This analysis focuses on spies who have been identified by the government, following the methods of the Defense Personnel Security Research Center, by relying upon publicly available information.85 They are identified by being convicted of espionage or an espionage‐related offense if they died after their discovery but before their conviction, or fled abroad after or before charges were filed against them. Their ethnicity is mainly assigned by their country of origin. If a spy’s country of birth and ethnicity are different, that is reflected in the coding; for instance, Taiwan‐born Ko‐Suen Moo is of Korean ethnicity, so he is recorded as having been born in Taiwan and as ethnically Korean. The ethnicity of American‐born spies is also recorded in broad categories including “white American,” “black American,” and “Hispanic American,” where the specific ethnicity based on their family histories is unavailable. Specific ethnicities for native‐born Americans are included if their ancestors’ country of origin is known.
The sample of identified spies in this analysis could bear little resemblance to the population of total spies because of the secretive nature of espionage. Spies do not want to be caught and seek to conceal their illicit activities. This is unlike analyses of terrorists, who intend to make their crimes well known in order to inspire terror and trigger political or social reforms.86 This might be less of a problem with amateur Chinese spies, especially those engaged in commercial espionage, because they are more likely to get caught than well‐trained spies engaged in state espionage. Although this analysis seeks to build as comprehensive a list of spies as possible, some spies are undoubtedly unidentified, so this analysis could suffer from systematic undercounting of spies, espionage, and espionage‐related crimes.
One problem that could lead to the undercounting of spies is that the government might not want to make known the individual spies that it identifies. Thus, the number of spies in this analysis could be undercounted because the government handles some espionage in secretive settings beyond the public view. However, if this does in fact happen, it likely involves few cases. The U.S. government does prosecute spies in court, and the DOJ typically issues press releases announcing when it has uncovered a spy and charged that person with a crime. The government even does this in some of the most embarrassing cases that expose significant government security failures. Furthermore, law enforcement and the DOJ have an incentive to publicize that they are capturing spies to act as a deterrent to other spies and to justify their budgets and investigative initiatives. The DOJ’s China Initiative even takes credit for individuals convicted of espionage or espionage‐related crimes if indictments against them were filed years prior to the initiative’s creation, such as in the cases of Hao Zhang, Weiqiang Zhang, Ying Lin, and Xiang Haitao.87 The DOJ would not announce its China Initiative and speak at prestigious conferences about it if it could not brag about how the program succeeds.88
Other government agencies could be more secretive in trying to uncover spies, but the FBI and other agencies charged with investigating espionage are mostly civil law enforcement agencies operating under the same incentives. Even when they cooperate with more secretive government agencies and don’t publicize that cooperation, they publicize the indictments and prosecutions that result—as in the case of Edward Snowden.89 Even if the DOJ’s China Initiative was secret when it was started, it would only remain secret if it was a failure and didn’t result in the identification or prosecution of any spies. It is tempting to ascribe a super‐secretive identification and punishment process for spies beyond the public’s view, especially in a culture rife with entertaining fiction about espionage, but the actual U.S. counterintelligence system is not set up that way to punish such offenders.
Overcounting of spies, espionage, and espionage‐related crimes could also be a problem. Spies usually must be discovered and rarely announce their presence, so the discovery of spies is based on government and private‐sector investigations of espionage. Political mandates to discover and prosecute spies could lead to an overidentification of spies. Mark Rasch, former computer‐crime prosecutor, said, “If you’re looking everywhere for spies, you will find spies everywhere, even where they don’t exist.”90 For instance, prosecutors who are empowered and pushed to discover spies would likely result in marginal individuals being convicted of espionage‐related crimes, such as making false statements or tax fraud, that were discovered in the course of an espionage investigation that wouldn’t otherwise be prosecuted. In other cases, scientists in the United States may share data with scientists in foreign countries while collaborating on research projects that could result in the data being used as part of intelligence‐gathering efforts by foreign governments. Those scientists in the United States who share the data could be spies, or they could just be scientists whose findings or data are used as a source of intelligence—but these marginal individuals are included in this analysis if they’re prosecuted for espionage or espionage‐related crime.
