> ## Content Index
> Fetch the complete content index at: https://ishangirdhar.com/llms.txt
> Use this file to discover other available public pages before exploring further.

# The Hacker and the State
- URL: https://ishangirdhar.com/the-hacker-and-the-state/
- Published: 2023-04-25T04:56:00.000Z
- Updated: 2026-11-17T17:36:25.000Z
- Description: by Ben Buchanan
- Author: Ishan Girdhar
- Tags: Books, #reading-list, #ebook, #toc, #book

## Highlights

> Rather than realizing this apocalyptic vision, however, cyber attacks have become a low-grade yet persistent part of geopolitical competition. They happen every day. Government hackers play an unending game of espionage and deception, attack and counterattack, destabilization and retaliation. This is a new form of statecraft, more subtle than policymakers imagined, yet with impacts that are world-changing.

> The more competitive aspects of statecraft rely on two overlapping but distinct approaches: signaling and shaping. The distinction between the two is vital.5 If international relations are like a game of high-stakes poker, to signal is to hint credibly at the cards one holds, in an attempt to influence how the other side will play its hand. To shape is to change the state of play, stacking the deck or stealing an opponent’s card for one’s own use.

> Since the dawn of the nuclear age, the theory and practice of international relations has focused on signaling, and with good reason: humankind’s most powerful weapons have become so destructive that they cannot be used except in the most extreme of circumstances. The canonical scholarship of the Cold War period thus explained not how to win a conflict, but how to avoid it on one’s own terms.6 Theorists like Thomas Schelling, who received the Nobel memorial prize in economics for his studies of game theory, described how a state can gain and retain an edge without firing a single shot. To hear Schelling tell it, much of statecraft is about manipulating the shared risk of war, coercing an adversary with carefully calibrated threats so as to gain a peaceful advantage.

> Military mobilization is an example of statecraft by signaling. Deploying armed forces highlights one’s fighting capabilities and demonstrates resolve to adversaries. It suggests that any aggression will bring significant consequences. For this reason, during the Cold War, the United States regularly positioned forces in Western Europe. Since there were not nearly enough American troops to stop a Soviet invasion, one might have wondered what these warriors could do. Schelling had a ready answer: “Bluntly, they can die. They can die heroically, dramatically, and in a manner that guarantees that the action cannot stop there.”7 The Soviets knew that no president could suffer the loss of thousands of Americans and not retaliate. The troops lent credibility to the United States’ signal of commitment to the continent. Their presence helped keep the peace.

> Many scholars ignored how clandestine activities subtly shaped the global environment. These operations were hard to spot and harder still to study, but they mattered. A few American policymakers argued early on that aggressive shaping needed more attention. The famed diplomat George Kennan wrote in 1948 that American policymakers hewed too blindly to an overly simplistic worldview, in which times of war were neatly separated from times of peace, failing to “recognize the realities of international relations—the perpetual rhythm of struggle, in and out of war.”9 Kennan suggested that the inevitable conflict between states’ divergent interests would lead to a constant competition for advantage—a vision that proved prescient.

> Pushed far enough, deception can morph into sabotage. Late in the Cold War, the Soviet Union began to spy aggressively on American technological production and markets, obtaining thousands of technical documents and sample products for their engineers to study. Once the CIA realized what was happening, the agency saw an opportunity. It fed flawed designs to the Soviets, ones that looked real but would fail before too long. As a result, the CIA managed to get poor-quality computer chips installed into Soviet equipment, faulty turbines installed into Soviet gas pipelines, and much more. Duped Soviet engineers unwittingly used defective designs for chemical plants and tractor factories. The Soviet space shuttle, which never flew, was a variant of a rejected NASA blueprint. Until well after the end of the Cold War, the Soviets never knew about the CIA’s program, which is just how the American saboteurs wanted it.12 Almost always out of view, shaping mattered, too.

> The two parts of this home-field advantage—collection from telecommunications sites and access to data from internet firms like Google and Facebook—work together to play a significant role in the American intelligence apparatus. They represent a shift from the millennia-old form of espionage, in which a well-placed human source steals a few vital secrets at a time, and also differ from the most common type of cyber operation: hacking an individual target or device. Together, the two parts of this advantage feed powerful analytic tools that provide intelligence analysts with near-real-time insights on targets, offering powerful new means of finding people of interest.

