August 12th, 1996, Vienna, Leopoldstadt District.
A man named Dariush Karimi sits at a drafting table and draws a line that will kill many people a decade later.
>> He does not know this.
He is an asset of Israeli intelligence.
The flaw he is drawing into the schematic is supposed to make Iranian missiles fail, to set the program back, to leave no trace.
Iranian engineers found the flaw years later.
>> [music] >> They spent 11 weeks trying to correct it.
Correcting it led them somewhere no one in Tel Aviv had anticipated.
A new combustion cycle, 40 extra kilometers of range, less visible on radar.
For Iran, it was an accident.
For Israel, it was a catastrophe with Mossad’s own signature on it.
>> [music] >> How does a deliberate flaw, drawn by hand in Vienna, become the breakthrough an Iranian weapons program had been searching for? And how does an intelligence service >> [music] >> spend seven years looking away from what it built? What you are about to hear is the most expensive mistake in the history of intelligence disinformation, and the one no one has ever officially admitted.
Subscribe if you want more Mossad operations that sound impossible.
They were real.
Real people, real decisions, real >> [music] >> consequences.
Most are still classified.
The ones that aren’t are stranger than the ones that are.
It gets darker.
In the spring of 1993, the Iranian aerospace engineering community sent a delegation to a conference in Vienna.
It was not a large conference, 47 registered attendees, papers on combustion efficiency, propulsion systems, turbofan design.
Mossad had two people in the building.
One was watching the delegation, the other was watching one specific member of it, a propulsion engineer from the Isfahan Aerospace Research Center, named Dariush Karimi, who who had submitted a paper on fuel feed optimization in compact propulsion systems.
The paper had been read before Karimi boarded his flight.
It had been read carefully.
Karimi was 34 years old.
He had been at the Isfahan facility for 6 years.
He spoke German well enough to ask for directions.
His English was better than that.
After his presentation, 11 people in the room, polite questions, no follow-up offers.
He went to the hotel bar.
A man named Daniel sat down next to him.
Daniel introduced himself as a consultant for a European Aerospace firm.
He had specific questions about the paper.
Specific enough to be flattering.
[music] Karimi answered them.
Daniel asked more.
By the second drink, the conversation had moved away from turbofan design and towards something harder to name.
The gap between what Karimi knew and what the people above him were willing to use.
“Do they understand what you have built?” Daniel asked.
Karimi looked at his glass.
“Some of it.
” Daniel nodded slowly as if that were exactly the answer he had been waiting to hear.
With the particular loneliness of a man who believed his work mattered and could not get anyone in his immediate vicinity to agree, Karimi answered every question Daniel asked that evening.
He would answer more of them over the following 6 months.
Over the following 6 months, Karimi and Daniel met four more times.
Frankfurt, once.
Vienna, twice.
Once in a rented apartment in Geneva that Karimi was told belonged to a friend of Daniel’s.
The meetings were long.
They involved wine and technical drawings.
And eventually, a number that Daniel wrote on a piece of paper and slid across the table without comment.
Karimi did not pick it up immediately.
He looked at it for a while, then he folded it and put it in his pocket.
He was not asked to steal anything.
Not then.
He was asked to answer questions.
Detailed questions about specific systems.
The kind of questions that told him, without being stated, that the people asking them already knew a great deal and were filling in specific gaps.
By the end of 1993, Dariush Karimi was on Mossad’s payroll.
His official designation was a string of letters and numbers that appeared in no file he would ever have access to.
His handler changed twice over the following 2 years.
The operation that would use him took 2 more years to design.
[music] It had a name inside the Technical Intelligence Division.
Operation Closed Circuit.
It described the mechanism.
A loop of false information that would feed back into the Iranian program and cause it to fail from the inside.
The target >> [music] >> was the propulsion system of a short-range ballistic missile the Iranians were developing for deployment in the Western theater.
The missile itself was not new.
The propulsion system was.
Iranian engineers at the Isfahan facility had been working on a modified fuel feed architecture that was more compact and more reliable than the previous generation.
If they completed it successfully, the missile’s operational range would extend far enough to reach targets [music] that had previously been outside its envelope.
Mossad wanted that program to fail.
