Israel’s Wall Fell to Water Cannons
The wolf at the door, part two: notes from the Israeli frontier
Audio Version:
Israel’s Wall Fell to Water Cannons
The wolf at the door, part two: notes from the Israeli frontier
Jill and I spent the last week of July 2026 in Israel with a delegation organized to push back against Jew hatred, specifically the version now circulating on the American right. That was the purpose, and I expected to spend the week looking outward.
Instead, the country kept handing me a mirror.
The detail that did it was small and administrative. Young women assigned to watch the Gaza fence had spent months reporting rehearsals against mock Israeli compounds and drills on breaching the barrier. Their reports went up the chain and stopped, and some of them were told to quit filing. God lives in the details.
I know the shape of that. I lived something similar during COVID. So has anyone who has ever been the person whose report was never forwarded, at any scale, in any setting. The recognition comes with a jolt of vindication, and that is the thing to be careful about.
Seeing your own experience reflected in someone else’s catastrophe is not evidence of anything. It is the same mechanism that failed the Israelis, now running in me rather than in an intelligence directorate. A story that flatters the man telling it deserves to be held at arm’s length, and that includes this one.
The Israeli failure was not villainy. Nobody in Tel Aviv wanted that fence to fall. The officers who read the observers’ reports and moved on were competent people working inside a structure that made their judgment seem almost inevitable. Fifty years earlier, different people made the same call about the same kind of evidence in the same building.
If that charity is owed to them, it is owed elsewhere. I have spent five years criticizing American public health institutions, and I retract none of it. But I also have to take seriously that most of the people inside those institutions were not conspirators. They were doing what the observers’ superiors did: receiving information that contradicted a settled model and dismissing it as noise.
That is a less satisfying story than the villain version, and the villain version is the one that sells. I am not immune to its appeal. But I think the duller story is closer to the truth, and I think it is more useful, because a structure can be redesigned, while a villain can only be replaced.
The thought does not stay in Israel, and it does not stay in 2023.
Defeating a Fortress with Water Cannons
In 1973, many military planners regarded Israel’s Bar-Lev Line along the Suez Canal as one of the strongest defensive positions in the world. Built after the 1967 Six-Day War, it was intended to make any Egyptian assault across the canal prohibitively costly.
At two in the afternoon on October 6, 1973, Egyptian engineers pointed high-pressure hoses at the Israeli sand rampart along the Suez Canal and turned on the water.
The rampart formed the outer face of the Bar-Lev Line. Israel had spent several years and much of its defense budget building it. Thirty-odd fortified strongpoints, a wall of sand and rock rising sixty feet above the waterline, and buried pipes intended to set the canal itself on fire.
The water cut through it in a matter of hours. Egyptian assault troops crossed on rafts and walked through the gaps.
By the next morning, most of the strongpoints were surrounded. A single reserve infantry brigade of older men held them, roughly 450 soldiers spread across a hundred miles of canal. Only sixteen positions were manned.
The Bar-Lev Line had effectively ceased to exist as a defensive barrier. The problem was not that the fortifications were weak. It was that the assumptions behind them were wrong.
The Maginot Myth: The Defenses Held, The Strategy Failed.
The Maginot Line has become shorthand for a spectacular military failure, a monument to misplaced confidence in fixed defenses. That is not what happened.
Built along France’s border with Germany after the First World War, it was a vast system of underground fortresses, artillery positions, bunkers, tunnels, and obstacles designed to make a direct German invasion prohibitively expensive. It was never intended to defend every inch of France. It was meant to anchor the frontier so France’s field armies could fight elsewhere.
The Maginot Line was not breached in May 1940. German forces did not overwhelm it, did not outfight it, and, for the most part, did not attack it.
The fortifications performed the mission their designers assigned them, which was economy of force. Hold the German border cheaply so the mobile armies can concentrate somewhere else. The line did exactly that.
France then squandered what the line had bought. Maurice Gamelin, the French supreme commander, sent his best formations north into Belgium under the Dyle-Breda Plan, committing the strongest Allied divisions to a forward line inside Belgium and the Netherlands. His most capable mobile reserve went furthest of all.
That left the hinge of the whole position at Sedan, on the Meuse River. Two Series B reserve divisions held it, formations of older men with obsolete equipment and little training (Horne 1969).