> Unlike telegraphs, modern digital communications feel ephemeral. Cellular networks are wireless. The cloud filled with emails and files is invisible. Yet, while geography may seem unimportant, it still matters tremendously. All digital messages must take a physical route. Whether they go through the air or along cables, whether they travel short distances or long, they nonetheless have a real presence and must pass, however briefly, through some points on Earth between sender and receiver.

> Another legal authority the NSA uses to access information traveling through AT&T’s network is known as “transit authority.”21 Under this justification, the agency can target communications that pass through the United States but do not originate or terminate within its borders. At American sites of collection, the agency and its partners first employ technical filters to attempt to ensure that the collection sweeps up only foreign-to-foreign communications. But some domestic data does get misclassified and included. The on-site collection systems cull the information down to the communications of most interest, sending these back to NSA headquarters for further storage and analysis.

> Much of the hard data on passive collection, including that done using transit authority, comes from the documents leaked in 2013 by Edward Snowden, a contractor to the NSA. This means that the evidence of NSA activities is ample but imperfect. On the one hand, some of the leaked documents are likely to overstate NSA capabilities and successes; just as in every organization, employees have incentives to make their programs look good. On the other hand, since most of the leaked files are over five years old, it is likely that NSA capabilities have gotten much better in the intervening years as technology has progressed.

> When asked about its activities at 33 Thomas Street and elsewhere, AT&T says that it does not “allow any government agency to connect directly to or otherwise control our network to obtain our customers’ information.” Instead, the company says its practice is to “simply respond to government requests for information pursuant to court orders or other mandatory process and, in rare cases, on a legal and voluntary basis when a person’s life is in danger and time is of the essence.”29 Other telecommunications companies, including British firms, have given similar answers when questioned about their activities.30 For all the techno-utopian talk of a borderless internet, cyberspace is still physical space.31 National laws compelling corporate cooperation still apply—a fact that the Five Eyes are happy to use for their geopolitical benefit.

> This modern system creates an intelligence opportunity. Some of the users of internet platforms are foreign individuals of great interest to the NSA. Luckily for the agency, the dominance of American firms means that these foreign targets will likely send their emails, online messages, and other data through internet platform companies that are legally required to cooperate with the United States government. Under the Foreign Intelligence Surveillance Act, the agency can compel internet companies operating in the United States to turn over data on a target if two criteria are met.

> When NSA analysts are tracking foreign intelligence targets, they can use PRISM to gather more information. The analyst usually begins with what the agency calls a selector, which is often an email address or an IP address that corresponds to the target. With the target’s virtual identity in hand, PRISM enables two kinds of collection: surveillance and stored communications. The former provides forward-looking collection of the target’s activities online. It can give analysts great insight into what a target is up to, enabling a response if warranted. The latter is backward-facing, and arranges for the transfer of a company’s records on a target to the government. This kind of collection sheds light on past activities, including messages the target sent to other possible accomplices, messages received, and files the target uploaded, such as photos and videos. Between the two methods, analysts can gain extraordinary insight into the target’s digital and even physical life.

> That week’s activity level seems not to be exceptional. Annually, PRISM informs tens of thousands of intelligence reports on a wide variety of subjects, including counterterrorism, counterproliferation, weapons development, space programs, cyber defense, regional assessments, counterintelligence, and much more. Across the broad US intelligence mission PRISM provides incredible insight, all because foreign intelligence targets rely on American businesses to communicate and coordinate. Perhaps most crucially, PRISM serves as one of the most significant sources of information for the President’s Daily Brief—a sign of just how important the Five Eyes home-field advantage is to the intelligence community’s top priority of informing senior policymakers.

> But when the Games ended and the athletes went home, the United States did not dismantle its collection capabilities as it had promised.36 Instead, it used the well-positioned spying apparatus for its own ongoing missions. The job, for the most part, was well done. The American espionage software built upon the lawful interception mechanisms that Greek law enforcement used but, crucially, it bypassed key auditing and oversight systems.37 This made possible the secret wiretapping of more than a hundred targets, including the prime minister and his wife, cabinet members, the mayor of Athens, and various Greek journalists.