>> [music] >> The difficulty was not finding the flaw.
The difficulty was making it undetectable.
Designing an error that would pass every quality check the Iranians ran, survive laboratory testing, and only reveal itself under the specific conditions of an operational launch sequence.
That required someone who understood the system from the inside.
That was what Karimi was for.
In January of 1996, Karimi was brought to a safe house outside Tel Aviv.
The drive from Ben Gurion Airport took 40 [music] minutes.
He was not told the address.
The room was small, a folding table, two chairs, a whiteboard with nothing written on it.
A man named Avi was already there with a folder open in front of him.
Avi was not an engineer, but the folder contained schematics Karimi recognized, drawings from the Isfahan facility, details he had helped produce.
He did not ask how Avi had them.
Avi placed one drawing on the table.
“You know this system.
” Avi said.
It was not a question.
Karimi looked at the schematic.
Fuel feed manifold, short-range ballistic configuration.
Third generation casing, modified inlet geometry.
He had seen versions of this before.
“Yes.
” Avi placed a second drawing [music] next to the first.
The two schematics looked identical at first reading.
Karimi leaned forward.
He looked at the nozzle assembly.
He looked at the inlet geometry.
He found the difference.
It was in the angle of the secondary nozzle throat.
Less than 1° of deviation from the specification on the first [music] drawing.
In a controlled environment, a component built to the second specification would perform within acceptable parameters.
The variance would not register on standard quality checks.
But under the stress of a full operational burn cycle, the deviation would introduce an inconsistency in fuel atomization.
Fuel flow would become irregular.
Across multiple burns, the [music] irregularity would compound, the combustion cycle would degrade, the missile would fail.
Not catastrophically, not in a way that destroyed the component and left clear evidence of a design flaw.
It would fail in the way things fail when nobody is quite sure what caused it.
A slightly short burn, a guidance deviation, a test result [music] that did not match the model.
The kind of failure that makes engineers doubt their assembly [music] process before they doubt their design.
“How long before they find it?” Karimi asked.
Avi folded his hands on the table.
“We estimate [music] 2 years, possibly 3.
” Karimi looked at the second drawing.
“And if they find it faster?” “Then they find a schematic with an error in it.
” Avi said.
“Engineers make errors.
” 6 months later, Karimi was alone in his apartment in Vienna.
The drafting table had been his for 11 years.
He had bought it second-hand from a student moving back to Graz, and it had survived four apartments since then.
The surface was marked with the faint indentations of old drawings.
The angle lamp above it had a broken hinge he had fixed twice with electrical tape.
On the table was a single sheet of A3 graph paper, blank.
He did not keep a copy of Avi’s drawing.
He had been instructed not to keep a copy, and he had not.
What was in front of him had to come from his own hand.
A photocopied schematic from an unknown source created questions.
A hand-drawn schematic from a trusted engineer at the Isfahan facility created trust.
He picked up the pencil.
He drew the manifold housing, then the inlet geometry, then the nozzle assembly, and the secondary throat at the angle that was not quite right.
His hand did not hesitate.
He had drawn the deviation so many times in practice that it no longer required thought.
When he finished, he looked at the sheet for a moment.
Then he set the pencil down.
He put the sheet in a Manila envelope, sealed it, and placed it in the inside pocket of his travel bag.
Three days later, he booked a flight to Tehran.
Karimi arrived in Tehran on the 2nd of October, 1996.
He stayed at a hotel near Vali-e Asr Square that he had used on two previous visits.
The room was on the fourth floor.
The window faced a courtyard.
He had requested that room specifically.
From the window, he could see the entrance of a tea house that had a landline telephone in the back room.
He had used it on a previous visit.
He did not intend to use it on this one.
Knowing it was there was enough.
The first meeting was not at the facility.
Meran, the contact in the documentation unit, >> [music] >> had suggested they meet first at a tea house on Enghelab Avenue.
“Routine,” he said.
“A chance to review the submission protocols before the formal handoff.
” Karimi agreed.
They met on a Tuesday afternoon.
Meran ordered tea and talked about the new logging procedures the facility had introduced in October.
He was not suspicious of anything.
He was simply a man who liked procedures to be followed correctly.
