Ask an intelligence historian why one secret action succeeds and another collapses, and the honest answer is that the difference is rarely luck. It is structure. A campaign of covert violence is not a single decision but a chain of them, and the chain holds or breaks at predictable joints. Bangladesh in 1971 held. The Bay of Pigs in 1961 broke within seventy-two hours. Both were planned by competent professionals who believed they understood the terrain. The question this analysis sets out to answer is therefore not whether covert action can work, since the historical record proves that it sometimes does, but rather what separates the operations that age into quiet triumphs from the ones that age into national embarrassments, criminal trials, and strategic disasters.

That question is not academic for anyone reading the news out of South Asia. Since 2021, a sequence of wanted militants has been shot dead on Pakistani soil by assailants who arrive on motorcycles, fire at close range, and vanish into traffic. Indian officials have claimed nothing. Pakistani officials have alleged much. The pattern, examined across dozens of incidents, looks less like coincidence and more like a sustained program, and that program now has enough operational history behind it to be measured rather than merely described. The measurement requires a yardstick, and the only honest yardstick is the comparative record of every other state that has tried something similar. India’s shadow war is new. The dilemmas it faces are not.

This piece builds a diagnostic out of that record. It takes ten covert operations from the past sixty years, five widely judged successes and five widely judged failures, and runs each of them through eight tests that intelligence scholars consistently identify as the load-bearing variables: objective clarity, intelligence accuracy, operational discipline, political patience, exit criteria, deniability maintenance, collateral-damage control, and strategic-goal alignment. Then it places India’s campaign of targeted killings inside that same frame and asks how it scores. The conclusion, stated plainly so the reader can argue with it, is that the shadow war satisfies most of the tests a covert program needs to pass, and fails one of them in a way that should worry its architects more than any Pakistani protest note.

The Cases

A diagnostic is only as good as the cases that populate it, so the ten chosen here were selected to span methods, decades, and outcomes rather than to flatter any single thesis. Three involve assassination or elimination, three involve arming a proxy, two involve regime change, one involves cyber sabotage, and one involves kidnapping. They include operations by the intelligence services of the United States, Israel, and India, and they range from the early Cold War to the post-2001 era. What unites them is that each has been studied long enough for the verdict to settle, which is precisely what India’s shadow war still lacks.

Begin with the successes, because the word deserves caution. None of these operations was clean in a moral sense, and naming an operation successful here means only that it achieved its stated objective at a cost its sponsor was willing to pay and without producing consequences that reversed the gain. The first is the Indian intelligence effort behind the liberation of Bangladesh in 1971. Under R. N. Kao, the Research and Analysis Wing trained and armed the Mukti Bahini, the Bengali resistance fighting Pakistani military rule in what was then East Pakistan, and fed operational intelligence into the conventional Indian campaign that followed. The objective was bounded: support an uprising that was already underway, hasten the birth of a state that Bengali resistance and Pakistani atrocity were already producing, and conclude when the war concluded. It did conclude, in December 1971, and the agency walked away from a finished job. Srinath Raghavan’s history of the conflict places the intelligence dimension correctly, as one important input among several rather than the decisive cause, and that modesty is itself part of why the effort reads as a success. RAW did not try to own the outcome. It contributed to an outcome that local actors were already driving. Readers tracing how this episode shaped everything that came after can follow the success case in detail.

The second success is the Israeli capture of Adolf Eichmann in Buenos Aires in 1960. Mossad spent years confirming that the architect of the Nazi deportation machinery was living under an assumed name in Argentina, then seized him on a suburban street, held him in a safe house, and flew him to Israel disguised as a sedated airline crew member. The objective was a single human being and a single courtroom. When Eichmann landed in Israel, the operation was over. There was no second phase, no mission creep, no open-ended commitment. The sovereignty of Argentina was violated, and Argentina protested, but the bounded nature of the act limited the diplomatic damage and the affair faded within months.

The third success is Operation Neptune Spear, the 2011 raid that killed Osama bin Laden in Abbottabad, Pakistan. A decade of patient intelligence work, much of it tracing a single trusted courier, located one man in one compound, and a Navy SEAL element executed a raid with a defined endpoint. The objective never expanded. The United States did not try to use the Abbottabad intrusion as a lever for broader influence over Pakistan; it took its target and left. The raid strained Washington’s relationship with Islamabad, but the strain was survivable precisely because the action was narrow.

Standing slightly apart is the fourth case, because it involved no bullets at all. Stuxnet, the cyber weapon widely attributed to a joint American and Israeli program, infiltrated the industrial controllers governing Iran’s uranium enrichment centrifuges at Natanz and quietly destroyed a meaningful fraction of them around 2009 and 2010. The objective was to delay Iran’s nuclear progress and to substitute for the kinetic strike that hardliners in both countries were urging. Within those limits it worked. The malware was eventually discovered and analyzed, which ended its useful life, but discovery did not reverse the centrifuge losses, and the operation bought the time it was designed to buy.

The fifth success is the most contested, and it is placed here deliberately as the bridge between the two halves of the diagnostic. Operation Wrath of God was Israel’s response to the murder of its athletes at the 1972 Munich Olympics, a sustained campaign to hunt and kill members of Black September and the Palestinian networks associated with the attack across Europe and the Middle East. It eliminated many of its intended targets and announced, to anyone watching, that the state would pursue its enemies anywhere and indefinitely. By the narrow measure of targets reached, it succeeded. By two other measures it did not, and those failures, the killing of an innocent man in Norway and the absence of any defined endpoint, are examined later because they are exactly the failures India’s planners must now study. The full operational chronicle of that campaign, including the operation that went wrong, is treated as a partial success for reasons that this diagnostic will make precise.

Turn now to the failures, and notice that none of them failed for want of resources or talent. The first is the Bay of Pigs invasion of April 1961, in which the Central Intelligence Agency landed a brigade of roughly fourteen hundred Cuban exiles on the southern coast of Cuba in the expectation that the landing would trigger a popular uprising against Fidel Castro. No uprising came. Castro’s forces pinned the brigade on the beach within hours and captured most of it within three days. The objective was regime change, the operational plan rested on an assumption about Cuban popular sentiment that the agency’s own analysts could have punctured, and there was no recovery option once the assumption proved false. The Kennedy administration’s subsequent internal inquiry, the review led by General Maxwell Taylor, catalogued a failure of planning and assumption rather than a failure of courage.

The second failure is Iran-Contra, the mid-1980s scheme in which officials of the Reagan administration sold weapons to Iran and diverted the proceeds to fund the Contra rebels in Nicaragua, evading the congressional Boland Amendment that had cut off Contra aid. The objective was never singular. It tangled hostage recovery in Lebanon, ideological support for an anti-communist insurgency, and a desire to operate beyond legislative reach into one undertaking that satisfied none of its purposes cleanly. When the scheme was exposed in late 1986, the resulting Tower Commission and congressional investigations consumed the administration’s credibility for the remainder of its term.

The third failure is India’s own, and it is the most painful entry in the series. Beginning in the early 1980s, RAW trained and armed Tamil militant groups, including the organization that became the Liberation Tigers of Tamil Eelam, as a means of pressuring the Sri Lankan government. The proxy then slipped its leash. When India deployed the Indian Peace Keeping Force to Sri Lanka in 1987, that force found itself fighting the very militants Indian intelligence had helped create, and it lost more than twelve hundred soldiers before it withdrew. In 1991 the Tigers assassinated former Prime Minister Rajiv Gandhi on Indian soil. The episode is the textbook case of proxy blowback, and its lessons were absorbed so deeply into Indian institutional memory that they help explain the methodology of the present campaign. Readers can examine that reversal as a failure case whose shadow still falls across every proxy decision India makes.

Operation Cyclone, the fourth failure, was the CIA program that armed the Afghan mujahideen against the Soviet occupation through the 1980s. By one measure the program succeeded, since the Soviet Union withdrew from Afghanistan in 1989. But the measure that matters for strategy is whether the operation advanced the sponsor’s long-term position, and on that measure Cyclone is a cautionary monument. The weapons, the networks, and the ideological infrastructure built to bleed the Soviets did not dissolve when the Soviets left. They fed an Afghan civil war and seeded the militant ecosystem from which al-Qaeda later emerged. The operation won its battle and lost the larger argument because it never planned for the day after victory.