> The United States has another way to extend its home-field advantage overseas: through its embassies and consulates all over the world. Many of these buildings have antennas hidden on their roofs or behind facades. Small teams of intelligence officers from both the NSA and CIA run collection programs that steal secrets in foreign lands.52 In one of the operations that sparked the most controversy when it became public, the NSA apparently tapped German Chancellor Angela Merkel’s cell phone, using the secret listening devices on top of the United States embassy in Berlin.53 That the United States would spy on an allied leader, particularly one so tightly aligned with President Obama, shows how friends and enemies are all sometimes targets of intelligence collection.

> But the analysis can extend offline, too. One relevant Five Eyes program, ROYAL CONCIERGE, replete with a crown-wearing and scepter-carrying penguin for its logo, extracts hotel reservation confirmations from emails collected by the signals intelligence apparatus. Analysts use it to track the travels of foreign diplomats and other targets, enabling intelligence agencies to position additional electronic and human intelligence gathering capabilities in advance if needed.70 In this way, cyber espionage can beget physical espionage.

> To manage the problems posed by cryptography’s convergence, the NSA settled on an approach that is often informally referred to as “Nobody But Us,” or NOBUS (unlike the other all-caps terms the agency uses, this is not a code-name for a secret program, but just an acronym for a phrase).16 Originally, the idea behind the term was that there were mathematical ways to ensure that only the United States could use certain espionage capabilities. In current usage, though, it often refers to a more general policy goal: when there is tension between the offensive and defensive missions—perhaps because both targets and citizens use the same kinds of encryption—the NSA tries to secure communications against all forms of eavesdropper decryption except those decryption capabilities that are so complex, difficult, or inaccessible that only the NSA can use them. In other words, if the NSA is the world’s best high-jumper, then the agency wants to set the bar just an inch above where its competitors can jump but low enough that it can still clear it.

> To most people, this distinction might seem meaningless. But Tye drew an alarming conclusion from the White House’s modifications: some American intelligence activities were beyond the reach of citizens’ democratic process. As Tye knew, intelligence agencies’ overseas activities against foreign targets are most closely governed by a presidential executive order signed by Ronald Reagan, known as EO 12333, and updated several times since. EO 12333 gives the NSA and other intelligence agencies a much freer hand in their overseas programs than they have on American soil.

> The Snowden leaks, which began several years later, shed more light on Dual\_EC’s flaws and how they came to be. With the benefit of hindsight and internal documents, the NSA’s subtle hand becomes visible throughout Dual\_EC’s creation and standardization process. Initially, the agency worked to influence the design of the number generator. “Eventually,” an in-house memo states, the “NSA became the sole editor” of the Dual\_EC specification, and presumably had enough free rein to manipulate the requirements to suit its espionage purposes.16 It was likely here that the NSA managed to introduce the backdoor and pick faulty default versions of P and Q.17 If this is the case, the agency fulfilled the first and second requirements for enabling the backdoor: choosing non-random points and knowing the mathematical relationship between them.

> While the precise launch date of China’s cyber espionage campaign is uncertain, it probably began in earnest not long after 2000\. It has only grown in the decades since. Spear-phishing—the practice of sending socially engineered emails in order to dupe a target into surrendering vital information or opening malicious code—was and is a common Chinese tactic. China’s hackers, spread across organizations like the People’s Liberation Army and the Ministry of State Security, used the technique again and again because it worked so well. From the comfort of their homes and offices, they could blast out thousands of emails, some more carefully crafted than others, and wait to see what catch of the day came back.

> The Chinese hackers aimed not just to gain secrets of tactical and operational value, but to reshape the strategic relationship with the United States in their favor. The speed, scale, and scope of their efforts were of a different magnitude than the traditional espionage that predated the digital age. In large part, this is because digital information is much more portable than paper. For example, before Daniel Ellsberg, the Vietnam War whistleblower, could leak the Pentagon Papers, he and his family and friends spent more than a year photocopying the seven thousand pages, one at a time, racking up thousands of dollars in fees as they went.2 By contrast, once computer hackers gain access, they can often vacuum up and freely move similar volumes of information in hours, if not minutes.