Karimi listened and said the right things.
He had the envelope in his jacket.
He did not take it out.
They agreed to complete the formal submission the following week.
The second meeting was at the facility.
Karimi signed in at the outer gate, received his visitor badge, and was escorted to building four by a junior administrator who did not speak during the walk.
The documentation unit was on the second floor.
Meran was at his desk.
The submission process took 22 minutes.
Forms, signatures, a brief review of the materials list.
Karimi handed over the first two sheets, the manifold housing diagram and the inlet geometry schematic.
He said the third sheet, the nozzle assembly specification, was still being prepared.
A calibration issue with one measurement.
He would bring it in two to three weeks.
>> [music] >> Meron wrote this down without comment.
That night, in the hotel room, Karimi sat on the edge of the bed and looked at the third sheet.
He had not planned to withhold it.
The plan had been to deliver all three together, but in the car on the way to the facility that morning, something had shifted in him.
A calculation he could not fully articulate about the weight of what he was doing >> [music] >> and what it would cost him if anything went wrong.
He had held the third sheet back almost instinctively.
He told himself it was operational caution.
He put the sheet back in the envelope and placed the envelope under the mattress.
He flew back to Vienna the following morning.
The three weeks in Vienna were not quiet.
He had two calls with his handler, brief, procedural, nothing that required a decision.
He redrew the nozzle assembly specification once from memory, alone at the drafting table, to make sure he still had it right.
He did.
He returned to Tehran on the 5th of November.
The third meeting was scheduled for the 7th.
Karimi went back to the facility.
He signed in at the outer gate.
He walked to building four.
He reached the internal security checkpoint on the third floor.
>> [music] >> The guard at the checkpoint was not the same man as before.
His name tag said Miri.
He was methodical in a way the previous guards had not been.
He asked Karimi to open his jacket.
He saw the envelope.
Miri removed the single sheet inside.
He looked at it for a long time.
It was the nozzle assembly specification, the most complex of the three, with the secondary throat geometry and the angle that was not quite right.
Miri was not an engineer, but he had been doing checkpoint inspections for four years and he knew what detailed technical drawings looked like and he knew when something in a submission did not match what had come before.
He picked up the previous submission log.
He compared the materials list from the second meeting against what he was holding.
The first two sheets cover the manifold housing and inlet geometry.
He said, >> [music] >> “This is the nozzle assembly.
” “That is correct.
” Karimi said.
“The nozzle assembly specification is typically submitted with the inlet geometry.
They are part of the same document set.
” “There was a calibration issue.
” Karimi said.
“It required additional review before submission.
” Miri looked at him.
Then he looked at the sheet again.
He placed it on the desk.
>> [music] >> He picked up a form.
A different form from the standard submission log with a red border at [music] the top.
He began filling it out.
Karimi understood what the red bordered form meant.
He had been briefed on facility security procedures.
[music] The red form was a secondary review request.
It meant the material would be held at the checkpoint [music] for inspection by the technical security unit before being forwarded to the documentation unit.
It meant the sheet would not reach Meran that day.
It meant Karimi would leave the facility without completing the delivery.
He stood at the checkpoint for another 4 minutes while Miri finished the form.
Miri handed him a receipt.
He told Karimi the sheet would be available for collection or forwarding within 5 to 7 business days [music] pending technical review.
Karimi took the receipt.
He thanked Miri.
[music] He walked back to the elevator.
He did not go to Meran’s unit.
There was nothing to deliver.
That evening, back in the hotel room, Karimi sat at the small desk by the window [music] and thought through what had happened.
The third sheet had been detained.
This was not catastrophic.
It would pass review and reach the engineering team within 2 weeks.
The operation was not compromised.
>> [music] >> But the detention meant the nozzle assembly specification would arrive at the engineering team with a security annotation [music] attached.
Engineers noticed annotations.
They paid attention to documents that had been held and cleared.
They asked why.
There was no clean answer to that question.
Karimi flew out of Tehran the following morning.
He was at Mehrabad Airport at 6:00 >> [music] >> waiting for the first flight west when Miri arrived at his desk for the day shift and opened his checkpoint log.
They were in the same building for exactly 40 minutes.
Karimi sent the prearranged signal from the inner 3 days later.