The fifth failure is the United States extraordinary rendition program, the post-2001 practice of seizing terrorism suspects without judicial process and transferring them to secret facilities or to third countries known to use torture. The program’s objective drifted from intelligence collection toward something closer to indefinite extrajudicial detention. Its deniability collapsed under journalistic and judicial scrutiny. It swept up innocent men, the German citizen Khaled el-Masri being the most documented case of mistaken identity. It produced criminal convictions of intelligence officers in absentia in an Italian court, and a 2014 study by the United States Senate Select Committee on Intelligence concluded that the harsh techniques the program enabled had not produced the unique intelligence its defenders claimed. The episode is examined in full as a failure case whose legal wreckage is still being cleared.

Two of the failures deserve a closer look, because their operational texture carries lessons that the summary verdicts compress. The Bay of Pigs was not undone by a lack of force but by a cascade of compromises made to preserve a deniability that the operation could never realistically have possessed. The landing site was shifted late in planning, partly to make the assault look less like an American invasion, to a location that was militarily worse, ringed by swamp and cut off from the mountains into which a defeated brigade had been expected to retreat and survive as a guerrilla force. Air cover was curtailed for the same cosmetic reason, to keep American fingerprints faint. The result was an action crippled by the attempt to disguise it, neither honestly covert nor honestly overt, and it fell into the gap between the two. The lesson, which the deniability dimension will return to, is that an operation whose scale cannot genuinely be hidden should not pretend that it can, because the pretense degrades the action without delivering the concealment.

Iran-Contra carries a different operational lesson, the danger of a scheme run outside the institution’s own machinery. It was driven by a small circle operating partly around the Central Intelligence Agency’s normal procedures and the normal channels of legislative accountability, and that very freedom from process, which its architects no doubt experienced as welcome efficiency, became the source of its collapse. Process inside an intelligence service is not merely bureaucratic friction to be cleared away. It is the accumulated mechanism by which assumptions get challenged, records get kept, and some identifiable person remains answerable for the result. An action that escapes the process escapes the friction and escapes the safeguards in the same motion, and it tends to fail in ways that a slower, more procedural undertaking would have caught early. These two failures, the over-disguised invasion and the under-supervised scheme, frame the diagnostic from opposite directions, and India’s shadow war is best read against both of them at once.

Eleven cases, then, once India’s shadow war is added as the subject the other ten exist to illuminate. The diagnostic that follows does not score the operations on a single axis of good and bad. It tests each one against eight specific questions, grouped into four analytical dimensions, and it is in the pattern of answers, not in any individual verdict, that the lessons live.

One feature of the selection deserves to be made explicit before the diagnostic begins, because it shapes everything that follows. The eleven cases were not chosen to be morally comparable, and a reader should resist the pull to rank them on a single scale of approval. A reader may find the Eichmann capture defensible and the rendition program indefensible while still recognizing that the diagnostic is not asking that question. It is asking a narrower and more technical one: given what the sponsoring state set out to do, did the operation deliver it at a cost the state would accept, and did the operation avoid generating consequences that reversed the gain. That narrower question is the only one on which sixty years of cases can be meaningfully compared, because the moral questions differ so sharply from case to case that no common scale survives contact with them. The value of holding the technical question separate is that it produces a tool, and a tool can be pointed at India’s shadow war and made to yield something more useful than approval or disapproval.

The First Dimension: Knowing What You Want and Knowing Where It Is

The first two tests of any covert action concern its foundations. The first asks whether the objective was clear, bounded, and achievable. The second asks whether the intelligence underpinning the action was accurate enough to act on. An operation that fails either test is built on sand, and no amount of operational skill applied later can compensate for a foundation that was wrong from the start.

Objective clarity is the variable that most cleanly sorts the eleven cases. Look at the successes and a pattern jumps out immediately. The Eichmann capture had an objective that could be written on an index card: locate this man, seize this man, deliver this man to a courtroom. Neptune Spear was scarcely more complex in its definition, however complex its execution. Stuxnet aimed to delay a specific industrial process by a measurable interval. Even the Bangladesh effort, the broadest of the successes, had a bounded definition once it is read correctly, since RAW was not trying to govern Bengal or to install a client regime but to assist an uprising toward a finish line that the conventional war would draw. Loch Johnson, whose study of American intelligence is the most systematic survey of the success and failure question, makes the point in institutional terms: covert action behaves like a controllable instrument only when the policymaker can state, in advance and in one sentence, what completion looks like. When completion cannot be stated, the instrument controls the policymaker.

Now read the failures through the same lens. The Bay of Pigs aimed at regime change, an objective that sounds clear but is not, because regime change is not an event that an operation can deliver; it is an outcome that depends on the political behavior of an entire society, behavior that no foreign service can command. Iran-Contra had no single objective at all, only a bundle of incompatible ones, and a bundle cannot be completed because its components compete. Operation Cyclone had an objective that was clear in the near term, expel the Soviets, and entirely undefined in the long term, with the result that the operation simply kept running after its near-term goal was met because no one had specified what running it had ultimately been for. The RAW-LTTE relationship suffered the subtlest version of the disease, objective drift, in which an operation that began as leverage against Colombo mutated, without any single decision, into the sponsorship of an armed movement that India could neither control nor abandon. Richard Betts, whose work on why intelligence organizations fail is built around the idea that the most dangerous errors are structural rather than personal, would recognize each of these as a failure that was latent in the design and merely waiting for circumstances to expose it.

A useful refinement separates a clear objective from an achievable one, because the two are not the same thing and the failures repeatedly confused them. The Bay of Pigs had, in one narrow sense, a clear objective, since everyone involved could name it in a phrase, the removal of Castro, and yet it was not an achievable objective for a covert action, because no covert action can manufacture the popular consent that a genuine change of government requires. Clarity without achievability is a trap, a goal stated crisply enough to win a policymaker’s approval and impossible enough to guarantee the operation’s defeat. The genuinely sound objectives in this set were both clear and achievable in the same breath, and the test that exposes the difference is disarmingly simple: can the operation, by its own efforts, actually produce the result, or does the result depend on the choices of people whom the operation does not control? Eichmann’s capture depended only on Mossad. Castro’s fall depended on the Cuban people. That single distinction, more than any other in the diagnostic, is why one operation is studied as a model and the other as a warning.

Where does India’s shadow war sit on objective clarity? Strikingly high. The campaign’s objective, inferred from its target selection rather than from any statement, is narrow and stateable: eliminate specific individuals on a list of wanted militants, men associated with attacks on Indian soil and with the organizations that sponsor them. It is not regime change. It does not seek to topple the Pakistani state, to install a client government, or to redraw a border. It does not arm a proxy whose loyalty might curdle. It seeks the removal of named people, one at a time, and the removal of each name is a completed task in itself. By the standard that sorts the Eichmann capture from the Bay of Pigs, the shadow war belongs firmly with the successes on this first test. Its planners learned the lesson of the Tamil disaster and learned it well: do not build a movement you will have to disown, simply reach the individual and withdraw.

The second test, intelligence accuracy, sorts the cases almost as cleanly, and it exposes a truth that intelligence professionals state more bluntly than outsiders expect. Operations do not usually fail at the trigger. They fail in the months before the trigger, in the quality of the picture the operators are given. The Eichmann capture rested on years of painstaking confirmation, including the famous corroboration that the suspect’s son had been seen with a young woman whose family connection pointed back to the fugitive. Neptune Spear rested on a decade of work that narrowed a courier’s identity, then his movements, then a compound. Stuxnet rested on an intimate, almost loving knowledge of the exact industrial controllers and centrifuge cascades inside Natanz, knowledge so specific that the weapon could damage the machines while telling their operators that nothing was wrong.

Contrast the Bay of Pigs, where the operation’s central intelligence assumption, that the Cuban population would rise on cue, was not a piece of confirmed intelligence at all but a hope dressed in the language of analysis. The agency’s planning directorate wanted the assumption to be true and therefore did not subject it to the adversarial scrutiny that its own analytical directorate could have provided. The rendition program offers the inverse pathology, an intelligence process so distorted by the pressure to produce that it generated false leads, and el-Masri’s months of wrongful detention were the direct product of an identity confusion that disciplined verification would have caught. Betts has a precise phrase for this family of errors, the failure of intelligence as a process rather than as a fact, and the rendition cases are his thesis in operation.

