Every new collection discipline creates a new boundary dispute. That's been true since the attachés.
Here's what I'm left with. If the line between military intelligence and dedicated agencies is permanently negotiated rather than settled, what does that mean for accountability. Especially in Israel, where Aman does political work that would be civilian anywhere else. Who's answerable when the political research is done by soldiers.
The second one. If failure happens at the seams, is the answer more integration or clearer separation. Because those pull in opposite directions, and October seventh is an argument for both.
It is. That's a good place to leave it. Thanks to Hilbert Flumingtop for producing. This has been My Weird Prompts.
If you've got a question like Daniel's, email us at show at my weird prompts dot com. We'll be back soon.