The message said the submission had been completed.
The third sheet had been held for review and would be forwarded through the facility’s internal procedures.
His handler replied in four words.
Understood.
>> [music] >> Stand by.
The nozzle assembly specification sat in Miri’s tray for 14 days.
The technical security reviewer found nothing irregular.
The sheet [music] was forwarded to Maran on November 22nd.
Maran logged it and routed the complete three-sheet package to the propulsion engineering team.
In the weeks between Karimi’s departure and the arrival of the full package, the propulsion team had not been idle.
They had the first two sheets.
That was enough to begin modeling the upstream components of the fuel cycle.
When the nozzle assembly finally arrived, the team had already built assumptions about how the interface would connect.
Those assumptions would shape every reading of the third sheet that followed.
The package arrived on the desk of the team’s lead engineer on the 26th.
He placed all three sheets in a row.
>> [music] >> The two earlier sheets were still crisp.
The nozzle assembly specification had a faint crease from where it had been folded during the security review.
He smoothed it flat.
The secondary throat geometry was drawn with a precision that made the first two sheets look preliminary by comparison.
He looked at the point where the nozzle assembly was supposed to connect with the inlet geometry from sheet two.
The interface did not resolve the way he expected.
He made a note in his working file.
Recheck nozzle inlet interface geometry.
Below it, a second note.
Why staged [music] delivery? He set the sheets aside and moved on to other work.
In the checkpoint log on the third floor, Miri’s end-of-shift entry recorded the incomplete original submission, the secondary review request, and the clearance.
[music] In the remarks column, source appeared uncertain regarding submission timeline.
Recommend routine vendor file review.
No one read that line for 11 weeks.
The propulsion team at the Isfahan Aerospace Research Center spent the first quarter of 1997 trying to make the three sheets work as a single system.
They could not.
The manifold housing and the inlet geometry connected cleanly.
That part was not the problem.
The problem >> [music] >> was the nozzle assembly.
Specifically, the secondary throat geometry and the angle at which it was supposed to interface with the inlet specification from sheet two.
The angle did not match.
Under modeling conditions, it produced an inconsistency in fuel atomization that compounded across the burn cycle and degraded output by a margin no one could account for.
The team ran the numbers 11 times across 11 weeks.
The third attempt used a modified nozzle throat with adjusted tolerances.
The fuel flow improved.
The deficit remained.
The fifth test changed the manifold inlet radius.
By the eighth week, three members of the team had begun to discuss, quietly, whether the external source had made an error.
The lead engineer did not join those conversations.
They brought in an external reviewer from the propulsion faculty at Sharif University, who spent [music] two days with the drawings and left without a conclusion.
The lead engineer filed three internal progress reports, each one describing the same problem in different [music] language.
The problem did not change.
It was in the eighth week that Farid Mohammadi asked [music] to look at the submission file.
Mohammadi worked in the adjacent systems unit, which handled integration modeling for the guidance and control components.
He was not part of the propulsion team.
He had heard about the nozzle problem the way people in small facilities hear about things, through the wall, over lunch, in the corridor outside the documentation unit.
He had not paid close attention.
>> [music] >> Then, one afternoon, he asked the lead engineer if he could see the original submission documents.
The lead engineer gave him the folder.
Mohammadi took it back to his desk.
He read the submission log.
He read the security annotations.
The red form detention on November 7th, the clearance 14 days later, the note about staged delivery in the remarks column.
He read the lead engineer’s marginal notes on the nozzle assembly specification.
He looked at sheet three for a long time.
Then he went back to the submission log and looked at the dates again.
The first two sheets had been submitted in mid-October.
The third arrived three weeks later.
The source had cited a calibration delay.
But the third sheet was not a continuation of the first two.
>> [music] >> It was drawn with a different hand pressure, a different density of line, a level of detail the first two sheets did not approach.
An engineer who had produced sheets one and two would not produce sheet three three weeks later as a routine follow-up.
He would have produced it in the same session, at the same desk, with the same pencil.
That night, Mohammadi stayed at the facility past midnight.
He had his own drafting paper.
He worked through the problem from first principles, not as an engineer trying to reproduce the source’s design, >> [music] >> but as one trying to solve the problem the source might have been solving when he drew sheet three.