There is a second pattern inside the intelligence-accuracy test, visible only when the cases are read against one another, and it concerns who is permitted to challenge the picture. The Bay of Pigs failed in part because the branch that planned the operation was also, in effect, the branch that validated the assumptions behind it, and an organization grading its own homework rarely awards itself a failing mark. The strongest successes, by contrast, kept the people who wanted the operation separate from the people whose job was to test whether it would actually work. Betts treats this separation as the structural heart of intelligence failure, the presence or absence of an adversarial check, and the point applies with particular force to a deniable campaign, because deniability tends to shrink the circle of people cleared to know, and a shrunken circle is a circle with fewer skeptics inside it. The standing risk for any long covert program is that secrecy and self-validation slowly merge until no one inside the compartment is positioned, or incentivized, to say out loud that the picture is wrong.

India’s shadow war scores well on intelligence accuracy, though here the assessment must be more careful, because the evidence is necessarily indirect. The pattern visible across the targeted killings, the specificity of the individuals reached, the relative absence of widely reported wrong-target incidents, the apparent knowledge of routines and locations that close-range attacks require, all of it suggests an intelligence picture of considerable quality. A motorcycle assassination is an unforgiving method. It demands that the operators know where a particular man will be, at what time, and by what route he will leave, and it offers no standoff distance to absorb an error. That the method has been used repeatedly without the kind of catastrophic misidentification that defined the rendition program is itself a data point about the underlying intelligence. The campaign’s analytical foundation, the way its picture of the Pakistani militant landscape is assembled, connects to the longer story of how Indian intelligence built its doctrinal framework for offensive action. On the first dimension, then, the foundations dimension, India’s campaign passes both tests, and it passes them for a reason worth naming: it was designed by an institution that had already paid, in Sri Lankan jungles, the full price of getting the foundations wrong.

Read together, the two foundational tests deliver a verdict that is more demanding than it first appears. An operation does not pass this dimension by being clear about its goal or by gathering a great deal of intelligence. It passes by having a goal that is clear, achievable, and bounded, supported by a picture that has survived adversarial challenge from people who had no stake in launching the action. By that strict combined standard, India’s shadow war passes, and it passes because the institution behind it had already failed the same dimension once, in a way too costly to forget.

The Second Dimension: The Temperament of Execution

If the first dimension concerns the foundations laid before an operation begins, the second concerns the temperament with which it is carried out. Two tests define that temperament. Operational discipline asks whether the people executing the action did so with rigor, restraint, and adherence to the plan. Political patience asks whether the sponsoring government allowed the action the time it required rather than demanding visible results on a political calendar. These are tests of character more than of design, and they sort the eleven cases in ways that the first dimension does not.

Operational discipline is the quality that the public most associates with the word professional, and the successes display it abundantly. The Eichmann team rehearsed, scouted, built fallback plans, and held a kidnapped man in a safe house for days without incident before the extraction. The Neptune Spear element trained against a full-scale mock-up of the Abbottabad compound and adapted in real time when a helicopter was lost on insertion. The Stuxnet developers built a weapon so disciplined that it was engineered to spread cautiously and to act only against a precisely fingerprinted configuration, a restraint designed to limit its spread beyond the intended target. Discipline in these cases was not merely the absence of mistakes; it was a positive practice, a culture of rehearsal and contingency that treated the plan as a living thing to be stress-tested rather than a script to be recited.

The failures divide on this test in an instructive way, because some of them were undisciplined in execution and others were disciplined in execution but doomed by design, and the diagnostic must keep those two situations separate. The Bay of Pigs was, in its tactical execution, not wildly incompetent; the brigade fought hard. It was the design that was fatal, and discipline cannot rescue a design that was wrong. Iran-Contra, by contrast, was undisciplined in the deepest sense, a freelancing operation run partly outside the normal chain of accountability, with the predictable result that no one was fully in control of it. The rendition program was disciplined in a narrow procedural sense, with its aircraft and its black sites and its protocols, yet that very procedural discipline, applied to a fundamentally lawless practice, only made the eventual exposure more damning, because the paper trail of a disciplined illegal program is itself the evidence against it.

Wrath of God is the case that makes operational discipline vivid, and it does so through its single worst moment. In July 1973, in the Norwegian town of Lillehammer, a Mossad team killed a Moroccan waiter named Ahmed Bouchiki, having mistaken him for a senior Black September figure. Norwegian police arrested several of the agents. The Lillehammer affair is studied to this day because it demonstrates that even an intelligence service with a deserved reputation for competence can suffer a discipline collapse, a failure of verification under the momentum of a long campaign, and that the cost of that collapse is both a human life and a strategic embarrassment that no number of successful eliminations can erase. Avery Plaw’s analysis of targeted-killing campaigns treats Lillehammer not as an aberration but as a structural risk inherent to any program that runs long enough, because tempo erodes verification and a campaign that never ends will eventually have its Lillehammer.

Discipline, properly understood, is not a fixed trait that an agency either has or lacks. It is a quantity that erodes under specific, identifiable pressures, and naming those pressures is what makes the Lillehammer warning useful rather than merely ominous. The first pressure is tempo. An agency conducting one operation a year can rehearse exhaustively; an agency conducting one a month cannot, and the verification that prevents a fatal error is the first thing quietly sacrificed when the calendar tightens. The second pressure is success itself, because a long run of operations that go well breeds the unspoken conviction that the next one will go well too, and that confidence is the natural enemy of the paranoid double-checking that sound tradecraft demands. The third pressure is institutional fatigue, the simple wearing-down of the experienced people who must personally sustain the standard year after year. None of these pressures is visible in a snapshot. All of them compound over time. This is why the discipline of a young campaign tells an observer far less than it appears to, and why the only honest assessment of operational discipline is one that asks not how disciplined the campaign looks today but how disciplined it will still be after another five years at the same tempo.

India’s shadow war scores high on operational discipline, and the evidence is the method itself. The motorcycle-borne close-range shooting, repeated across many incidents with a consistent signature, is a disciplined method executed disciplinedly. The assailants reach the target, fire, and escape through congested streets where pursuit fails, and the consistency of that pattern across years and locations is the fingerprint of trained operators working to a refined template rather than improvising. The campaign’s modus operandi, the recurring choreography that ties the incidents together, is examined in the analysis of the unknown gunmen pattern, and the relevant point for this diagnostic is that disciplined repetition is a strength. Yet the Lillehammer warning sits directly on top of this strength and must be stated honestly. The discipline India has displayed so far is a snapshot, not a guarantee. The longer a campaign of this kind runs, the more its tempo presses against the verification that prevents a Lillehammer, and the diagnostic returns to this danger when it reaches the fourth dimension.

One further observation about discipline deserves its place before the analysis moves on, because it cuts against an intuition that outsiders commonly hold. The public tends to assume that the most disciplined covert operation is the one that looks most aggressive and decisive. The cases suggest close to the opposite. The most disciplined operations in this set were frequently the most restrained, the ones whose operators declined opportunities, aborted when conditions were wrong, and accepted a slower result rather than a riskier one. Stuxnet’s engineering of caution into the weapon itself, its deliberate limits on how far and how fast it would spread, is discipline of exactly this restrained kind. The Eichmann team’s willingness to wait, to hold rather than to rush, is another instance of the same quality. Discipline, in covert action, is mostly the discipline of not acting, of tolerating a target who slips away today because acting today would have been wrong, and an agency that cannot tolerate the missed opportunity is an agency that will eventually convert a missed opportunity into a Lillehammer.

Political patience is the second test of temperament, and it is the one that democracies fail most often, because democratic timelines are short and covert results are slow. The successes were uniformly patient. The Eichmann hunt took years. The bin Laden hunt took roughly a decade, spanning two American presidencies, and the patience was institutional rather than personal, sustained across a change of administration. The Bangladesh effort was patient in a different sense, willing to build the Mukti Bahini’s capacity over months rather than demanding a premature spectacle. Patience, in these cases, was the political class granting the intelligence service the most precious and least glamorous resource it can give, which is time.