> China launched such an aggressive campaign for a reason: it does not have the home-field advantage that the Five Eyes enjoy. Though the country has more than a billion people, it is far less optimally placed on the world’s data flows and its technology companies do not command the world’s data in the way that Google and Facebook do. Passive collection and corporate access serve the regime well within its borders, but if the government wants to learn more about what is happening on the other side of the Great Firewall, it has to hack those targets proactively. Spear-phishing is often the easiest way in. For that reason and others, the flood of fake messages began.

> But the Chinese hackers were out to do more than just repress dissent. Their second objective was to undermine the home-field advantage that so greatly benefited the United States. They wanted to know which targets in China the United States was surveilling through Google’s systems. After hacking their way in, the Chinese operators began querying names on the company’s internal “legal-discovery” portal. This portal managed the requests for information that law enforcement made to Google for information pertaining to ongoing investigations; it was probably separate from the PRISM system used by the NSA.

> A third motive for the Chinese operation against American technology firms was that the hackers wanted access to those companies’ secrets—starting with their source code. Code is itself a valuable form of intellectual property, often highly coveted by foreign governments and their favored state-owned enterprises. The source code of the companies Operation Aurora targeted would have been of enormous value to their Chinese competitors trying to gain an edge in the global marketplace.

> The beginning of high-profile private-sector reporting on government-backed cyber operations can be traced back to a precise date: February 19, 2013\. That was the day that a firm called Mandiant released its detailed account of the activity of a Chinese hacking group it named Advanced Persistent Threat 1, or APT1\. While other private-sector companies had published reports on foreign hackers before, Mandiant’s revelations provided much more detail and received widespread media attention, including a front-page story in the New York Times. In one fell swoop, the company put geopolitically motivated hacking in a spotlight, to which it would return again and again in years to follow.

> In 2012, it seems, the NSA’s signals intelligence apparatus picked up a series of emails among three Chinese nationals. Two of them were military hackers in China. The third was a late-forties businessman in Canada named Su Bin, who supervised the British Columbia office of an aviation company called Lode-Tech. Lode-Tech’s business was to track global aerospace technology developments for its clients; Su Bin’s mission was to track American military aerospace developments for China. He used his knowledge of the industry, aviation, and English to guide the Chinese hackers toward the most valuable targets.

> This is the power of counterintelligence. While the discipline is often seen as defensive, the Ōshima case and others like it offer reminders that the most powerful counterintelligence missions are proactive. Traditional spying focuses on stealing political, military, and economic secrets, but counterintelligence goes further, penetrating an adversary’s intelligence service to find out what it knows and how it knows it. The American spymaster Allen Dulles, who headed the Central Intelligence Agency in the early years of the Cold War, ordered his officers to compromise the sources, methods, people, and codes of foreign intelligence services—and that mission continues.5 Achieving it requires intrusive and aggressive collection. It also necessitates spying broadly, since sometimes the best source of information on a foreign intelligence service is in another state entirely.

> Faced with such hard-to-spot threats, NSA analysts realized they had to be proactive. They recognized that the basic forms of cyber defense that took place within the United States’ own computer networks—including software updating, antivirus scanning, and network security monitoring—were necessary but insufficient against sophisticated hackers. To defend against nimbler adversaries, it was also necessary to take the initiative. Defenders sometimes had to venture out from their own networks into a more active posture. For better cybersecurity, the NSA realized, it was best to hack the hackers.

> It is fairly common for multiple signals intelligence agencies to hack the same computers and networks. What is of interest to one state is often of interest to another. Perhaps the most notable example is an unnamed research institute in the Middle East that the Russian cybersecurity company Kaspersky Lab called “The Magnet of Threats.” Six different major groups hacked the institute’s systems: two English-speaking groups (likely corresponding to the NSA and GCHQ), two Russian groups associated with different intelligence agencies, a French intelligence agency, and a Spanish-speaking group of hackers.12 If any of these hacking groups had realized what was going on, they would likely have spotted the opportunity to spy not just on their target, but also on the others spying on the same target.

> To spot and seize these kinds of counterintelligence opportunities, the NSA has its Territorial Dispute program, which agency employees often abbreviate to TeDi. By assembling information the NSA has gathered about a wide range of significant hacking groups all over the world, the TeDi effort builds a “signature” for each group, identifying its distinguishing elements. These signatures can include telltale file names, snippets of code, or habitual behaviors.13 Using two to five indicators per actor, the NSA is able to differentiate among its adversaries; each group has a unique combination of indicators making up an unintentional calling card.