The Germans came through the Ardennes instead and hit the hinge. The Maginot Line had done its job. The French command had failed to do theirs.
The Caveat That Got Lost
The hinge was weak because the Ardennes had been judged impassable to armor. But that judgment came with a condition, and the condition stopped being repeated.
Philippe Pétain, then France’s most decorated living soldier, told a Senate commission in 1934 that the Ardennes forest was impenetrable, provided special dispositions were made. The conclusion survived in French planning for six years. The condition did not.
By late 1944, the Allies had reached much the same conclusion about the same forest. In December 1944, they judged that Germany no longer had the capacity for a major offensive and treated the Ardennes as a quiet sector where tired divisions could rest. Three German armies came through it (MacDonald 1985). Same terrain, same assessment, different army.
A qualified judgment hardens into an unqualified premise. Nobody formally decides to discard the condition. It simply stops being repeated. After enough repetitions, the premise begins to look like fact rather than opinion.
Israeli intelligence did this twice. In 1973, the governing assumption held that Egypt would not attack without the ability to strike Israeli airfields in depth, and that Syria would not attack without Egypt. In 2023, it held that Hamas had been deterred and had become a government with something to lose. Israelis call the pattern HaConceptzia, the Conception.
Zvi Lanir distinguished between two kinds of surprise. Situational surprise concerns timing and place. Fundamental surprise occurs when the model itself collapses (Lanir 1983). Sedan, the Suez Canal, and the Gaza envelope were all fundamental surprises. In each case, the enemy did not merely arrive unexpectedly. It arrived through a possibility the prevailing model had ruled out.
Wingtip to Wingtip in Hawaii
On November 27, 1941, Washington sent a war warning to the Hawaiian commands. Lieutenant General Walter Short read the threat as sabotage by local agents rather than attack from the sea, and ordered the lowest of three readiness states, one designed to guard against exactly that.
Aircraft came out of dispersed revetments and were parked in tight rows on open ground, wingtip to wingtip, where a small guard force could watch them all. Ammunition went into locked storage. The precaution was rational for the threat Short believed he was facing. It also arranged the Hawaiian Air Force into an ideal strafing target, and most of it burned on the ground within the first minutes of the attack (Prange 1981).
The Navy’s assumptions failed in much the same way. Pearl Harbor is about forty feet deep, and American planners believed aerial torpedoes could not operate in water that shallow. In November 1940, British aircraft had crippled three Italian battleships at Taranto in comparable depth using torpedoes modified for the purpose. The Japanese studied Taranto and adapted their torpedoes with wooden fins that allowed them to run in shallow water. American planners read the same reporting. The information was accepted. The implication was not.
Data collection was not the failure. American cryptanalysts had broken the Japanese diplomatic cipher, and in September 1941 they decrypted an instruction from Tokyo directing its Honolulu consulate to divide Pearl Harbor into five sub-areas and report warship berthings by grid. Analysts interpreted it as Japanese thoroughness rather than targeting data.
At 7:02 on the morning of December 7, two privates at the Opana Point radar station detected a large formation about 130 miles north. The duty officer they telephoned told them not to worry because a flight of American bombers was expected from California. A flight of American bombers was, in fact, expected. The first Japanese bombs fell fifty-three minutes later.
Roberta Wohlstetter’s classic study of the attack settled the broader question. The problem was not a lack of information. The signals were present, buried in far more noise, and they appear obvious only after you know which ones mattered (Wohlstetter 1962).
What Sharon Lost
The Bar-Lev Line was built after an argument, and the people who lost the argument were right.
Ariel Sharon and Israel Tal favored a mobile defense. Hold the canal thinly, keep the armor back, let the Egyptians cross, and destroy them on open ground where Israeli tanks held the advantage. Chaim Bar-Lev and Avraham Adan favored the fortified line. Bar-Lev was Chief of the General Staff, and the line carries his name because he won the argument (Rabinovich 2004).
Nobody disputed the engineering. Both camps agreed the fortifications would be formidable. They disagreed about whether strength in a fixed position was the relevant variable. That was a judgment, not a technical question.