The failures often failed because patience ran out or was never extended. The Bay of Pigs was, in part, a victim of a calendar; the planning had a momentum that made postponement feel like cancellation, and a new administration inherited an operation it was reluctant either to fully own or to fully stop. Iran-Contra was the opposite of patient, an attempt to produce results, hostage releases above all, on a timeline driven by political anxiety rather than by the slow rhythm of sound tradecraft. India’s shadow war, by contrast, displays conspicuous patience. The campaign has unfolded over years at a deliberate tempo, without the appearance of being accelerated to satisfy a domestic political moment, and the absence of official acknowledgment, whatever else it costs, has the side effect of insulating the campaign from the pressure to perform on cue. A program no one admits to cannot be rushed by a public demanding to see its results. On the second dimension, therefore, India’s campaign again passes both tests, with the single, serious asterisk that the discipline is being tested against time, and time has not yet finished its examination.

Considered as a whole, the second dimension tests an agency’s temperament rather than its planning, and temperament is harder to audit than design because it does not show up on a chart. An operation can be inspected before it launches and found sound; its discipline and the patience it will be granted can only be observed as it runs. India’s shadow war displays both qualities at present, and that present-tense observation is precisely the one the fourth dimension will complicate, because temperament under five years of pressure is not the same as temperament in year one.

The Third Dimension: Staying Invisible and Staying Clean

The third dimension turns from the temperament of execution to the management of consequences. Two tests govern it. Deniability maintenance asks whether the sponsoring state was able to keep its hand sufficiently hidden to avoid the political and legal costs of acknowledged action. Collateral-damage control asks whether the operation confined its harm to its intended targets rather than killing, imprisoning, or ruining people it never meant to touch. These two tests are linked, because the fastest way to lose deniability is to harm an innocent, and the rendition program is the case where both tests fail together.

Deniability is a strange asset, because the public tends to view it as mere cowardice, a state declining to own its actions, when in fact it is often the load-bearing wall of a covert program. Deniability is what allows an adversary to absorb a blow without being forced, by domestic humiliation, into a retaliation that escalates. It is what keeps an operation below the threshold that triggers a wider war. The successes in this set understood that asymmetrically. The Eichmann capture sacrificed deniability deliberately, because Israel wanted the trial and the trial required ownership, and the operation was bounded enough that the sacrifice was affordable. Stuxnet, by contrast, was built to be deniable and remained formally unacknowledged even after it was technically attributed, and that formal silence preserved a margin of maneuver for everyone involved. Neptune Spear was acknowledged within hours, but its deniability had never been the point; the United States wanted the world, and bin Laden’s followers, to know.

The failures show deniability collapsing in slow motion. Iran-Contra was exposed and then could not be contained, and the exposure was the disaster. The rendition program leaked steadily, through aircraft tail numbers tracked by aviation enthusiasts, through host countries whose courts and parliaments began to investigate, until the deniability was simply gone and the United States stood publicly accused of running a global archipelago of secret prisons. Operation Cyclone is an interesting middle case, because its deniability was always thin, the arming of the mujahideen being a poorly kept secret, and the thinness mattered less than it should have only because the operation enjoyed broad international tolerance.

It is worth pausing on why deniability erodes, because the cases show that it rarely collapses through a single dramatic leak. It erodes the way a coastline erodes, through the steady accumulation of small disclosures, no single one of which seems decisive. The rendition program was not exposed by one whistleblower; it was exposed by aircraft tail numbers logged at airports by hobbyists, by a contractor’s billing records, by a host country’s magistrate who would not let an investigation drop, by a survivor who lived to tell his story, each fragment joining the others until the composite picture became impossible to deny. This is the quiet arithmetic of a long covert program. Every operation generates a small quantity of evidence, a witness, a shell casing, a frame of surveillance footage, a recurring pattern, and while any single fragment remains deniable, the fragments accumulate, and they do not decay. A campaign that has run for five years has generated five years of fragments, and the question is never whether the fragments exist but only whether anyone has yet troubled to assemble them.

India’s shadow war scores extremely high on deniability maintenance, and this is, by some distance, the campaign’s most impressive feature. No Indian official has claimed responsibility for any of the killings. The Indian government’s posture has been consistent denial. The strongest public evidence connecting India to the pattern, including a notable 2024 investigation by The Guardian citing unnamed intelligence sources, remains circumstantial, and no foreign government has produced the kind of proof that would force the issue. The motorcycle method is itself a deniability instrument, because a close-range shooting by men who escape into traffic leaves a far thinner forensic trail than a missile, a drone, or an airstrike, and it is plausibly deniable in a way that a crater is not. The campaign has, in effect, engineered the Stuxnet property, action that everyone suspects and no one can prove, and it has sustained that property across years. On this test it outperforms most of the historical successes.

Collateral-damage control is the test where the moral and the strategic finally fuse, because in covert action the unintended victim is not only a human tragedy; the unintended victim is the single most reliable destroyer of deniability. The successes were, by the brutal standard of this field, comparatively contained. The Eichmann capture harmed no one but Eichmann. Neptune Spear, though it occurred in a populated compound, was a precise raid rather than an area attack. Stuxnet, remarkably, killed no one at all, which is part of why it is studied as a model; it degraded machines. The failures, predictably, were not contained. The rendition program’s collateral damage has names, el-Masri foremost among them, and the existence of those names is the reason the program could not survive scrutiny. Operation Cyclone’s collateral damage was the entire downstream ecosystem of violence its weapons enabled, a collateral cost so large and so delayed that the accounting is still open.

Wrath of God, once again, is the bridge case, and Lillehammer is once again the reason. The killing of Ahmed Bouchiki was a collateral-damage failure and a deniability failure in the same instant, because the innocent victim produced the arrests that produced the exposure. The two tests of the third dimension are not really separate; they are the same wall seen from two sides.

The link between the two tests of this dimension deserves to be stated as a principle, because it is the single most useful thing the third dimension teaches. An innocent victim is not merely a moral cost to be weighed against an operational gain. An innocent victim is the mechanism by which a deniable program becomes an undeniable one. The reason is structural rather than incidental. The intended target of a covert killing is, almost by definition, a person whose death few will investigate energetically, whose associates have reasons of their own to stay silent, and whose own government may quietly prefer the matter closed. The unintended victim is the opposite in every respect. The innocent leaves behind a grieving family with no reason at all for silence, a home government with every incentive to pursue the case loudly, and a press that will treat the death as a scandal rather than as a settling of accounts among violent men. This is why Lillehammer destroyed what it destroyed. The dead waiter was not simply a failure of one operation standing alongside the campaign’s successes; he was the loose thread that, once pulled, unravelled the deniability of everything else the campaign had done.

How does India’s shadow war score on collateral-damage control? Well, so far, and the qualifier is doing heavy work. The targeted killings, as reported, have been narrowly aimed at specific individuals, and the close-range method, for all its risks, is in one respect a precision instrument, because it allows the operators to confirm the target visually at the moment of action in a way that no remote strike permits. This is the consideration that the comparison of methods of elimination returns to repeatedly: the human assassin sees the face. A drone sees a heat signature and a pattern of life. But the Lillehammer precedent forbids any comfortable conclusion. The shadow war’s collateral record is clean for now, and for now is precisely the phrase that the fourth dimension is about to interrogate, because every long campaign believes its record is clean right up until the day it is not.

A closing observation about the third dimension connects it to everything the fourth dimension is about to argue. Deniability and collateral-damage control are not merely two more tests on a checklist; they are the two tests most sensitive to the passage of time, and that sensitivity is what makes them the hinge of the whole analysis. An operation’s objective clarity is fixed at the moment of design and does not decay. Its intelligence foundation, once sound, tends to stay sound. But deniability is consumed by every operation, and the probability of a collateral-damage failure compounds with every operation, which means a campaign can pass both tests cleanly in its early years and still be quietly accumulating the conditions of its own exposure. This is the precise sense in which a strong third-dimension score on a young campaign is less reassuring than it looks. The score is real, but it is a measurement of a quantity that is being spent, and the rate of spending, not the current balance, is what the fourth dimension forces into view.