By 2:00 in the morning, he had a schematic that was different from all three original sheets, but derived [music] from all of them.
He photographed it with his personal camera.
He put it in his coat pocket.
He went home.
The following afternoon, he brought the folder back to the lead engineer.
“What if the third sheet is not a continuation?” he said.
“What if it is an alternative?” The lead engineer looked at him.
“The source had two approaches,” Mohammadi said.
“He submitted the first approach in October.
Three weeks later, he submitted a competing design for the nozzle assembly, one he had developed separately.
He never indicated they were meant to connect directly.
” The lead engineer was quiet for a moment.
“Then the interface problem is intentional.
” “The interface problem is a feature,” Mohammadi said.
“The two sets of geometry are not supposed to merge.
They are supposed to run in parallel.
” >> [music] >> He took a sheet of paper from the desk and drew two lines that did not converge.
Between them, a narrow channel.
A dual circuit feed.
“The inlet geometry from sheet two handles primary combustion.
The nozzle assembly from sheet three handles a secondary ignition cycle, a staged burn [music] that activates after the primary cycle reaches peak pressure.
The inconsistency you have been trying to correct is the gap [music] between the two circuits.
It is not a flaw.
It is the gap through which the second circuit breathes.
” The lead engineer looked at the drawing on the desk.
>> [music] >> He looked at sheet three.
He looked at sheet two.
“If you are wrong,” he said, “we have lost [music] six months.
” “If I am right,” Mohammadi said, “we have gained something the source never intended to give us.
Build it.
The lead engineer said.
It took 4 months to model and 4 months to test.
The first full test of the dual circuit system was run on a cold morning in late November on a test stand at an outdoor facility south of Isfahan.
Mohammadi was there.
The lead engineer was there.
Seven other people were there.
The ignition sequence ran to completion.
The secondary cycle activated [music] at the correct pressure threshold.
The output readings were taken.
Nobody said anything for approximately 30 seconds.
Then the lead engineer walked to the data terminal and looked at the numbers.
He stood there for a long time.
The program director was briefed on December 3rd.
He read the test [music] results for 20 minutes without speaking.
He asked two questions about the radar suppression data.
He approved the next phase of development without further discussion.
He did not ask how the team had arrived at the dual circuit configuration.
Engineers were not expected to explain their methods, only their results.
The system produced results that exceeded the program’s original targets by margins that required a new set of benchmarks.
40 extra kilometers of range, the radar signature at launch, the thermal and electromagnetic bloom that made missiles visible to early warning systems in the seconds after ignition, reduced by just under a third.
The staged burn suppressed it.
The second ignition happened inside the primary exhaust envelope, invisible to sensors calibrated for a single [music] combustion event.
On the wall of Mohammadi’s workspace, pinned above his drafting table, was a large sheet of paper that had started as a copy [music] of Karimi’s original nozzle assembly specification and had become something else entirely.
The secondary throat geometry had been crossed out and redrawn four times.
The margins were covered in calculations and arrows and three different colors of ink.
The original lines from sheet three were still visible underneath, but they were now the least visible thing on the page.
By the following year, the design had been handed to the manufacturing division.
By the end of 1999, the first production units had been completed and passed acceptance testing.
In the spring of 2000, a convoy of sealed trucks left a facility in the western Iranian province of Kermanshah and drove northwest.
It crossed into Syria through a passage that had been in use since the early 1990s.
>> [music] >> It did not cross at a designated border point.
The trucks arrived at a storage facility in the Bekaa Valley on a Tuesday morning.
A logistics officer named Walid was [music] there to receive them.
Walid had been receiving deliveries in this valley for nine years.
He knew what different systems looked like.
He signed the transfer documents.
He walked the length of the storage room while two men opened the first crate.
The casing was gray.
The stenciled markings were in Farsi.
The dimensions were different from the previous deliveries.
Longer by approximately 20 cm with a wider exhaust housing at the rear.
He looked at the exhaust housing for a moment.
Then he signed the second form and walked [music] back to his office.
He filed the transfer documents under the date and went back to his other work.
Outside, the trucks were already leaving.