Institutions are good at answering technical questions. They are much less comfortable with questions of judgment. So they resolve what they can measure, and seniority of staff settles the rest.
Why Walls Cannot be Questioned
A large fortification is a political object before it is a military one. It is expensive, and it has a name attached. Budgets were defended for it, and the public was told it would keep them safe.
After that, doubting the wall means doubting the people who approved it. In any hierarchy, that is costly for a junior person to say and uncomfortable for a senior person to hear. Neither of those facts has anything to do with evidence.
Israel spent roughly a billion dollars on the Gaza barrier and completed it in 2021. Confidence in the sensors rose, manpower behind the fence thinned, and units shifted to the West Bank. Hamas studied the system, disabled the remote weapon stations first, and opened roughly thirty breaches in the first hour.
None of this requires concrete. Any expensive countermeasure that prominent officials have publicly defended acquires the same protection. Questioning it stops being an empirical act and becomes a political one. Once that happens, evidence no longer competes with reality. It competes with reputation.
Vincennes Had No Radio
In 1940, French commander Maurice Gamelin directed the defense of France from the Château de Vincennes outside Paris. His headquarters had no radio. Orders and reports moved by telephone and motorcycle courier. André Beaufre, who served on Gamelin’s staff, later described the headquarters as “a submarine without a periscope” (Horne 1969).
French aerial reconnaissance did spot the German columns moving through the Ardennes. The concentration was among the largest traffic jams in military history, stretching for more than a hundred miles toward the Rhine. But the reports reached a headquarters that could neither process nor respond at the speed events demanded. Worse, they contradicted an assumption that no one was prepared to question.
On October 7, 2023, something similar happened in southern Israel. The Gaza Division headquarters at Re’im was itself overrun. Divisional command stopped functioning for most of the day, and the opening fight fell to local ready squads, off-duty soldiers, police officers, and civilians.
Centralized command has a characteristic failure mode. It works when information flows reliably, and events unfold slowly. It struggles when communications break down, and decisions must be made immediately, which is precisely when it is needed most.
Groupthink: When the Explanation Becomes the Problem
Irving Janis offered the framework most people reach for.
Janis studied the Bay of Pigs, Pearl Harbor, and the escalation in Vietnam. He argued that cohesive groups under stress converge early on a course of action while suppressing the doubts of their own members (Janis 1972). The symptoms that matter here are collective rationalization, the illusion of unanimity, direct pressure on dissenters, and what Janis called mindguards, people who shield the group from disconfirming information.
His remedies were structural. The leader withholds his own preference until others have spoken. Someone is assigned to argue the opposing case. Independent subgroups work the same problem separately.
The Agranat Commission, Israel’s state inquiry into the 1973 surprise, reached nearly identical prescriptions in 1974 from an entirely different direction. It broke the military intelligence’s monopoly on national assessment, added an intelligence adviser to the Prime Minister, and created a standing office inside the intelligence directorate whose only job was to write the contrary assessment. Israelis call that office ipcha mistabra, an Aramaic phrase from the Talmud meaning the opposite seems likely.
Groupthink is also contested, and has been for thirty years. Ramon Aldag and Sally Riggs Fuller showed that the model was built from case studies selected because they ended badly, which guarantees the symptoms will be found (Aldag and Fuller 1993). Roderick Kramer reexamined the Bay of Pigs and Vietnam records and concluded that political self-interest explained the decisions better than the group cohesion that the theories behind groupthink did (Kramer 1998).
A seductive framework is exactly what trapped the Israelis. Used carelessly, groupthink becomes the very kind of unquestioned assumption it was meant to expose. Anyone who diagnoses it only in organizations he dislikes and never in those he belongs to is participating in the phenomenon rather than analyzing it.
What survives is narrower and better established. Groups under time pressure converge early. Dissent is expensive to voice. Senior people hear less than they think.
The Cost of Deference
One of history’s clearest examples of the cost of deference came before Germany’s invasion of the Soviet Union in June 1941.
Barton Whaley later counted eighty-four separate warnings that reached Soviet leadership before the attack (Whaley 1973). Stalin believed Hitler would not open a second front while Britain remained undefeated, so every report of an impending invasion was dismissed as British provocation. He returned one report from the NKVD, the Soviet secret police, with a note that its source inside the German Air Ministry was a disinformer rather than a source (Murphy 2005). Officers who pressed the point risked execution.