The Fourth Dimension: Knowing When to Stop and Why You Started

The fourth dimension is the one that the public discussion of covert action almost always neglects, and it is the one on which India’s shadow war is most exposed. Two tests define it. Exit criteria asks whether the operation had, from the outset, a defined condition that would constitute completion, a stated answer to the question of when it ends. Strategic-goal alignment asks whether the operation, even when tactically successful, actually advanced the sponsoring state’s larger and longer-term position rather than winning a local point while losing a strategic argument. These two tests are the deepest in the diagnostic, because an operation can pass all six of the preceding tests and still, by failing these two, become a slow-motion disaster.

Consider exit criteria first, and consider how rare a genuine exit criterion is. The Eichmann capture had a perfect one, built into its physics: the operation ended when the man reached Israel, and it could not continue past that point because its objective was a single, completable act. Neptune Spear had the same property; bin Laden could be killed exactly once, and his death was self-evidently the end. Stuxnet had a softer but real exit criterion, since the operation’s useful life ended when the malware was discovered, and the planners appear to have understood that the program had a finite shelf life. The Bangladesh effort had the cleanest exit of all, because it was tethered to a conventional war, and conventional wars end with instruments of surrender and dated ceasefires. RAW’s role in 1971 ended when the war ended, on a specific day, and the agency went home.

Now consider the operations that had no exit criterion, and notice that the absence of an exit criterion is a quieter flaw than a botched landing or a leaked cable, which is exactly why it is so dangerous. Operation Cyclone had no exit criterion worth the name. The program’s near-term goal, Soviet withdrawal, was reached, and then the operation did not end, because no one had ever specified that withdrawal was the finish line, and the infrastructure simply continued to exist and to metastasize. The RAW-LTTE relationship had no exit criterion either, which is why it could drift from leverage to sponsorship to open warfare without anyone ever making a decision to cross those lines; in the absence of a defined endpoint, an operation does not end, it transforms, and the transformation is rarely toward something better. Wrath of God, the bridge case, had no exit criterion, and this is the second of its two structural flaws. The campaign to avenge Munich had no stated condition under which Israel would declare the account settled and stand down, and so it ran, in various forms, for decades, and a campaign that runs for decades is a campaign that will, by the simple mathematics of repeated risk, eventually produce its Lillehammer.

There is a psychological dimension to the exit-criteria problem that the institutional framing can obscure, and it explains why the flaw proves so persistent across otherwise dissimilar agencies. An operation with a defined endpoint has, built into it, a moment of decision that some identifiable person must own: the moment of declaring the thing finished. An operation without a defined endpoint never produces that moment, and so it never requires anyone to take responsibility for stopping. Stopping becomes, instead, a positive act that some individual would have to initiate, against the institutional momentum, against the colleagues whose careers are invested in continuation, and against the uncomfortable implication that the program might no longer be necessary or, worse, might never have been. It is far easier, inside any bureaucracy, to let a successful program continue than to be the person who argued for ending it and was then proved wrong by an attack the program might have prevented. The absence of an exit criterion is therefore not a mere oversight in the planning. It is a structural arrangement that quietly removes the obligation to ever ask the hardest question, and programs designed that way tend to outlive not only their usefulness but the strategic conditions that once justified them.

This brings the diagnostic to the question that the analysis of India’s shadow war cannot avoid, and that intelligence scholars would frame as the central analytical disagreement of the entire subject. Does the shadow war have an exit criterion? And if it does not, does that absence make it unsustainable, or is a permanent, low-intensity covert campaign a viable strategic posture that a state can simply maintain indefinitely the way it maintains a coast guard?

The honest answer begins with an observation about the campaign’s trajectory. A program with a defined endpoint tends to wind down as it approaches that endpoint. A program without one tends, instead, to escalate, because each completed task creates the appetite and the confidence for a more ambitious one, and the absence of a stated finish line removes the natural brake. The reported pattern of the shadow war, read across its full timeline, shows a campaign that has not contracted but expanded, reaching, over time, for higher-value targets and operating across a wider geography. That trajectory is the signature of a campaign without an exit criterion. It is not the signature of a campaign that is failing; it is, on the contrary, the signature of a campaign that is succeeding so reliably that no one inside it has any incentive to ask when it stops. And that, precisely, is the danger, because the historical record is unambiguous on the point that Loch Johnson states most directly: the covert programs that end well are the ones that were designed to end, and the covert programs that end badly are, with remarkable consistency, the ones that were never designed to end at all.

Both positions in the disagreement deserve their fairest statement. The case that a permanent campaign is viable runs as follows. India’s shadow war is disciplined, deniable, cheap relative to conventional military action, and tightly bounded in its objectives, and a state that maintains those properties can, in principle, run such a campaign at low cost for a very long time, treating the elimination of wanted militants as a routine function of statecraft rather than as a finite operation. By this reasoning, the demand for an exit criterion is a category error, because the campaign is not a project to be completed but a capability to be sustained, and capabilities do not need endpoints.

The opposing case, which this analysis finds more persuasive while acknowledging the genuine force of the first, runs as follows. Every covert capability that was treated as permanent has, eventually, produced a failure that a finite operation would have avoided, because permanence multiplies the number of times the dice are rolled, and the dice, rolled enough times, produce a Lillehammer, an el-Masri, a captured operative, an operation that kills the wrong man, a piece of evidence that finally converts circumstantial suspicion into undeniable proof. Deniability is not a renewable resource; it is a finite one, depleted a little by every operation, and a campaign with no exit criterion is a campaign that has chosen to spend that resource until it runs out. The risk is not that the shadow war fails tomorrow. The risk is that it succeeds, and succeeds, and succeeds, and that the very reliability of the success becomes the reason no one notices the deniability draining away until the day a single operation goes wrong and converts a deniable program into an acknowledged act of war between two nuclear-armed states. The adjudication, therefore, is not that a permanent campaign is impossible. It is that a permanent campaign is a wager that the dimension-three wall will hold forever, and no covert program in the historical record has ever won that wager indefinitely.

One objection to this adjudication should be met directly, because it is the strongest the other side can make. It might be argued that the absence of a written exit criterion does not prove the absence of a real one, and that India’s planners may well hold, privately, a clear internal sense of the conditions under which the campaign would wind down, perhaps the exhaustion of a finite list of high-value names, perhaps a verifiable change in Pakistani state behavior toward the organizations in question. This is entirely possible, and the analysis cannot disprove it. But the historical record suggests that an unwritten, unshared exit criterion provides little of the protection that a defined and external one provides, because the value of an exit criterion lies precisely in its being a fixed commitment that constrains the program even when momentum and success are arguing loudly for continuation. An exit criterion held only in the minds of the current leadership is one that the next leadership will not inherit, that no one outside the room is accountable to, and that will quietly expand as each near-completion creates the appetite for one more name. A criterion that bends whenever the campaign is going well is not a criterion at all. It is a hope, and the Bay of Pigs is the enduring monument to what happens when a hope is mistaken for a plan.

The second test of the fourth dimension, strategic-goal alignment, sharpens the same concern from a different angle. An operation passes this test when its tactical success genuinely serves the state’s larger position. It fails when the tactical success is real but the strategic effect is neutral or negative. Operation Cyclone is the monument to this failure, a tactical and even an operational success, the Soviets did leave, that was a strategic catastrophe, because the operation’s downstream consequences harmed the sponsor far more than the original Soviet presence ever had. The rendition program failed this test as well; whatever fragments of intelligence it produced, the strategic cost in alliance friction, legal exposure, and moral authority was not remotely repaid.

For India’s shadow war, the strategic-goal-alignment question is the one that should be debated inside the Indian system with the most rigor, and it is genuinely open. The campaign’s tactical record is strong. But the strategic question is whether the elimination of individual militants, however many, actually degrades the organizations and the state apparatus that produce them, or whether it removes replaceable operatives from structures that simply promote successors, while the strategic relationship with Pakistan, and the international perception of India, absorbs a slow accumulating cost. This is the question that connects the present campaign to the deepest argument of the series, and it is the question that the pillar analysis of the shadow war frames as the campaign’s true test. On the fourth dimension, then, the verdict on India’s shadow war is split. On exit criteria, it scores poorly, not because it has failed but because it has never been designed to stop. On strategic-goal alignment, the verdict is genuinely undecided, and it will be decided not by the next successful elimination but by whether the cumulative campaign moves the larger relationship in India’s favor or merely keeps a wound open.