Tel Aviv, March 2003.
The technical intelligence division occupied the fourth floor of a building that did not have a sign on the front.
Yael B had worked there for six years.
Her desk was the second from the window.
She had a view of a parking structure.
>> [music] >> She was building a technical profile of the Fateh-110.
The Fateh-110 was an Iranian short-range ballistic [music] missile that had been in the intelligence community’s awareness since the late 1990s.
It was not new.
What was new was a set of intercept data from the previous autumn.
Signals intelligence collected during a series of Iranian test launches that had produced readings inconsistent with the known specifications of the system.
The radar signature at launch was lower than the models predicted.
The range measurements [music] were longer.
Both anomalies were within the margin of instrumentation error.
The previous analyst on the file had noted them and moved on.
Yael did not move on.
She pulled the full signals file.
She pulled the previous technical assessments going back to 1999.
She pulled the original source reporting on the Fateh-110 propulsion system from the late 1990s.
She read everything.
The anomalies were consistent across three separate test events.
They were not instrumentation error.
Something in the propulsion system was producing a lower ignition signature than a standard single burn combustion cycle would generate.
[music] The range extension was a related effect.
The same modification that suppressed the ignition signature was also increasing effective burn efficiency.
She wrote that conclusion in a working note and looked at it for a long time.
She went to the technical [music] archive.
The archive was a set of filing cabinets on the third floor supplemented by a digital index that had been built in 2001 and was already partially incomplete.
She searched by subject.
Propulsion modifications.
Iranian tactical systems.
Dual circuit combustion.
She found three documents.
None of them were relevant.
She searched by date range.
1994 to 2000.
She searched by facility.
Isfahan Aerospace Research Center.
She found 47 documents.
She read the index entries for all 47.
Most were assessment summaries, signals reports, satellite imagery analysis.
One was an operational file summary.
The summary was three pages.
It described [music] a technical disinformation operation run by the technical intelligence division between 1994 and 1997.
The operation had used a recruited asset to deliver false propulsion schematics to the Isfahan facility.
>> [music] >> The target had been a short-range ballistic missile propulsion program.
The schematics had contained a hidden flaw designed to degrade the combustion cycle.
The operation was designated closed-circuit.
Yale read the summary twice.
Then she went back to her desk and looked at the Fateh-110 technical profile she’d been building.
The propulsion anomalies she’d identified, the suppressed ignition signature, the range extension, were consistent [music] with a dual-circuit combustion modification.
It was exactly the kind of modification an engineering team might develop if they were trying to reconcile two incompatible propulsion schematics, one designed as a conventional system, one designed as an alternative nozzle configuration, and happened to interpret the incompatibility as a design feature >> [music] >> rather than a defect.
She sat with that thought for a while.
Then she wrote an internal request.
The request asked for access to the full operational file for closed-circuit, [music] including the technical schematics that had been delivered to the Isfahan facility.
She marked it routine.
She sent it to the archive unit.
The response came 18 days later.
It said, “Access to requested [music] file requires divisional authorization.
Request forwarded to relevant authority.
” She wrote to the relevant authority.
That response came in 11 days.
It said, “The file referenced [music] in your request was reviewed at division level in 2001.
Conclusions from that review are reflected in the current technical assessment of the program.
No further review is indicated at this time.
She read that response three times.
Then she wrote again.
She explained in [music] two pages why the current technical assessment did not reflect what she had found in the signals data.
She attached her working note.
She attached the anomaly documentation.
>> [music] >> She sent it to the same address.
The response came in six days.
It said four words.
This matter is closed.
Yael printed that response >> [music] >> and put it in a folder on her desk.
She labeled the folder with a date and a reference number.
She kept working.
She was reassigned to a different file in December, a signals backlog from the Syrian border that required two analysts and had been sitting unaddressed for four months.
She worked on it through the winter.
She did not touch the Fateh 110 file during that [music] period.
In February of 2004, she finished the Syrian backlog and returned to her desk to find the Fateh 110 folder exactly [music] where she had left it.
She reopened it.
Over the following seven months, she built the comparative analysis on her own time, at her own desk, using only materials she already had access to.
She compared the Fateh 110 propulsion signature data against [music] the technical parameters described in the closed circuit operational summary.