> The NSA presentation with the milkshake cover explains what is known as fourth-party collection.20 This term builds on some older concepts. First-party collection is information collected by the agency itself. Second-party collection is intelligence gathered and shared by another Five Eyes member. Third-party collection is information obtained by an intelligence agency outside the Five Eyes and shared with the NSA. Fourth-party collection is the realm of illicit milkshake-drinking: it involves knowing that one state has intelligence on another state and helping oneself to it.

> Stuxnet was different. It was not an attempt to inflict visible harm or to signal the prospect of more to come. Instead, it was an attempt to shape the situation, to buy time and undermine confidence, key pillars of the United States’ strategy for dealing with Iran. No worm could delay the Iranian nuclear program forever, but Stuxnet could give the United States a better chance of overall success. As the worm tore apart Iranian centrifuges and slowed their enrichment progress, the noose of economic sanctions tightened, aided in part by Wiper’s sabotage.

> In the most favorable interpretation, Stuxnet provided not just time but leverage. A New York Times analysis after the Iran deal’s signing concluded that the United States’ power to thwart Iran’s ambitions mattered. The analysis noted that, for as much as the Obama administration praised diplomacy and economic sanctions, equally important were the “covert actions that repeatedly, if briefly, set back the nuclear program and convinced Iranian elites that its secrecy had been compromised.”44 Whether this was an overall success depends in part on one’s view of the deal itself; the Trump administration later withdrew from the agreement.

> That the worm came to light at all is perhaps the most striking part of the entire saga. An ad hoc group of cybersecurity professionals that most people had never heard of uncovered a classified operation ordered at the highest levels of the American government. The United States carried out Stuxnet in quiet partnership with an ally, tested it in extreme secrecy and at great expense, delivered it to another country’s secret nuclear facility on the other side of the world, and then watched as its details eventually emerged for all to see. While bad luck had previously ruined covert missions and intelligence operations, this was more than misfortune. The Stuxnet outcome hinted at a pattern that would become increasingly apparent as time went on: states had tremendous power in cyber operations, but nonstate actors including researchers and journalists were increasingly important, as well.

> Once it got onto a target computer, the hackers’ code loaded itself to a folder it labeled Shamoon, which quickly became investigators’ name for both the code and the operation. Shamoon’s code consisted of three components. The first of these copied itself to various places on the computer and the Aramco network. It also configured itself to run in the future whenever the targeted computer started up. This is a fairly common technique for achieving what cybersecurity experts call persistence: the capacity to remain on a targeted system despite actions taken by users, such as rebooting the machine or trying to delete malicious code. With this foothold established, the Shamoon code deployed its other two components: one to wipe files from the target system, and one to let the attackers know how the operation was proceeding.

> While Stuxnet and Wiper tried to shape the environment, these public-facing attacks were about signaling. But they just were not very effective as signals, failing for three reasons. First, the message was hardly nuanced: Iran was displeased with geopolitical actions and statements. Conveying this displeasure revealed no new information about Iran’s geopolitical views and ambitions. Long before Shamoon, every observer knew that Iran and Saudi Arabia were regional adversaries. Before Operation Ababil, it was no secret that Iran was greatly aggrieved by Stuxnet. Similarly, the mutual enmity between Iran and Sheldon Adelson was widely known, even if Adelson did escalate the rhetoric with his call for a nuclear test in the Iranian desert. Signals are an attempt to change behavior by revealing credible information about one’s priorities; the Iranian operations fell well short of this bar.

> Third, a key element of nuanced geopolitical signaling is having the capacity to inflict carefully measured amounts of violence, with the threat of more to come. The Iranian attacks all failed in this regard. In each case, the particular cyber capability was mostly spent as soon as the attack was launched. In the cases of Aramco and Sands, the Iranians burned much of their access and capability with the attack, and in the denial-of-service campaigns, the additional harm that could be inflicted was minimal. Even when the Iranians inflicted less damage than they might have, as in the Sands attack, the limitation did not come across as deliberate signaling but seemed much more likely to be an error on the part of the attackers.