Where the Comparison Breaks Down

A diagnostic of this kind earns trust only if it admits what it cannot do, and a comparison across sixty years and three intelligence cultures has real limits that an honest analysis must put on the table rather than hide.

The first limit is the problem of unique circumstance. Each of the eleven cases occurred in a specific historical setting that cannot be reproduced, and pattern extraction always risks flattening that specificity into a false universality. The Bangladesh effort succeeded partly because it rode a genuine popular uprising and a conventional war that was going to be won regardless; lift RAW’s 1971 conduct out of that context and drop it into a different one and the same conduct might fail completely. The Bay of Pigs failed partly because of a specific transition between two American administrations with different appetites for risk; in a different political moment the same plan might have been postponed and reworked rather than launched. To say that objective clarity correlates with success is true, but it is a correlation drawn from cases whose other variables were never held constant, and no reader should mistake the diagnostic for a formula that guarantees an outcome.

A second limit is the asymmetry of evidence, and it bears most heavily on the case the analysis most wants to assess. The ten historical operations have been studied for years or decades. There are official inquiries, declassified documents, memoirs, and the slow sediment of scholarship. India’s shadow war has none of that. It is assessed here through pattern, inference, and a small body of investigative journalism, and the campaign’s sponsor has confirmed nothing. Every score assigned to the shadow war in this diagnostic is therefore a score assigned to a reconstruction, not to a documented record, and the reconstruction could be wrong. It is possible, though the analysis judges it unlikely, that some of the killings attributed to a coordinated Indian campaign are in truth the product of internal Pakistani militant rivalries or of the Pakistani state’s own settling of accounts. The diagnostic’s verdict on India is offered with a confidence proportional to the evidence, which is to say with real but bounded confidence, and a reader who weighs the evidence differently is entitled to a different verdict.

The third limit is that the very framework of success and failure smuggles in a sponsor’s-eye view of the world. To call the Eichmann capture a success is to adopt, for the length of the sentence, the objectives of the Israeli state, and to call the Bay of Pigs a failure is to adopt the objectives of the American one. The diagnostic measures whether an operation achieved what its sponsor wanted at a cost the sponsor would pay. It does not, and cannot, measure whether the sponsor should have wanted that thing at all. A covert killing can be a flawless operational success and a moral catastrophe at the same time, and nothing in the eight tests captures that second judgment, which belongs to law and ethics rather than to the analysis of tradecraft. Readers who want that second judgment will find the legal and ethical argument treated directly in the examination of the targeted killing debate, and they should hold the operational verdict and the moral verdict in separate hands.

A fourth limit concerns the survivorship bias that haunts any study of covert action, and it cuts against the diagnostic’s confidence in both directions at once. The operations available to be studied are, overwhelmingly, the ones that became known, and an operation becomes known either because it succeeded spectacularly enough to be claimed or because it failed badly enough to be exposed. The vast middle, the covert actions that were modest, inconclusive, or quietly abandoned before they ripened into anything, leaves little trace in the public record, which means the ten historical cases here are drawn from the two tails of the distribution rather than from its body. This distorts the lessons in a subtle but real way. It may make success and failure look more sharply separated than they truly are, and more cleanly attributable to the eight variables than a complete record would show, because the ambiguous cases that would blur the pattern are precisely the ones that never surface to be counted. A reader should therefore treat the diagnostic as a tool sharpened on extreme cases, reliable for identifying the structural flaws that produce dramatic failure, and considerably less reliable for predicting the muddy, partial outcomes that are probably the true statistical norm of covert action.

The fifth limit is the one the analysis is least able to resolve, the limit of time. The diagnostic can score a completed operation because a completed operation has revealed all of its consequences. It cannot fully score an operation that is still running, because the most important consequences of a covert campaign are frequently the delayed ones. Operation Cyclone looked like a success in 1989 and revealed itself as a strategic failure only across the following decade. India’s shadow war, in 2026, is roughly at the point in its life where Cyclone was in 1989, the point of apparent success, and the diagnostic is therefore scoring it before its delayed consequences, whatever they prove to be, have arrived. This is not a flaw in the framework. It is a flaw in the timing of the question, and the only honest response to it is to say plainly that the verdict on India’s shadow war is provisional, and that the variable that will revise it is not skill but time.

What the Comparison Teaches

Strip the eleven cases down to their pattern and a small number of lessons survive, durable enough to be stated as conclusions rather than as observations, and useful enough that an intelligence service ignoring them does so at a known and documented risk.

The first lesson is that covert action is governed by its foundations far more than by its execution. Across the cases, the operations that failed almost always carried the seed of failure in their design, in a muddled objective or an unverified assumption, and the operations that succeeded almost always carried the seed of success in a clear objective and a sound intelligence picture. Brilliant execution cannot rescue a bad design; the Bay of Pigs brigade fought hard and still lost, because no amount of fighting can win an operation whose central premise was false. This is the lesson that India’s planners appear to have internalized most completely, and the evidence is the contrast between the LTTE disaster and the present campaign. The shadow war is the work of an institution that learned, at the cost of more than twelve hundred soldiers and a former prime minister, that the foundations are the operation.

A second lesson is that the most dangerous flaws are the quiet ones. A botched landing announces itself within hours. A leaked cable announces itself within weeks. But the absence of an exit criterion announces itself only after years, sometimes decades, when the operation that was never designed to stop has run long enough to produce the failure that a finite operation would have avoided. This is the lesson that the comparison of Operation Cyclone and Wrath of God teaches with the most force, and it is the lesson that bears most directly on India’s campaign, because the shadow war’s single clear vulnerability, measured against the historical record, is not a weakness anyone can see in 2026. It is the structural fact that the campaign has no stated condition under which it ends.

The third lesson follows from the second, and it is the answer to the question of what failure would actually look like for India’s shadow war, a question worth answering concretely rather than leaving as a vague unease. Failure, if it comes, will almost certainly not look like a single dramatic catastrophe announced on a single day. It will look like one of a small number of slow or sudden converging events that the historical cases have already rehearsed. It could look like a Lillehammer, an operation that kills the wrong person and converts an innocent’s death into the arrests and the evidence that destroy deniability. It could look like a captured operative whose interrogation or whose documents finally convert circumstantial suspicion into the kind of proof that a government cannot wave away, the proof that would oblige Pakistan, for reasons of its own domestic humiliation, to respond in a manner that neither state could then control. It could look like a strategic-alignment failure, in which the campaign continues to eliminate replaceable individuals for another decade while the organizations regenerate and the larger relationship with Pakistan, and India’s standing with the Western governments whose territory the wider dispute has touched, slowly absorbs a cost that no single elimination ever seemed to justify. Or it could look like the quietest failure of all, the campaign simply continuing forever, never defeated and never concluded, a permanent low-grade hemorrhage of deniability and diplomatic capital that no one inside the system is ever incentivized to stop because stopping it would require admitting it existed.

There is a further lesson, less about operations than about the institutions that run them, and it is the single note of genuine optimism the comparison permits. Institutions can learn, and the proof is India itself. The distance between the agency that armed the Tamil Tigers and the agency that, if the allegations hold, designed the present campaign is the distance of a hard lesson absorbed at full cost. The earlier institution built a proxy and lost control of it; the later institution appears to have built a methodology that owns no proxy at all, that reaches the individual directly and then withdraws, leaving nothing behind that can later be turned against its maker. That is not an accident of circumstance. It is the visible residue of an organization that studied its own worst failure and rebuilt its doctrine around never repeating it. The encouraging implication is that the one test the shadow war currently fails, the absence of an exit criterion, is not a fixed feature of the campaign but a design choice, and design choices can be revised by exactly the same institutional learning that fixed the proxy problem. The discouraging implication is that institutions usually learn a lesson only after paying its full price, and the price of the exit-criteria lesson, in every historical case examined here, has been a failure severe enough to force the reckoning.