That is the authority gradient with the safety catch removed, and the gradient is a spectrum rather than a category.
Psychologist Solomon Asch explored the power of social conformity with a deceptively simple experiment. Volunteers were shown a set of lines and asked to identify which two were the same length. The correct answer was obvious. Unknown to the volunteer, however, everyone else in the room had been instructed to give the wrong answer.
The experiment’s most famous result is not its most important one. Roughly a third of the volunteers denied the evidence of their own eyes and agreed with the unanimous group. The more important finding came when Asch planted a single confederate who broke with the majority. Conformity fell by about three-quarters (Asch 1955). Unanimity generates the pressure. Breaking unanimity releases it
One industry took that lesson seriously. In December 1978, United Airlines Flight 173 circled Portland while the captain focused on a landing gear problem. The first officer and flight engineer both knew the aircraft was running out of fuel. They mentioned the fuel repeatedly, but neither directly challenged the captain or declared that they needed to land immediately. The aircraft crashed into a suburb with its tanks dry.
In response, the industry built Crew Resource Management, a training discipline aimed at the authority gradient between a captain and his crew (Helmreich, Merritt, and Wilhelm 1999). Junior officers learn escalating assertion. Captains learn to solicit challenge and to state their reasoning aloud so it can be attacked.
Aviation solved a problem that intelligence services and health agencies have not. Airline crashes produce bodies, a mandatory investigation, and a public report with named recommendations. Intelligence failures produce arguments.
Lieutenant Siman-Tov
By now, the pattern should look familiar. The warning exists. It reaches the institution. Then it stops moving.
Benjamin Siman-Tov was a lieutenant serving as an intelligence officer at Israel’s Southern Command. On October 1 and again on October 3, 1973, he wrote assessments arguing that Egyptian activity along the canal was not another exercise but a deployment for war. He was junior, he was right, and his superior did not forward the reports.
Israel had no shortage of information. Ashraf Marwan, the best-placed human source Israel ever ran inside Egypt, gave warning. Soviet advisers quietly evacuated their families. The Egyptian buildup was visible from the far bank.
Eli Zeira, who headed military intelligence, told the political leadership that certain special collection systems had been activated when they had not (Bar-Joseph 2005). Uri Bar-Joseph and Arie Kruglanski later analyzed the failure through the psychology of cognitive closure, arguing that those involved became committed to an answer they already had and stopped searching for evidence that might overturn it (Bar-Joseph and Kruglanski 2003).
Richard Betts drew the conclusion that should trouble anyone who designs institutions. Intelligence failures persist because the binding constraint is rarely collecting information. It is whether decision makers are willing to hear unwelcome conclusions, and no procedural reform can guarantee that (Betts 1978).
Fifty Years Later
The examples are separated by half a century. The pattern is almost unchanged.
In 1973, the Egyptians massed divisions along the Suez Canal in plain view, and Israeli intelligence classified the activity as an exercise. In 2023, Israeli intelligence possessed a Hamas planning document, reported afterward as Jericho Wall, describing a cross-border assault at roughly the scale of the one that came. Senior officers classified it as aspirational.
Exercise and aspirational are, in effect, the same word. Both mean the enemy is doing the thing, but the thing does not count because our model says he will not follow through.
Siman-Tov wrote his assessments, and his superior filed them away. Fifty years later, the tatzpitaniyot, young women conscripted to watch the Gaza fence, spent months reporting rehearsals against mock Israeli compounds, drone reconnaissance along the barrier, and other preparations. Their reports moved up the chain and stopped. Some of them were told to quit filing.
A junior person with a view of the ground writes it down. A senior person with a model decides it does not matter.
On the night of October 5, mass activation of Israeli SIM cards inside Gaza was detected. A consultation took place around three in the morning. The alert status was not raised.
The ipcha mistabra office was staffed and operating throughout. Israel had built the precise institutional answer to the precise institutional failure, maintained it for half a century, and it still did not fire.
The lesson is uncomfortable. Institutions can preserve reforms for decades without preserving the habits of mind those reforms were meant to encourage.