She modeled what the Isfahan facility’s propulsion team would have produced if they had received two incompatible schematics and attempted to reconcile them.
She compared the output of that model against the actual Fateh 110 test data.
The match was not exact.
It was close enough to require an explanation.
She wrote the full analysis in October of 2004.
It was 41 pages.
The first page said, >> [music] >> “The propulsion modification currently observed in the Fateh-110 system is consistent with an unintended consequence of Operation Closed Circuit.
” The following pages explained why.
[music] She brought it directly to the head of the division on a Thursday morning.
She left a copy on his desk.
She went back to her own desk and waited.
By 4:00 in the afternoon, he had not responded.
She sent a message [music] asking if he had received the document.
He replied, “Yes.
Come to my office.
” She went.
He was sitting behind his desk.
The report was in front of him.
He had a pen in his hand but had not written anything on the cover page.
“You understand what you are saying in this document?” he said.
“Yes.
” Yael said.
“You are saying that a modification currently present in a weapons system deployed by a hostile organization was produced as a direct result of an operation run by this division.
” “I am saying the evidence is consistent with that conclusion.
” she said.
“I’m not saying it is certain.
” He looked at the report.
He looked at her.
“This does not leave this floor.
” he said.
She went back to her desk.
On it [music] was her working copy of the Fateh-110 technical profile, 30 pages of signals analysis, anomaly documentation, and comparative modeling.
Next to it, a photocopy of the three-page Closed Circuit Operational Summary she had pulled from the archive.
Two documents on one desk.
The original schematic that Karimi had drawn was somewhere in a sealed file on the third floor, inaccessible without authorization she no longer had any reason to believe she would receive.
The following morning, she received a formal notification.
Her report had been classified at the highest available level and transferred to divisional custody.
The notification included one line.
The matter described in the referenced document >> [music] >> is under review by appropriate authorities.
No further action is required from the originating analyst.
The folder on her desk, the one labeled with a date and a reference number, containing the four-word response, she put in her bag and took home.
October 21st, 2004.
The stamp on the classification notice was still wet when she walked out of the building.
Two years later, it was summer.
The first missiles hit Haifa on the 13th of July, 2006.
By the 16th, [music] the technical intelligence division had a confirmed proportion identification.
It came not from analysis, but from a field tracking report filed by a signals unit operating in the north.
Proportion signature identified as combustion configuration, range consistent with launch [music] point estimates in the eastern Bekaa Valley.
The analyst who wrote that report was named Dove.
He was 28 years old.
He had been in signals for 3 years.
He had never heard of Operation Closed Circuit.
He had read the standard Fateh-110 technical profile in the division’s assessment library, the one without the classified annex, the one that described the proportion configuration as a development of unknown origin.
He filed his tracking update at 11:47 in the morning and moved to the next [music] intercept.
Three days later, he filed a second report, noting that the dual circuit configuration appeared consistent across all incoming rounds in the northern theater.
He flagged this as operationally significant.
>> [music] >> The flag was acknowledged.
No follow-up action was assigned.
On the 17th of July, a rocket came down two streets from a pharmacy on Hanassi Boulevard in Haifa.
The pharmacist, a woman named Miriam, was in the back room when the siren sounded.
She went to the reinforced storeroom and waited.
The impact came 40 seconds later.
When she came [music] out, the front window of the shop was intact.
The dispensing counter was not.
The pressure wave had moved through the ventilation and taken the upper shelves with it.
Boxes of medication on the floor, a glass partition in pieces.
On the counter, undisturbed, was a prescription she had been filling when the siren went off.
She picked up the broom from the corner and started sweeping.
33 days Haifa, Nahariya, Afula, Tiberias.
165 people were killed on the Israeli side of the border.
On the Lebanese side, the toll was higher and harder to verify.
The Fateh-110 variant functioned as a suppression system.
Its purpose was reach and volume, not accuracy.
It reached places previous [music] generations of this weapon could not.
The dual circuit configuration shortened the early warning window.
By the time northern tracking systems registered the launch plume, the incoming rounds were already past the apex of their trajectory.
By the time the ceasefire came into effect on the 14th of August, the division had a full propulsion profile of every variant used in the conflict.