Prosecutors might also entirely ignore spies from friendly foreign powers. For instance, spying by Israel in the mid‐1990s was rarely prosecuted, but the government claims that it was likely significant.91 Commercial espionage conducted by the French government is also a big problem, according to former defense secretary Robert Gates.92 Spies from those countries could either be deported from the United States; forced out by other means, such as diplomatic pressure on the foreign government to recall them; given stern warnings to stop; or be entirely ignored because the foreign government is not considered a threat to the United States. This source of bias would result in a relative undercounting of spies from friendly countries and a relative overcounting of spies from less‐friendly countries.
In still other cases, individuals could be accused of espionage based on unfounded allegations, investigated or charged with crimes, convicted, and then cleared, as in the case of Russian‐born Sergey Aleynikov. He was convicted of theft of trade secrets and another related offense and sentenced to eight years in prison. The Federal Court of Appeals unanimously reversed his conviction and entered a judgment of acquittal.93 Aleynikov was lucky, and his counsel was especially competent, but some people convicted of espionage or espionage‐related crimes are likely innocent and are either waiting for their convictions to be overturned or else they will continue to languish in prison for crimes they didn’t commit.
Another problem is the selectiveness of government espionage and espionage‐related prosecutions. For instance, there are many leakers of classified information, and very few of them are prosecuted. When they are investigated and prosecuted, political and policy reasons are likely the paramount justifications rather than a commitment to enforcing the law.94 The government also releases selective information about certain alleged spies without prosecuting them. The government did this in the case of Chinese‐born entrepreneur Liu Ruopeng, who earned his masters and doctorate degrees from Duke University and allegedly turned over research secrets to the Chinese government while using stolen research to start a successful firm in China.95 Despite the serious allegations and convincing evidence that was leaked to the public, the government did not charge Ruopeng with any crimes, which indicates that the government likely selectively released evidence to show Ruopeng in the worst light.
Government efforts to identify spies from China, a new focus on commercial espionage, or other initiatives could boost the number of prosecutions and convictions but reduce them elsewhere. In other words, spies will always exist, but the spies that are identified are those that the government targets. If the government directs law enforcement resources to targeting Chinese spies, then it could relax efforts to identify Iranian or Russian spies, for instance. Government targeting is based on political and administrative objectives that might be related to the changing nature of espionage or could be reactive to other political or administrative considerations.
The sample of spies in this analysis could be too small and undercounted, too large and overcounted, or systematically skewed because of the selective nature of government prosecutions. Thus, the sample of spies in this analysis could be biased and unrepresentative of the total population of spies in the United States—a phenomenon called sampling bias. In short, sampling bias in this analysis could mean that the identified spies here could differ to a statistically significant extent from the total population of spies. They could come from different countries, spy for different countries, or commit different crimes. It is tempting to use this analysis to cherry‐pick marginal cases to include those who probably should be charged with espionage and to exclude those who probably shouldn’t have been, but that would introduce an entirely new layer of bias that would significantly reduce the utility of this analysis. As a result, this analysis includes the widest possible sample without cherry‐picking individual spies for inclusion or exclusion. Regardless, one must keep the above problems in mind when considering the findings of this analysis.
The government’s intense public focus on espionage and espionage‐related crimes, growing media and public scrutiny of espionage, the building geopolitical and economic conflict with China, and the lack of any thorough account of all espionage and espionage‐related crimes justify this analysis. Regardless of the above problems with identifying spies, this analysis is a first step in analyzing the totality of espionage and espionage‐related crime in the United States. Systematic efforts like this to build a dataset will help identify whether a problem exists, the scale of the problem, and characteristics of the problem far better than a few anecdotes selectively used in public statements. The main hope is that this analysis will help answer those questions. The secondary hope is this analysis will inspire other researchers who will take the data here, add to it, refine the methods, and produce better research in the future.