The fourth lesson is the one that should temper any reader’s verdict in either direction. By the eight-test standard built from the historical record, India’s shadow war is, as of 2026, a well-run covert campaign. It has a clear and bounded objective. It rests on an intelligence picture of evident quality. It is executed with discipline and sustained with patience. It has maintained deniability to a degree that outperforms most of the historical successes, and its collateral record, so far, is clean. On six of the eight tests it scores at or near the top of the eleven cases. It fails one test outright, exit criteria, and it leaves one test, strategic-goal alignment, genuinely undecided. That is, by any fair reading, a strong scorecard, and the analysis would be dishonest if it pretended otherwise in order to manufacture a tidy warning. The campaign is not the Bay of Pigs. It is not Iran-Contra. It is not, on its current record, the LTTE.

But the final lesson is that a strong scorecard on a campaign still running is a snapshot, and the historical record is a film. Operation Cyclone had a strong scorecard in 1989. The men who armed the mujahideen could have pointed, with justification, to a clear objective, sound execution, sufficient patience, and a Soviet army in retreat. What they could not point to was an exit criterion or a clear-eyed account of strategic alignment, and the two tests they failed were the two that the following decade would reveal as decisive. India’s shadow war today is a more disciplined and more deniable campaign than Cyclone ever was, and it deserves the credit that this analysis has tried to give it precisely. The warning that the comparison leaves is not a prediction of failure. It is a single, specific, historically grounded observation: the covert campaigns that ended well were the ones that were designed to end, and the most important decision India’s planners have not yet visibly made is the decision about what, exactly, would have to be true for the shadow war to be declared finished. Until that decision is made, the campaign’s most impressive strengths are also the reason no one inside it will ever be the person who decides to stop.

A final word belongs to the reader who finishes this analysis wanting a single sentence to carry away. It is this: the covert operations that aged well were the ones whose designers could answer, before launching, the question of what completion would look like, and the operations that aged badly were the ones whose designers never asked. India’s shadow war answers six of the eight questions a covert program must answer, and answers them well, which places it among the more competent campaigns in the historical record rather than among the cautionary disasters. But the question it has not visibly answered, the question of when and on what condition the campaign ends, is not a minor omission. It is the question that the comparative record identifies, again and again, as the one whose absence does the eventual damage. The campaign’s strengths are genuine and this analysis has tried to credit them without flinching. The warning it leaves is equally genuine, and it is narrow enough to be acted upon: a program this disciplined deserves an ending designed with the same discipline, and the design of that ending is the one piece of work that remains undone.

Frequently Asked Questions

What makes covert operations succeed or fail?

The single most consistent finding from the historical record is that covert operations are decided by their foundations rather than by their execution. Operations with a clear, bounded, achievable objective and an accurate intelligence picture tend to succeed; operations built on a muddled objective or an unverified assumption tend to fail, and skilled execution cannot rescue a flawed design. Beyond the foundations, the recurring variables are operational discipline, political patience, the preservation of deniability, the control of collateral damage, the presence of a defined endpoint, and genuine alignment between the operation’s tactical result and the sponsor’s larger strategic position.

What are the eight criteria for assessing a covert operation?

This analysis uses eight tests drawn from the patterns visible across sixty years of intelligence history: objective clarity, meaning a goal that can be stated in one sentence and completed; intelligence accuracy, meaning a verified rather than a hoped-for picture; operational discipline, meaning rigor and contingency in execution; political patience, meaning the willingness to grant the action the time it needs; exit criteria, meaning a defined condition that constitutes completion; deniability maintenance, meaning the ability to keep the sponsoring hand hidden; collateral-damage control, meaning harm confined to intended targets; and strategic-goal alignment, meaning that tactical success genuinely advances the state’s long-term position.

Which covert operations in history were most successful?

By the standard of achieving a stated objective at an acceptable cost, the strongest historical performers include the Israeli capture of Adolf Eichmann in 1960, a bounded operation that ended cleanly when its single target reached an Israeli courtroom; the Indian intelligence support for the liberation of Bangladesh in 1971, which assisted an existing uprising toward a defined finish line and then concluded; Operation Neptune Spear, the 2011 raid that killed Osama bin Laden; and the Stuxnet cyber operation, which delayed Iran’s nuclear enrichment without firing a shot. Each succeeded partly because its objective was narrow and its endpoint was clear.

Which were the biggest covert-action failures?

The most instructive failures include the 1961 Bay of Pigs invasion, which pursued regime change on the false assumption of a popular uprising; the Iran-Contra affair of the 1980s, which tangled incompatible objectives into a scheme that collapsed when exposed; India’s sponsorship of Tamil militant groups in Sri Lanka, which produced a proxy that India later had to fight and that assassinated a former Indian prime minister; Operation Cyclone, the CIA program to arm the Afghan mujahideen, which won its near-term goal but seeded a long-term strategic disaster; and the United States extraordinary rendition program, which lost its deniability, harmed innocent people, and produced little reliable intelligence.

Does India’s shadow war meet the criteria for success?

Measured against the eight tests, India’s campaign of targeted killings scores well on six of them. Its objective is clear and bounded, its intelligence picture appears to be of high quality, its execution is disciplined, its tempo is patient, its deniability has been maintained to an unusual degree, and its collateral record, so far, is clean. It scores poorly on one test, exit criteria, because the campaign has no stated condition under which it ends, and it leaves a second test, strategic-goal alignment, genuinely undecided. The overall scorecard is strong, with the important qualification that it is being assessed while the campaign is still running.

What is the most important predictor of covert-action success?

If the eight tests had to be reduced to one, objective clarity would be the strongest single predictor. Across the historical cases, operations with an objective that could be stated in a single sentence and completed as a discrete act, capturing one man, killing one man, delaying one industrial process, almost uniformly succeeded, while operations whose objective was a broad political outcome such as regime change, or a bundle of competing aims, almost uniformly failed. A clear objective also tends to carry a built-in exit criterion, which is why it matters so much.

Does the shadow war have exit criteria?

On the available evidence, no. India’s campaign of targeted killings has no visible, stated condition under which it would be declared complete. This is the campaign’s clearest structural vulnerability when it is measured against the historical record, because the covert programs that ended well were generally the ones designed to end, while the programs that ran indefinitely, such as Operation Cyclone and Operation Wrath of God, tended to accumulate the delayed failures that finite operations avoided. The reported expansion of the campaign over time, rather than its contraction, is itself a symptom of the absence of a defined endpoint.

Can a permanent covert campaign be sustainable?

This is the central analytical disagreement of the subject. One position holds that a disciplined, deniable, cheap, and tightly bounded campaign can be maintained indefinitely as a routine capability rather than a finite project. The opposing position, which this analysis finds more persuasive, holds that deniability is a finite resource depleted by every operation, that permanence multiplies the number of times risk is rolled, and that every covert capability treated as permanent has eventually produced a failure a finite operation would have avoided. A permanent campaign is therefore best understood as a wager that the program’s strengths will hold forever, and the historical record contains no example of that wager being won indefinitely.

What would failure look like for India’s shadow war?

Failure, if it comes, is unlikely to be a single dramatic event. It is more likely to resemble one of the patterns the historical cases have already rehearsed: an operation that kills the wrong person and converts an innocent’s death into the evidence that destroys deniability, as the Lillehammer affair did to Operation Wrath of God; a captured operative or a recovered document that converts circumstantial suspicion into undeniable proof and forces an escalation neither state can control; a strategic-alignment failure in which the campaign removes replaceable individuals for years while the organizations regenerate and the larger relationship absorbs a mounting cost; or the quietest failure of all, the campaign simply continuing forever as a permanent low-grade drain on deniability and diplomatic capital.

Why did the Bay of Pigs fail?

The 1961 Bay of Pigs invasion failed because its design was flawed before a single exile landed. Its objective, the overthrow of Fidel Castro, was not a discrete act an operation could deliver but a political outcome dependent on the behavior of Cuban society. Its central intelligence assumption, that the landing would trigger a popular uprising, was a hope rather than a verified finding, and the agency’s planners did not subject that hope to adversarial scrutiny. The operation had no recovery option once the assumption proved false, and the brigade was pinned and captured within three days. The Kennedy administration’s own internal review identified the failure as one of planning and assumption.

Why is the Eichmann capture considered a model operation?