> Coercion is one of the bedrock goals of foreign policy, as it offers the possibility of attaining desirable outcomes without full-scale conflict. It is signaling in its purest form. Whereas Iran was limited to retaliating for past behavior and actions it did not like, North Korea had an opportunity, in advance of The Interview’s release, to force Sony to change its mind.

> With the benefit of hindsight, and thanks to the work of some diligent investigators, the story takes on more importance. Three big lessons have emerged from the blackouts in Ukraine. First, this case counters the claim that cyber operations are the new nuclear operations, capable of destroying societies in a single blow with a weapon fired from afar. Even allowing that future blackouts might be more destructive, the force of cyber capabilities will remain well short of what nuclear weapons can do. Ukraine provides more evidence that the common analogy simply does not hold up.

> Second, the blackouts rebut the notion that cyber operations are akin to conventional operations, in which states deploy widely understood capabilities in ways other states know how to interpret. Because the attackers in Ukraine showed a restraint or impotency that did not lend itself to easy explanation, observers could variously view the blackouts as a signal, a test, or a failure. Even in the context of the irregular conflict in Ukraine, in which Russia has used a wide range of military and intelligence capabilities to advance its interests, it is not apparent what particular purpose the blackouts served. In general, the goals of hacking operations are less clear than many scholars and policymakers assume, and they are obscured still further by technical complexity. Even when the real-world effect is obvious, signaling intentions with cyber capabilities is still difficult.

> This organization, and its poll, never existed. British intelligence operatives had made it up and circulated it as part of their efforts to push the Republican Party toward choosing a pro-war candidate. The British fabrication efforts, lending weight to arguments made by Republicans aligned with British interests, helped catapult the candidate Wendell Willkie, previously a definite underdog, to the nomination. Willkie, a former Democrat, proved to be an enormous asset to Britain, acquiescing during the presidential campaign to Franklin Roosevelt’s transfer of American destroyers to the Royal Navy. Maybe most importantly, he lost the November election, leaving Roosevelt and his strong support for Britain in charge.

> Even after eight decades, however, the British effort stands above all—and neatly foreshadows the Russian interference in the 2016 US election. Here was direct interference in United States presidential politics by a foreign actor, aided by the spread of false information, the manipulation of popular media, the clever timing of leaks and lies, and the creation of propaganda that aligned with preexisting narratives—with lingering uncertainty, even in retrospect, about the overall impact. The Russian hacking and leaking mission has justifiably become one of the most-discussed cyber operations of all time, but its British antecedent lends much-needed historical context. More than anything, the British effort serves as a reminder that the true importance of cyber operations comes not from their damage to machines, but from their destabilizing effects on humans and societies.

> Soviet propagandists learned this long ago.84 They realized that the most effective influence operations use falsehoods that are not cut out of whole cloth, but built on preexisting notions. Fairly or not, Hillary Clinton had a reputation in some quarters of the American electorate as corrupt and selfish. The leaks of emails showing DNC senior staff discussing ways to undermine Sanders’s primary challenge played into this preconception. The social media campaign and real-world rallies exacerbated it still further. More generally, the Russian wedge widened underlying divisions in the United States on hot-button issues. It did not have to create racial tensions and ideological differences; they were already there waiting to be exploited.

> The situation has not gotten much better since the frantic days of 2016\. The United States could have treated the Russian operation as a harbinger of greater cyber destabilization to come. The two political parties could have come to agreement about the need for a united front against foreign interference. But the underlying divisions in the United States have only deepened. Perhaps worst of all, the differences seem to be not only of opinion or values, but of truth; Trump advisor Kellyanne Conway’s famous line about “alternative facts” highlights this divide.89 A democracy that cannot agree on fundamental realities—a foundation of truth that can prove impervious to foreign manipulation—is one that is highly susceptible to the shaping power of future wedges.

> One final theory seems to be the simplest and likeliest: this was sabotage in the form of exposure—and sabotage is a classic form of shaping. By dragging the NSA’s capabilities into the light of day, the Shadow Brokers eroded the agency’s capacity to use its powerful hacking tools. In the words of Leon Panetta, a former head of the CIA and former Secretary of Defense, “these leaks have been incredibly damaging to our intelligence and cyber capabilities.” The operational effect of a leak is devastating, he said. “Every time it happens, you essentially have to start over.”56 It would certainly make sense for a hostile foreign intelligence agency to want to inflict this kind of setback on the United States, and perhaps it would make sense for a traitorous insider to feel similarly.