The cost was not theoretical. On October 7, more than 1,200 people were murdered, hundreds were taken hostage, and entire communities were overrun before the system understood that the model had failed.
Agranat was convened within weeks of the 1973 war and reported in April 1974. Nearly three years after October 7, Israel still has no equivalent state commission. The debate over how to constitute one has moved through the Supreme Court and the Knesset without resolution.
The Knowledge Problem
An economist described this problem in 1945 without mentioning armies, intelligence agencies, or war.
Friedrich Hayek argued that the knowledge most important to a decision rarely exists in one place. Much of it consists of what he called “the particular circumstances of time and place,” held by people on the spot. Much of it cannot be fully articulated, even by the people who possess it. A central authority cannot simply collect that knowledge because, in compressing it for transmission, it loses the very details that matter (Hayek 1945).
The young women at the observation posts did not hold a threat assessment. They knew what the far side of the fence looked like on an ordinary Tuesday, well enough to notice when it stopped looking ordinary. That kind of knowledge does not survive translation into a summary paragraph, and a summary paragraph is the only form in which it can travel upward.
The pattern repeated itself again and again. Lieutenant Benjamin Siman-Tov, a junior Israeli intelligence officer in 1973, was close enough to events to recognize that Egyptian forces were preparing for war, but his warnings never reached the people making the final decisions.
Fifty years later, the local defense squads in communities such as Be'eri and Nir Oz were the only people with an accurate picture of what was happening on their own streets during the opening hours of the October 7 attack. Every one of these systems was designed to move local knowledge upward to a central authority that would decide. Every one of these systems was designed on the assumption that local knowledge could be transmitted upward without losing its meaning.
The loss is not random. Whoever writes the summary already has a model of the situation. Summarizing requires judging what is relevant. Relevance is judged against the model. The details most likely to challenge the model are therefore the first to be discarded as noise. At every level, the filter preferentially strips away disconfirming evidence. But God lives in the details.
Markets avoid this problem because they do not require all knowledge to travel to the center. Prices carry the signal, and local actors make local decisions. Armies and public health agencies cannot go that far because someone still has to decide where reserves, personnel, and resources should go.
What they can do is move authority closer to the knowledge instead of pulling the knowledge toward the authority.
The Germans had a word for that in 1940: Auftragstaktik, usually translated as mission command. A subordinate commander was given the objective and left to determine the method, on the assumption that the commander looking at the ground sees what the map cannot. Heinz Guderian, commanding the German armored spearhead, crossed the Meuse at Sedan on his own initiative and kept going, in places beyond his orders (van Creveld 1985).
Israel built much of its military doctrine on the same principle and has been unusually successful with it. Eitan Shamir’s comparative study identifies the IDF as the strongest modern practitioner of mission command among Western militaries (Shamir 2011). The fighting in the first hours of October 7 was carried by people exercising exactly that discretion.
The people closest to the fight performed as intended. The failure occurred higher in the chain of command.
The Phoenix Memo
The pattern repeated itself in the United States before 9/11.
Kenneth Williams was an FBI agent in Phoenix. On July 10, 2001, he warned headquarters that Osama bin Laden might be sending followers to American flight schools and recommended a nationwide canvass of aviation schools. His memo never reached senior FBI leadership before September 11 (National Commission 2004).
In August 2001, agents in Minneapolis arrested Zacarias Moussaoui after a flight school reported that a student with almost no flying experience wanted time on a Boeing 747 simulator. The field office asked headquarters for authority to search his laptop. Headquarters refused. Coleen Rowley, the division counsel in Minneapolis, later documented the sequence in a memorandum to Director Robert Mueller (Rowley 2002).
Two field offices. Two people close to the problem. Both warnings stopped at the same level. God lives in the details.
There was also "the wall," the informal name for the legal and procedural barriers separating intelligence and criminal terrorism investigations. Procedures dating from 1995 sharply restricted the flow of information between intelligence and criminal terrorism investigations. The barrier is legal rather than physical, and it exists for a defensible reason: protecting criminal prosecutions. Its unintended effect is to prevent exactly the fusion of scattered warnings that the intelligence system was supposed to provide.
The 9/11 Commission concluded that the central failure was one of imagination (National Commission 2004).