The 1960 capture of Adolf Eichmann is treated as a model because it satisfied almost every criterion of sound covert action simultaneously. Its objective was singular and completable: locate, seize, and deliver one man for trial. It rested on years of patient and verified intelligence work. It was executed with rehearsed discipline, including the holding of a captured man in a safe house for days without incident. And it had a perfect exit criterion built into its physics, since the operation ended, and could only end, when Eichmann reached Israel. Even its deliberate sacrifice of deniability, necessary because Israel wanted a public trial, was affordable because the act was so tightly bounded.

What made Operation Wrath of God a partial rather than a full success?

Operation Wrath of God, Israel’s campaign against the Palestinian networks linked to the 1972 Munich attack, reached many of its intended targets and so succeeded by the narrow measure of elimination. It is judged only a partial success because it failed two structural tests. The Lillehammer affair of 1973, in which a Mossad team killed an innocent Moroccan waiter mistaken for a senior target, was a simultaneous failure of operational discipline and collateral-damage control, and the resulting arrests damaged the operation’s deniability. The campaign also had no exit criterion, no stated condition under which Israel would consider the account settled, and so it ran in various forms for decades.

How did the RAW-LTTE relationship become a failure?

India’s intelligence relationship with Tamil militant groups in Sri Lanka failed through objective drift, the quiet mutation of an operation’s purpose without any single decision to change it. What began in the early 1980s as leverage against the Sri Lankan government, training and arming Tamil militants, gradually became the sponsorship of an armed movement that India could neither control nor abandon. When the Indian Peace Keeping Force deployed in 1987 it found itself fighting the very groups Indian intelligence had helped build, losing more than twelve hundred soldiers, and in 1991 the Liberation Tigers of Tamil Eelam assassinated former Prime Minister Rajiv Gandhi. The episode is the textbook illustration of proxy blowback.

What was the strategic problem with arming the Afghan mujahideen?

Operation Cyclone, the CIA program that armed the Afghan mujahideen against the Soviet occupation, achieved its near-term goal, since the Soviet Union withdrew from Afghanistan in 1989. Its strategic problem was that it had no exit criterion and no plan for the day after victory. The weapons, the fighters, and the ideological infrastructure built to bleed the Soviets did not dissolve when the Soviets left; they fed an Afghan civil war and contributed to the militant ecosystem from which al-Qaeda later emerged. The operation won its battle and lost the larger strategic argument, which is why it is studied as a warning about tactical success without strategic alignment.

Why did the extraordinary rendition program damage the United States?

The United States extraordinary rendition program damaged its sponsor on several of the eight tests at once. Its objective drifted from intelligence collection toward indefinite extrajudicial detention. Its deniability collapsed under journalistic and judicial scrutiny, as tracked aircraft and investigating host-country courts steadily exposed it. It harmed innocent people, the wrongful detention of the German citizen Khaled el-Masri being the most documented case, and a 2014 study by the United States Senate Select Committee on Intelligence concluded that the harsh techniques the program enabled had not produced the unique intelligence its defenders claimed. The strategic cost in alliance friction and lost moral authority was never repaid.

Is regime change ever a realistic covert objective?

The historical record is heavily discouraging on this point. Regime change sounds like a clear objective but is not, because the fall of a government is not an event a foreign operation can deliver; it is an outcome that depends on the political behavior of an entire society. The Bay of Pigs is the cautionary monument. Covert action has occasionally contributed to changes of government, but the operations that read as successes in this analysis succeeded precisely because their objectives were narrow and completable acts, not broad political transformations. A useful rule drawn from the cases is that the further an objective moves from a discrete act toward a societal outcome, the lower the probability of success.

How does intelligence accuracy affect operational outcomes?

Intelligence accuracy is one of the two foundational tests, and the cases show that operations rarely fail at the moment of action; they fail in the quality of the picture assembled in the months before. The Eichmann capture, Operation Neptune Spear, and Stuxnet all rested on intelligence pictures of exceptional and verified detail. The Bay of Pigs failed on an assumption that was never confirmed, and the rendition program’s wrongful detentions flowed directly from identity confusions that disciplined verification would have caught. For close-range methods in particular, where there is no standoff distance to absorb an error, the accuracy of the intelligence picture is the difference between a precise action and a catastrophe.

What did the Church Committee conclude about covert action?

The United States Senate investigation of 1975, commonly known as the Church Committee after its chairman, examined intelligence abuses including covert-action programs and assassination plots against foreign leaders. Its broad conclusion, relevant to this analysis, was that covert action conducted without clear objectives, adequate oversight, and accountability tended to produce operations that damaged the United States more than they served it. The committee’s findings became a foundational reference for later scholarship on why covert operations fail, and they reinforce the diagnostic’s central theme that the absence of clear objectives and defined limits is the most reliable predictor of long-term failure.

Why does deniability matter if everyone suspects the truth anyway?

Deniability is often misunderstood as a way of hiding the truth from the world, and judged by that standard it can look pointless once suspicion is already widespread. Its real function is different. Deniability is what allows an adversary state to absorb a blow without being forced, by domestic humiliation, into a retaliation that escalates a covert conflict into an open one. A killing that is suspected but unproven can be quietly tolerated by a government that would be politically unable to tolerate a killing that has been formally acknowledged and proven. Deniability, in other words, is not protection for the sponsor’s reputation so much as a controlled-escalation device that keeps a confrontation below the threshold of open war. That is why its slow erosion is dangerous even when suspicion is already universal, and why a campaign that runs without an endpoint is spending a resource it cannot replace.

How is a successful covert operation different from a justified one?

The diagnostic in this analysis measures success, not justification, and the two questions are genuinely separate. A covert operation is successful, in the operational sense used here, when it achieves its sponsor’s stated objective at an acceptable cost and without producing consequences that reverse the gain. Whether the operation was justified is a question of law and ethics: whether the objective was legitimate, whether the action was proportionate, and whether it respected the sovereignty of the state on whose soil it occurred. An operation can be a flawless operational success and a serious wrong at the same time, and it can equally be morally defensible yet operationally incompetent. Readers assessing any covert campaign, India’s shadow war included, should keep the operational verdict and the moral verdict in separate hands, because collapsing the two produces sloppy thinking in both directions.

Why is Stuxnet treated as a success when it was eventually discovered?

Discovery is sometimes assumed to mean failure, but that assumption misreads what the operation was designed to achieve. Stuxnet was built to delay a specific industrial process, the enrichment of uranium at Natanz, and to buy time as an alternative to a kinetic strike that hardliners were urging. By that standard it worked, because the centrifuge losses it caused were real and were not reversed when the malware was finally identified and analyzed. The discovery ended the weapon’s useful life, but a finite useful life had been a reasonable expectation from the start, and the operation had already delivered the delay it was meant to deliver. The deeper point is that an operation should be judged against the objective it actually had, not against an imagined standard of permanent secrecy. Stuxnet had a soft but genuine exit criterion built into it, the moment of discovery, and an operation that anticipates its own ending is far better positioned than one that assumes it will run forever.

What single decision would most improve India’s shadow war?

Measured against the historical record, the most valuable decision India’s planners could make is also the one they appear least likely to make voluntarily: the definition of an exit criterion, a stated condition under which the campaign would be declared complete. The campaign already passes the tests of objective clarity, intelligence accuracy, operational discipline, political patience, deniability, and collateral-damage control. Its single clear structural vulnerability is the absence of a defined endpoint, and that vulnerability is not a matter of skill or resources but of design. A finite list of names, a verifiable change in the adversary’s behavior, or a fixed time horizon would each convert an open-ended program into a bounded one, and a bounded program is the kind the historical record shows ending well. The difficulty is that defining an endpoint is precisely the decision that institutional incentives push against, because a successful program generates no internal constituency for stopping it.

How does India’s shadow war compare to Operation Wrath of God?

The two campaigns are closely comparable, which is why Wrath of God serves as the central reference point for assessing India’s shadow war. Both are sustained campaigns of targeted elimination conducted on foreign soil against members of organizations the sponsoring state holds responsible for attacks on its people. Both have been executed with evident discipline and reached many of their intended targets. The instructive difference is in the tests Wrath of God failed: the Lillehammer affair showed how a long campaign’s tempo can erode the verification that prevents the killing of an innocent, and the campaign’s lack of an exit criterion allowed it to run for decades. India’s shadow war, so far, has avoided a Lillehammer, but it shares the same structural absence of a defined endpoint, and the Israeli precedent is the clearest warning of where that absence can lead.