> The Shadow Brokers almost certainly did more harm to United States security and intelligence collection capabilities than Snowden did. Snowden took his documents to journalists and let them determine what was in the public interest to publish. While the NSA strenuously disagreed with many of those journalists’ decisions, the United States government was at least often consulted and alerted prior to publication. It could make the case for why some documents should be partially or entirely redacted.59 Indeed, as a result of these redactions, the published Snowden documents are comparatively silent on some notable categories of operations, especially the NSA’s active hacking efforts against non-democracies.

> North Korean operatives found a better way to rob banks. They did not have to break through reinforced concrete or tunnel under vaults to get at the money, and they had no need to use force or threats. Instead, they simply duped the bank’s computers into giving it away. To do this, they set their sights on a core system in international business called the Society for Worldwide Interbank Financial Telecommunication, or SWIFT. The SWIFT system has been around since the 1970s. Its eleven thousand financial institutions in more than two hundred countries process tens of millions of transactions per day. The daily transfers total trillions of dollars, more than the annual gross domestic product of most countries. Many financial institutions in the SWIFT system have special user accounts for custom SWIFT software to communicate their business to other banks all over the world.

> But, as this book has shown, the cyber operations that achieve such a broad range of objectives often fail at one important task: signaling. Cyber capabilities are ill-suited for communicating with other states to encourage or discourage future behavior. In this way, they depart from their predecessors, like conventional military arms and nuclear weapons. States sometimes try to signal with cyber operations, but usually fail. The rhetoric around topics like cyber deterrence can sometimes outrun the reality, which is that cyber capabilities rarely offer clear, credible, or calibrated means of signaling and coercion. There are four reasons for this.

> First, visibility enhances signaling, but cyber capabilities often benefit from or require secrecy. For decades, canonical international relations scholarship and policymaking has focused on those activities that all can see. Presidential summits and international diplomacy are fixtures in the minds of scholars and the public. Widely examined cases such as the Cuban Missile Crisis center on the capacity of leaders to walk up to—and then back from—the geopolitical brink in a way that shows resolve, benefits their own interests, and ultimately averts a war. For decades, this has been the art of modern statecraft.

> By contrast, many cyber operations can’t meet their goals—such as passive collection, surreptitious decryption, source code manipulation, and counterintelligence—if the operations come into view. Even in some acts of sabotage, such as Stuxnet and Wiper, perpetrators have tried to remain invisible as long as possible to increase operational effectiveness. “If you know much about it, \[cyber is\] very easy to defend against,” explained Michael Daniel, cybersecurity coordinator during the Obama administration, adding, “that’s why we keep a lot of those capabilities very closely guarded.”4 As seen throughout this book, exposing particular hacking capabilities tends to render those capabilities much less useful, especially against well-secured targets.

> Third, even visible and controllable cyber operations are much harder to interpret than conventional tools of statecraft. Robert Jervis argued that, even with conventional weapons, it is always easier to send signals than it is to interpret them, since policymakers assume that the nuances in their signals will be apparent to their counterparts.6 That said, some signals are more interpretable than others. It is readily apparent what message the United States sends when it mobilizes an aircraft carrier battle group; decision-makers on the receiving end of the signal understand and fear the displayed readiness to attack. The deployment of thousands of American soldiers to the European continent during the Cold War is similarly obvious in meaning. As noted in the Introduction, these troops might not be capable of repelling a Soviet invasion, but they could die trying. Every Soviet leader knew that the death of thousands of Americans would precipitate war. There was little ambiguity.

> In lieu of much effective signaling, states have found it better to use their cyber capabilities to grapple with one another for advantage. This book’s narrative spans several decades of cyber operations. This history demonstrates an important pattern: the harm that hackers can do is expanding faster than the deterrence or defenses against them. For years, increasingly aggressive activities have in many cases gone unblocked and unmet by meaningful retaliation.

53 highlights, exported from my Kindle via Readwise.