That conclusion is only partly convincing. The scenario had been imagined repeatedly.
The Bojinka plot uncovered in Manila in 1995 envisioned airliners used as weapons.
Algerian hijackers seized an Air France flight in 1994 with the reported intention of crashing it into Paris.
A report prepared for the National Intelligence Council in 1999 described al-Qaeda operatives flying an aircraft into a building in Washington (Hudson 1999).
The problem was not imagination. It was that institutions failed to act on what they had already imagined. Israeli intelligence filed the Hamas invasion plan later known as Jericho Wall as aspirational for much the same reason.
Political scientist Amy Zegart reached the same conclusion in organizational terms. The agencies had been built for a Cold War adversary, their structures resisted adaptation, and competent people operated inside systems that all but guaranteed the outcome (Zegart 2007).
Congress responded to this issue during the 9/11 attacks in 2004 by creating the Director of National Intelligence and the National Counterterrorism Center so that scattered warnings could be fused and delivered to the President. Institutional reforms followed. As elsewhere in this essay, the harder question is whether structures alone can overcome the habits of mind that produced the failure in the first place.
Five Microns
The same failure mode appeared during COVID.
For most of 2020, the World Health Organization and American public health agencies treated SARS-CoV-2 as spreading primarily by large droplets that fell quickly to the ground. That model produced the countermeasures Americans lived under, including surface disinfection, plexiglass barriers, and distancing rules. It de-emphasized ventilation and shared indoor air.
The model rested on a boundary between droplets and aerosols set at five microns.
Katherine Randall and her co-authors traced the origin of that number. It did not come from the physics of airborne particles. Instead, it appears to have migrated from a different literature concerned with how deeply inhaled particles deposit in the lungs, an entirely different question. The two became conflated in the mid-twentieth century, and the figure was repeated as settled science for decades (Randall et al. 2021). Megan Molteni’s account for WIRED is the most accessible summary (Molteni 2021).
In July 2020, Lidia Morawska and Donald Milton published an appeal in Clinical Infectious Diseases, co-signed by 239 scientists from 32 countries, asking health authorities to recognize airborne transmission (Morawska and Milton 2020). The signatories were aerosol physicists and engineers. They were the observation post. They held dispersed technical knowledge that central public health authorities could not absorb without abandoning guidance they had already publicly defended.
The institutional response was gradual and slow. Morawska and her co-authors published a retrospective in the same journal three years later titled Science Rejected, Lives Lost (Morawska et al. 2023).
The suppression of dissent in this case left a paper trail. On October 8, 2020, four days after three epidemiologists published the Great Barrington Declaration, Francis Collins, then director of the National Institutes of Health, emailed Anthony Fauci and Clifford Lane. He called the authors “three fringe epidemiologists,” noted that the declaration was receiving attention, and wrote that there needed to be a quick and devastating published takedown of its premises. He asked whether one was underway. The email surfaced through Freedom of Information Act requests and was entered into the congressional record (Bhattacharya 2023).
The email documents a process rather than resolving a scientific debate. Two senior officials responded to scientific disagreement by organizing its public rebuttal rather than commissioning its evaluation. Janis, with his Groupthink theory, called that behavior a mindguard: protecting a prevailing view from disconfirming evidence rather than exposing it to challenge.
Institutionalized dissent is not valuable because dissenters are usually right. They may be right or wrong, but if correct, they represent a threat to established authority and consensus. That is precisely why it is easy to dismiss them and difficult to hear the dissenting opinion that matters.
Institutionalized dissent is not valuable because dissenters are usually right. It is valuable because, at the moment a disagreement matters most, no institution can confidently tell the difference between a flawed objection and the one warning it, and which decision makers ignore at the peril of all concerned.
What Actually Works
The encouraging news is that we already know a great deal about how organizations become more reliable. None of it requires perfect people.
Karl Weick and Kathleen Sutcliffe studied organizations that operate dangerous systems with remarkably few catastrophes, including aircraft carriers and nuclear power plants. They found a common set of habits: constant attention to small failures, reluctance to accept simple explanations, and deference to expertise rather than rank during a crisis (Weick and Sutcliffe 2007). That last habit directly counters the authority gradient.