The dual circuit configuration appeared in 41 separate tracking reports.
A working group was formed to assess its implications.
Their final [music] report, 3 months later, described the configuration as an indigenous Iranian propulsion innovation of unknown development timeline.
The phrase “unknown development timeline” appeared four times in the document.
Nobody questioned it.
The working group was not given access to Yael’s report.
They did not know it existed.
The internal A of the technical intelligence function began in August while the war was still running.
Three people from the oversight unit were given a set of files and asked whether the division’s technical assessments had been adequate before the conflict.
The review team interviewed 14 analysts over 6 weeks.
They did not interview Yael.
She was not identified as having worked on the Fateh-110 file during the relevant period because her classified analysis had been removed from the accessible file record.
The 14 people they interviewed had never seen her work.
The review’s conclusion, filed in November, identified [music] specific gaps.
Failure to update range models, insufficient analysis of propulsion anomalies in the 24 months before the conflict.
It recommended restructuring the technical assessment unit.
It did not explain how the anomalies had come to exist.
The closed-circuit file, the full operational file, not the three-page [music] summary, remained in the classified archive on the third floor.
It had been accessed twice since its classification, once in 2001 for a divisional review, and once in October of 2004, the day Yael’s report was classified and transferred.
The file described an operation that had concluded [music] successfully.
It was never updated after 2001.
At the Isfahan facility, Farid Mohammadi had been promoted twice since the dual-circuit tests of 1997 and was now leading the integration unit.
He had no knowledge of Operation Closed Circuit.
[music] He had no reason to know that the drawings he had analyzed in the spring of that year had been provided by a foreign intelligence service.
The large sheet of paper that had [music] covered his wall, the one with Karimi’s original lines barely visible under four iterations of his own revisions, had been taken down years before.
He did not remember where it had gone.
Dariush Karimi’s operational file had not been updated since 2001.
The last entry was a routine check-in note.
>> Asset stable.
No current tasking requirements.
In 2006, a database cross-check flagged it for a status review.
A junior analyst confirmed no active components and recommended archiving.
The file was moved to the inactive archive in September.
Two months after the first Fateh-110 variant [music] struck Haifa.
Whether Greemy was still in Vienna, whether the drafting table with a broken hinge was still in the same apartment, none of this was recorded.
In February of 2007, Yael B submitted her resignation.
The form required a reason.
She wrote, “Personal.
” There was a box for additional comments.
She left it empty.
[music] On her last day, she arrived at the office at 7:00 in the morning before anyone else was in.
She cleared her desk methodically.
Notebooks in a box, two framed photographs, a ceramic mug she had brought from home 3 years earlier and never taken back.
At the back of the bottom drawer, she found two things.
The first was the photocopy of the three-page [music] closed-circuit operational summary, the one she had pulled from the archive in March of 2003.
The paper had gone soft at the edges.
The second was the printout of the four-word response.
“This matter is closed.
” She had thought she’d taken it home in 2004.
She had not.
It had been in the same drawer as the summary for 2 years, face down under a notebook.
She put both in the box with everything else.
She carried the box to the elevator.
The corridor was empty.
She pressed the button for the ground floor.
She put the box in the trunk of her car and drove north on Ayalon.
It was 9:00 in the morning.
Traffic was light.
She got off at the Petah Tikva Interchange and drove without purpose for a while.
She stopped at a petrol station.
She bought a coffee from the machine inside and sat in the car and drank it.
It was bad coffee.
She drove home.
She put the box on the kitchen table and did not unpack it that day.
The restructuring of the technical disinformation unit was approved in December and implemented through the first quarter of 2007.
The new protocols required for the first time that every technical disinformation package undergo adversarial engineering review, an assessment of whether a sophisticated team finding the planted flaw and attempting to solve it rather [music] than simply replicate it might arrive at a design superior to the original.
The requirement added four to six weeks to the development timeline of each operation.
>> Several offices argued this made certain operations impractical.
The argument was noted in the minutes and not acted [music] upon.
The requirement had a formal designation in the new protocol document.
Internally, the people who wrote it called it the closed circuit clause.
They did not explain the name in the document.
Nobody asked.