Sociologist Diane Vaughan, studying the Challenger disaster, coined the phrase normalization of deviance to describe how repeated anomalies gradually become accepted as normal simply because nothing bad has happened yet (Vaughan 1996). Physicist Richard Feynman reached much the same conclusion in his appendix to the Rogers Commission report: “Reality must take precedence over public relations, for nature cannot be fooled” (Feynman 1986). God lives in the details.
Psychologist Gary Klein proposed perhaps the simplest intervention of all: the premortem. Before adopting a plan, the group assumes it is a year later, and the plan has failed catastrophically. Everyone then writes the history of that failure. Treating failure as an established fact rather than a mere possibility removes much of the social cost of raising uncomfortable objections (Klein 2007).
Political scientist Philip Tetlock addressed a different weakness. Instead of rewarding confidence or seniority, he argued that institutions should track predictions, measure their accuracy, and learn who is actually well calibrated over time (Tetlock 2005).
None of these ideas is exotic. None is especially expensive. All cost far less than building a wall, and all are more likely to prevent the next surprise.
The Office Existed
The Agranat Commission created the Devil's Advocate Office in 1974. Israeli military intelligence still had it on October 6, 2023. A named unit with a written mandate and fifty years of institutional memory. It did not prevent October 7 from happening.
The United States built the Central Intelligence Agency in 1947 so that scattered warnings would be fused and carried to the President. Fifty-seven years later, after concluding that scattered warnings still were not being fused and carried to the President, Congress created a second office and assigned it much the same mission.
Political scientist Richard Betts, whose landmark 1978 study examined why intelligence failures persist, predicted this outcome. Procedural reform does not solve the deeper problem. It cannot compel people to hear what they do not wish to hear.
That is the narrower, more durable lesson. Every institution that suffers a major surprise writes down what it learned. Writing it down is the easy part. The difficult part comes years later, when a junior officer, an analyst, an engineer, or a scientist brings forward an unwelcome report that challenges an accepted model. In that moment, no reform, commission, or organizational chart makes the decision. A person does.
The Bar-Lev Line was made of sand. It fell to water in an afternoon.
Fifty years later, the Gaza barrier was among the most sophisticated border defenses ever built. It, too, failed when the assumptions behind it proved false, and on a failure to attend to the details of local intelligence. And God lives in the details.
Every institution builds its own walls.
Most of them are made of assumptions.
RWM/JGM
We spent a lot of time researching, checking primary sources, and trying to understand not just what happened but why institutions keep repeating the same mistakes, I hope this essay has given you something worth thinking about.
If essays like this are valuable to you, please consider becoming a paid subscriber.
The headlines are free. Digging through commission reports, memoirs, academic papers, declassified documents, and historical records to connect ideas across decades is not. That work takes time, and paid subscriptions are what make it possible.
They also allow us to stay independent. No advertisers, no sponsors, no foundations, and no institution deciding which questions are acceptable to ask.
If you want more long-form investigations that go beyond the news cycle and try to understand how systems actually work, your support genuinely makes them possible.
Thank you for reading, for sharing our work, and for helping us keep doing it.
References
Aldag, Ramon J., and Sally Riggs Fuller. 1993. “Beyond Fiasco: A Reappraisal of the Groupthink Phenomenon and a New Model of Group Decision Processes.” Psychological Bulletin 113 (3): 533 to 552.
Asch, Solomon E. 1955. “Opinions and Social Pressure.” Scientific American 193 (5): 31 to 35.
Bar-Joseph, Uri. 2005. The Watchman Fell Asleep: The Surprise of Yom Kippur and Its Sources. Albany: State University of New York Press.
Bar-Joseph, Uri, and Arie W. Kruglanski. 2003. “Intelligence Failure and Need for Cognitive Closure: On the Psychology of the Yom Kippur Surprise.” Political Psychology 24 (1): 75 to 99.
Betts, Richard K. 1978. “Analysis, War, and Decision: Why Intelligence Failures Are Inevitable.” World Politics 31 (1): 61 to 89.
Bhattacharya, Jay. 2023. Prepared Statement Before the Subcommittee on Health, Committee on Energy and Commerce, U.S. House of Representatives. March 28.
Feynman, Richard P. 1986. “Appendix F: Personal Observations on the Reliability of the Shuttle.” In Report of the Presidential Commission on the Space Shuttle Challenger Accident. Washington, DC.
Hayek, F. A. 1945. “The Use of Knowledge in Society.” American Economic Review 35 (4): 519 to 530.
Helmreich, Robert L., Ashleigh C. Merritt, and John A. Wilhelm. 1999. “The Evolution of Crew Resource Management Training in Commercial Aviation.” International Journal of Aviation Psychology 9 (1): 19 to 32.
Horne, Alistair. 1969. To Lose a Battle: France 1940. London: Macmillan.
Hudson, Rex A. 1999. The Sociology and Psychology of Terrorism: Who Becomes a Terrorist and Why? Washington, DC: Federal Research Division, Library of Congress.
Janis, Irving L. 1972. Victims of Groupthink. Boston: Houghton Mifflin.
Klein, Gary. 2007. “Performing a Project Premortem.” Harvard Business Review 85 (9): 18 to 19.
Kramer, Roderick M. 1998. “Revisiting the Bay of Pigs and Vietnam Decisions 25 Years Later.” Organizational Behavior and Human Decision Processes 73 (2-3): 236 to 271.
Lanir, Zvi. 1983. Fundamental Surprise: The National Intelligence Crisis. Tel Aviv: Hakibbutz Hameuchad.
MacDonald, Charles B. 1985. A Time for Trumpets: The Untold Story of the Battle of the Bulge. New York: William Morrow.
Molteni, Megan. 2021. “The 60-Year-Old Scientific Screwup That Helped Covid Kill.” WIRED, May 13.
Morawska, Lidia, and Donald K. Milton. 2020. “It Is Time to Address Airborne Transmission of Coronavirus Disease 2019 (COVID-19).” Clinical Infectious Diseases 71 (9): 2311 to 2313.
Morawska, Lidia, et al. 2023. “Coronavirus Disease 2019 and Airborne Transmission: Science Rejected, Lives Lost. Can Society Do Better?” Clinical Infectious Diseases 76 (10): 1854 to 1859.
Murphy, David E. 2005. What Stalin Knew: The Enigma of Barbarossa. New Haven: Yale University Press.
National Commission on Terrorist Attacks Upon the United States. 2004. The 9/11 Commission Report. Washington, DC: Government Printing Office.
Prange, Gordon W. 1981. At Dawn We Slept: The Untold Story of Pearl Harbor. New York: McGraw-Hill.
Rabinovich, Abraham. 2004. The Yom Kippur War: The Epic Encounter That Transformed the Middle East. New York: Schocken.
Randall, Katherine, E. Thomas Ewing, Linsey C. Marr, Jose L. Jimenez, and Lydia Bourouiba. 2021. “How Did We Get Here: What Are Droplets and Aerosols and How Far Do They Go? A Historical Perspective on the Transmission of Respiratory Infectious Diseases.” Interface Focus 11 (6): 20210049.
Rowley, Coleen. 2002. Memorandum to FBI Director Robert Mueller. May 21.
Shamir, Eitan. 2011. Transforming Command: The Pursuit of Mission Command in the U.S., British, and Israeli Armies. Stanford: Stanford University Press.
State of Israel. 1974. Agranat Commission of Inquiry, Interim Report. Jerusalem.
Tetlock, Philip E. 2005. Expert Political Judgment: How Good Is It? How Can We Know? Princeton: Princeton University Press.
van Creveld, Martin. 1985. Command in War. Cambridge, MA: Harvard University Press.
Vaughan, Diane. 1996. The Challenger Launch Decision: Risky Technology, Culture, and Deviance at NASA. Chicago: University of Chicago Press.
Weick, Karl E., and Kathleen M. Sutcliffe. 2007. Managing the Unexpected: Resilient Performance in an Age of Uncertainty. 2nd ed. San Francisco: Jossey-Bass.
Whaley, Barton. 1973. Codeword Barbarossa. Cambridge, MA: MIT Press.
Wohlstetter, Roberta. 1962. Pearl Harbor: Warning and Decision. Stanford: Stanford University Press.
Zegart, Amy B. 2007. Spying Blind: The CIA, the FBI, and the Origins of 9/11. Princeton: Princeton University Press.